Journal of the History of Biology, Vol. 2, No. 1

Spring1969/ Volume2: Number1

Journal of the History of

Biology Published by the Belk-napPress of Harvard University Press Cambridge Massachusetts

SPECIAL ISSUE Explanation in Biology

0 Journalof the Historyof Biology SPRING 1969: VOLUME 2, NUMBER 1

Editor: EverettMendelsohn,HarvardUniversity Assistant Editor: Judith P. Swazey, Harvard University THE BELKNAP PRESS OF HARVARD UNIVERSITY PRESS ? Copyright 1969 by the President and Fellows of Harvard College

Conferenceon Explanationin Biology: Historical,Philosophical,and Scientific Aspects CONTENTS

v

Editors'Foreword

PART I

UNIQUENESS AND CHANGE IN BIOLOGICAL EXPLANATION

Biologyand the Unity of Science

3

DUDLEY SHAPERE

Theories and Explanations in Biology

19

KENNETH F. SCHAFFNER

The Bases of Conflictin BiologicalExplanation

35

RICHARD C. LEWONTIN

Explanationin Biology

47

BENTLEY GLASS

Hugo de Vries and the Receptionof the "MutationTheory"

55

GARLAND E. ALLEN

Joseph Barcroft and the Fixity of the Internal Environment FREDERIC L. HOLMES

Commentary-Part I 123 Ernst Mayr Ernest Nagel 128 Reply to Ernest Nagel Everett Mendelsohn 140 John Platt

*-

134 135

89

PART II

SPECIFIC ALLEGED FACTORS OF UNIQUENESS IN

BIOLOGICAL EXPLANATION

Function and Teleology

151

MORTON BECKNER

From Machine-Theory to Entelechy: Two Studies in Developmental Teleology

165

FREDERICK B. CHURCHILL

Explanation in the Biological Sciences

187

MICHAEL SCRIVEN

Organizational Levels and Explanation

199

CLIFFORD GROBSTEIN

Commentary-Part

II

Edward Manier 207 David B. Kitts 214 William Coleman 216

PART III

ORGANISM, ENVIRONMENT, AND INTELLIGENCE

AS A SYSTEM

Organism, Environment, and Intelligence as a System

225

JOHN PLATT

Essay Review: What Philosophy of Biology Is Not

241

DAVID HULL

Special Historical Note: Darwin's Questions About the Breeding of Animals

269

PETER J. VORZIMMER

ini

Editorial Board: Bentley Glass, State University of New York, Stony Brook; Hebbel E. Hoff, M.D., Baylor University; Ernst Mayr, Harvard University; Everett Mendelsohn, Harvard University; Jane Oppenheimer, Bryn Mawr College. Advisory Editorial Committee: Enrique Beltrin, Mexico; Georges Canguilhem, France; John T. Edsall, M.D., U.S.A.; A. E. Gaissinovitch, U.S.S.R.; Ralph W. Gerard, M.D., U.S.A.; John C. Greene, U.S.A.; Marc Klein, M.D., France; Vladislav Kruta, M.D., Czechoslovakia; Joseph Needham, England; Dickinson W. Richards, M.D., U.S.A.; K. E. Rothschuh, M.D., Germany; Conway Zirkle, U.S.A. JOURNAL OF THE HISTORY OF BIOLOGY is published semiannually in the spring and autumn by the Belknap Press of Harvard University Press, 79 Garden Street, Cambridge, Massachusetts, 02138. Editorial Correspondence and manuscripts should be sent to Professor Everett Mendelsohn, Editor, Journal of the History of Biology, Holyoke Center 838, Cambridge, Massachusetts, 02138. Subscription correspondence should be addressed to Mrs. W. H. Carpenter, Harvard University Press, 79 Garden Street, Cambridge, Massachusetts, 02138. Subscriptions, which are payable in advance, will start with the first issue published after receipt of the order. Please make remittances payable to Harvard University Press. Subscription rates are $7.50 a year in the U.S.; $8.50 in all other countries; $4.50 for a single copy. Journal Design by David Ford

Part I Uniqueness and Change in Biological Explanation

Part II Specific Alleged Factors of Uniqueness in Biological Explanation

149

Part III Organism, Environment and Intelligence as a System

Editors'Foreword

The papers in this volume were originally presented at a Conference on Explanation in Biology held at Asilomar State Park, Monterey, California, in June 1968. The aim of the conference was to bring together historians and philosophers of science and biologists so that they might discuss the range of problems in which they had common interest. A full interchange of ideas between the working scientist, the philosopher, and the historian occurs seldom enough to have made this project one which merited attention. The format called for papers which served as the focus for formal commentary and informal discussion. The quality of the presentations and the reception accorded them led us to believe that a major group of the papers should be published as a unit. They demonstrate the interacting points of view brought to the discussion of very similar topics by the three groups of participants. No unified point of view was sought-nor, indeed, did one emerge in the course of lively discussion. No attempt has been made to develop an artificial continuity between the papers and commentaries, and only brief snatches of the discussions have been edited for presentation here. That further efforts of the sort represented by the "Asilomar" Conference are called for seems evident, both from the fruitfulness of this first joint venture and from the indications that much remains to be discussed and thrashed out. The conference at Asliomar developed from the joining together of interests on the part of a number of individuals each of whom had expressed a desire to bring into confrontation scholars with a common interest in historical and philosophical problems of the biological sciences. Our plans were facilitated by the Commission on Undergraduate Education in the Biological Sciences (CUEBS) which generously undertook the tasks of providing a conference secretariat. Financing for the conference came through a grant to CUEBS from the National Science Foundation for this specific meeting. We are grateful to staff of CUEBS for arranging travel and facilities, and particularly to Martin Schein, former director, Jeffry Baker, and Jane Livermore, who participated in the conference at Asilomar.

v

EDITORS' FOREWORD

The following persons were conference members: Mr. Mark Adams, Dr. Garland E. Allen, Dr. Morton Beckner, Dr. Ernst Caspari, Dr. Frederick B. Churchill, Dr. William Coleman, Dr. Ruth Schwartz Cowan, Dr. Hubert Dreyfus, Dr. Bentley Glass, Dr. T. A. Goudge, Dr. Clifford Grobstein, Dr. Adolf Grunbaum, Dr. Keith Gunderson, Dr. Thomas Hall, Mr. Jonathan Hodge, Dr. Frederick L. Holmes, Dr. David Hull, Dr. David Kitts, Dr. Hugh Lehman, Dr. Richard Lewontin, Dr. Edward Manier, Dr. Ernst Mayr, Dr. Everett Mendelsohn, Dr. Ernest Nagel, Mr. Herbert Odum, Dr. John R. Platt, Dr. Hilary Putam, Dr. L. J. Rather, Dr. Carl Sagan, Dr. Kenneth F. Schaffner, Dr. Michael Scriven, Dr. Dudley Shapere, Dr. George Gaylord Simpson, Dr. Judith P. Swazey, and Dr. William Weedon. Everett Mendelsohn Dudley Shapere Garland E. Allen

vi

Biologyand the Unityof Science DUDLEY SHAPERE Department of Philosophy, University of Chicago, Chicago, Illinois

A recurrent theme in the philosophical interpretation of biology has been the claim that the biological sciences are distinguished in some radical way from the physical sciences, and in particular from physics and chemistry. It used to be said by adherents of this view that the difference lay in the fact that the subject-matter of the life sciences was of a fundamentally different sort (e.g., a distinct kind of substance or force) from that dealt with by the physical sciences. And this difference was alleged to be radical in the sense that the sorts of substances or forces which are the special concem of biology operate according to laws (if indeed they operate according to any laws at all) which are different from and irreducible to those governing mere physical or chemical bodies. In more recent times, however, such positions have tended to be abandoned in favor of ones that located the distinction in some fundamental methodological or epistemological difference. Some have maintained, for example, that the distinctness of biology from physics (and chemistry) lies in the different mode or modes of explanation employed by the former, as opposed or in addition to those employed by the latter. This type of view of the difference, like the earlier type, is claimed by many of its adherents to imply the "irreducibility" of biology to physics. Views of the latter sort have had many advocates and have taken many specific forms. One version has been presented by Professor G. G. Simpson in his fascinating book, This View of Life. According to Simpson, the differences between the physical and the biological sciences can be expressed in terms of kinds of scientific explanations and kinds of questions that elicit them. "How?" is the typical question in the physical sciences. There it is often the only meaningful or allowable one. It must also be asked in biology, and the answers can often be put in terms of the physical sci-

3

DUDLEY SHAPERE

ences. That is one kind of scientific explanation, a reductionist one as applied to biological problems: "How is heredity transmitted?" "How do muscles contract?" and so on through the whole enormous gamut of modem biophysics and biochemistry. But biology can and must go on from there. Here, "What for?"-the dreadful teleological quesonly is legitimate but also must eventually be tion-not asked about every vital phenomenon . . . In biology, then, a second kind of explanation must be added to the first or reductionist explanation made in terms of physical, chemical, and mechanical principles. This second form of explanation, which can be called compositionist in contrast with reductionist, is in terms of the adaptive usefulness of structures and processes to the whole organism and to the species of which it is a part, and still further, in terms of ecological function in the communities in which the species occurs. . . . A further question is necessary: . . . "How come?" This is another question that is usually inappropriate and does not necessarily elicit scientific answers as regards strictly physical phenomena. In biology but not invariably in the physical sciences, a full explanation ultimately involves a historical-that is, an evolutionary-factor.' Thus, according to Simpson, three types of explanation are involved in biology: "reductionist" (answers to the question to the "How?"), "compositionist" (or teleological-answers question "What for?"), and "historical" (answers to the question "How come?"). Inasmuch as physics and chemistry generally concern themselves only with the first of these, it follows that physical explanations cannot do justice to the concerns of biology. Professor Simpson's more detailed discussion includes the following points, among others. While he admits that there are many legitimate approaches in biology, nevertheless "to be effective and to be in fact biological, all approaches must take into account the organization of organisms." 2 This requirement has important ramifications for the difference between characteristic biological and physical explanations: for "to understand organisms one must explain their organization. It is elementary that one must know what is organized and how it is organized, but that does not explain the fact or the nature of the organization itself. Such explanation requires knowledge of how an organism came to be organized and what functions 1. G. G. Simpson, This View of Life (New York: Harcourt, Brace and World, 1964), pp 104-105. 2. Ibid., p. 109.

4

Biology and the Unity of Science the organization serves. Ultimate explanation m biology is therefore necessarily evolutionary," 3 or, as he also says, "historical." This historical dimension in turn entails that biological explanation is concerned with "a sequence of real, individual events" 4 in all their complexity and uniqueness: in this respect biology is to be contrasted with physics, which tends to concentrate on "abstractions," on "the ideal and the generalized." 5 "That hierarchy of complexity and individual uniqueness from physics to geology to biology is characteristic of those sciences and essential to philosophical understanding of them." 6 Because of this concentration on the unique and individual, "the search for historical laws is, I maintain, mistaken in principle . . . historical events are unique, usually to a high degree, and hence cannot embody laws defined as recurrent, repeatable relationships." 7 Further, since "historical events are always unique in some degree," it also follows that "they are therefore never precisely predictable . . . prediction can only be general and not particular. In other words, it does not include any unique aspect of the event, and in historical science it is often the unique aspects that most require explanation." 8 In view of the fact that the biological sciences deal not only with the question "How?" which he finds to be typical of the physical sciences, but also with the further questions "How come?" and "What for?" Professor Simpson feels able to claim that in some ways biology is superior to the physical sciences: . . . the life sciences are not only much more complicated than the physical sciences, they are also much broader in significance, and they penetrate much farther into the exploration of the universe that is science than do the physical sciences. They require and embrace the data and all the explanatory principles of the physical sciences and then go far beyond that to embody many other data and additional explanatory principles that are no less-that are, in a sense, even more-scientific.9 A host of questions could be raised regarding Simpson's account. One might, for example, request a clear and precise analysis of such terms as "organization," "uniqueness," and "complexity" that would do justice to the demands made by the distinction based on them between physical and biological sciences. (After all, crystals and galaxies are organized," 3. Ibid., p. 113. 4. Ibid., p. 128. 5. Ibid., p. 148. 6. Ibid., p. 124. 7. Ibid., p. 128. 8. Ibid., pp. 139-140. 9. Ibid., p. 104.

5

DUDLEY

SHAPERE

planets are "unique," and, unless we look at it through the eyes of the physicist, the material world seems a highly "complex" place, full of colors, textures, shapes, smells, moving bodiesa booming, buzzing confusion. And on the other hand, in what sense, precisely, does biology deal with the "unique" and the "individual"? Certainly, insofar as he is concerned to understand vital phenomena, the biologist is interested in classes of, e.g., organisms-despite any individual differences-rather than in particular individual organisms for their own sakes.) We might complain that the term "law" requires further analysis before we could conclude that, for example, Mendel's Laws, or the Hardy-Weinberg Law, or the Watson-Crick "central dogma" of molecular genetics, or any other of a large number of statements expressing "a recurrent, repeatable relationship between variables that is itself invariable to the extent that the factors affecting the relationship are explicit in the law," 10 should be deprived of the status of "law." Again, we might demand more evidence that "What for?" "How come?" and "How?" are really as distinct questions as Simpson claims-that there is not, for any answer to one, an equivalent reformulation which would be an answer to one of the others. But I am not interested here in centering attention on such questions regarding the sort of view Simpson is espousing; the difficulties in the way of answering them are well known, and are certainly known to Professor Simpson himself. But more importantly, it seems to me that, in spite of all these and other questions of clarification, there is a great deal to what Simpson is saying. Biology does indeed, I will want to argue, deal with something that may appropriately be called "organization" in ways that the physical sciences do not; it is faced with problems of complexity and uniqueness that are not commonly found in physics; and it is concerned with types of explanation that may be called "functional" or "teleonomic." The question is, however, what such claims involve, and, most especially, what conclusions can be drawn from them about the relationships between the physical and the biological sciences. If biology is indeed concerned with organization, complexity, uniqueness, and if, in that concern, it utilizes types of explanation that are different from those employed in physics, does all this imply that biology is an "autonomous" discipline? And if so, in what sense, precisely? In particular, is this autonomy the result of peculiarities that imply the impossibility of unifying the results of biology and physics in any way at all? Or is the question of the "unification" of biology and physics, or even of the "reducibility"of the former to the latter, unaffected? If this is the 10. Ibid., p. 126.

6

Biology and the Unity of Science case, then what are we to make of so much of the philosophical literature on the subjects of explanation and reduction? For that literature, much of it at any rate, tends to take it for granted that, as regards the reduction of one theory or branch of science to another, the statement "Theory (or subject) B is reduced to theory (or subject) A" is equivalent to the statement, "Theory (or subject) A explains theory (or subject) B." In approaching these questions, my aim will not be to attempt a formalization of such concepts as "organization," in the manner of Professors Nagel and Beckner. Nor will I attempt to construct a "translation-schema" which will purport to establish the equivalence (despite appearances to the contrary) of physical and biological explanations. My reasons for avoiding those approaches is not that I do not think there is anything to be gained from them, but simply that the points which I will try to make concerning the preceding questions are not brought out successfully by such approaches. Rather, I will compare biological reasoning-or, more precisely, certain types of it-with a type of reasoning which is frequently found in physics, a type which I will call "entityexplanations." I have discussed numerous features of such explanations elsewhere,"1 and so will not enter into further detail here, but will concentrate only on those features of entity-explanations which are relevant to present concerns. For our purposes here, we need say no more than that an entity-explanation is one in which explanations are offered in terms of entities and their properties and behavior, entities which are alleged to exist and to be causally responsible for certain phenomena. Grouping cases together under the heading "entity-explanations" is to a large extent a matter of convenience: for, first, although there are similarities between the kinds of cases that can be called "entity-explanations," there are also differences between them; secondly, although there are clear cases which it is natural to call "entity-explanations," there are also borderline cases; and, finally, there are many cases in science which one is strongly inclined to call "explanatory," but which one is strongly disinclined to say are explanations in terms of "entities." II In the context of many problems in physics, an entity or group of entities can, if certain conditions are fulfilled, be 11. D. Shapere, "Notes toward a Post-Positivistic Interpretation of Science," to appear in P. Achinstein and S. Barker, eds., The Legacy of Logical Positivism for the Philosophy of Science (Baltimore: Johns Hopkins University Press, 1969.)

7

DUDLEY SHAPERE

considered as effectively isolated from the influence of other entities. The conditions under which this can be done are specified by the physical laws and theories relevant to the problem under consideration, and also by the requirements set by the problem itself. For example, classical mechanics specifies the strength of the gravitational force exerted by one body on another; if one of the bodies is sufficiently small, or sufficiently far removed from the other (so that the force exerted by the former on the latter is sufficiently small), the gravitational influence of the first body may be ignored in calculations regarding the motion of the second. This is done despite the fact that we know, on the basis of the theory, that that influence does exist. The sense of "sufficiency" involved is laid down by the limits of experimental error or by the accuracy required by the problem to be solved. It is indeed fortunate that the universe is such that systems can be considered in isolation from one another; for even where three bodies influence one another's motions, the mathematical complexities that result are forbidding. The possibility of so isolating systems accomplishes an enormous degree of simplification in the treatment of a problem. In biology, however, the common problems of interest, and the types of interactions considered, do not permit such isolation to any substantial degree. Although in general it is possible to describe the structure of an entity or a system in isolation from other entities in its environment, nevertheless when we turn to considering the behavior of that entity, its interactions with other entities, as well as the complex contributions of its own structure to that behavior, become too significant to ignore. The number of factors, and the types of interactions, great. which cannot be ignored is frequently-usually-very At the level of the gene and its influence on heredity, we already find that "the precise manner in which genes act depends on the larger gene complex in which they reside." 12 Phenotypic characters may be determined by many non-allelic genes, and one gene may affect more than one phenotypic character. Identical genes operating in different environments may result in different phenotypes. Further, the heterocatalytic function of any particular gene is not manifested at all times, and induction or repression of gene activity may be a function of other genes, or of still more general environmental factors. In regard to both heredity and development, it appears likely 12. A. W. Ravin, The Evolution Press, 1965), p. 198.

8

of Genetics

(New

York: Academic

Biology and the Unity of Science that at least some role is played by the cytoplasmic organelles as well as by the DNA structure in the nucleus. Waddington reminds us: "In the whole cellular mechanism of differentiation DNA is only one part, and perhaps a rather inert one at that. We have to consider, at a minimum, the DNA, the chromosomal proteins, the mRNA, the ribosomes, the activating enzymes, the pools of amino acids, the transfer RNA's, and the various replicases, polymerases, degrading enzymes, enzymes attaching or detaching mRNA and DNA, or mRNA and ribosomes, and so on." 13 It is thus not easy to isolate the activity of any one constituent of the cell from the activities of other constituents: the influence of one on the others, unlike cases so commonly encountered in physics, is too great to be ignored. And if this is true of the individual cell, how much more is it the case with multicellular organisms and higher organisms adjusting to their environment! Cases of multiple and strong (i.e., non-ignorable) interdependencies in biology are so common and well-known that it seems superfluous to labor this point further. However, there is one moral to be drawn from this distinction which is of considerable importance for our present concerns: namely, that this greater "complexity" (for this is certainly the natural word to use here) that characterizes biology in contrast to physics is-given the biologist's interest in his particular problems-a result of the facts of nature. That the entities with which physics commonly deals can in a great many cases of interest to physicists by considered as "isolated" to such an extent that they can be dealt with in a simple (and usually mathematical) fashion, whereas this is not so easy in biology, is due to contingent facts of nature. In addition to the number and strength of the interacting factors, there is a further aspect of the influence of "complexity" on isolability in biology: those various interacting factors are also very different, as are the contributions of those factors to the situations under consideration. Again, this is unlike the case in physics, where we generally deal with interactions of very similar bodies-and where, for the same sorts of physical reasons as were noted above, the differences between the bodies can be ignored. Too, the types of interactions that are dealt with in situations in physics are far more uniform in character than those dealt with in many biological contexts. Once more we must note that the difference be13. C. H. Waddington, "Theoretical Biology and Molecular Biology," in C. H. Waddington, ed., Towards a Theoretical Biology (Chicago: Aldine Publishing Company, 1968), p. 107.

9

DUDLEY SHPERE

tween biology and physics in these respects is a matter of contingent facts. Yet a further type of "complexity"difference between biology and physics, different in many respects from the technique of isolation, arises as follows. In physics, even when we have isolated a group of entities from their environment in the way outlined above, it is often possible to put the situation through a further simplifying procedure, which I have elsewhere analyzed and called "idealization."As an example, consider the treatmentof bodies (particles) in Newtonian mechanics. There, the rationale for considering bodies as if they were "mass-points"is perfectly clear, consisting roughly of the following three considerations. First: there are certain problems to be solved-problems relating the positions, velocities, masses, and forces of bodies. Second: mathematical techniques exist for dealing with such problems if the masses are considered to be concentrated at geometrical points (viz., the quasi-geometricaltechniques of Newton, and, for later scientists, the methods of the calculus). Third: it is, as Newton showed, possible to treat ordinary bodies as if their masses were concentrated at their centers. Without entering into details, we can see that such considerations amount to showing that the use of the "mass-point"idealization is both possible and convenient for treating problems with which Newtonian mechanics is concerned. It is important, for our purposes, to note again that the reasons for introducing such idealizations-for it being possible and convenient to do so-are scientific, factual, in character. The universe could have been constructed in such a way that employment of the idealization would not have been possible; the errors resulting might have been too great. The possibility of using such idealizations without substantial error is partly a function of the facts of nature (as well as of the problem and its requirements,and also of the state of development of our mathematicaltools). Now, again, such idealization is not generally feasible in biology, at least to the extent to which it is in physics. And the sorts of reasons why it is not feasible in biology are the same as those that make it possible in physics. The factual conditions for fruitful and easy idealization are simply not present in the usual situations of interest in biology, whereas they are in physics. The universe might have been constructed in such a way that those conditions would have been present in biology; but it happens not to have been so constructed. Thus, in the case both of isolation and of idealization, what 10

Biology and the Unity of Science makes the application of these techniques to the solution of problems feasible in physics but not (generally) in biology are facts of nature. If, therefore, we say that it is difficult or perhaps even "impossible" to "isolate" systems, or to "idealize" them in biology, we must understand clearly what must be meant. It is not that an analysis of some antecedently given list of characteristic "biological" problems-in abstraction from the facts with which biology will deal-will show that the very attempt to apply those techniques to the solution of those problems is a self-contradictory enterprise. Any such alleged "impossibility" cannot be a "logical impossibility" in this sense. If isolation and idealization are impossible in biology, then it is because of the kinds of entities and interactions with which biology deals. "But," one might reply, "it is nevertheless impossible to isolate or idealize in biology in one very good sense: for the facts which militate against the employment of those techniques in biology cannot be expected to change." True; yet there is still more to be said about this alleged impossibility. For the possibility of isolating systems, or of employing idealizations, is a function not only of the subject matter being dealt with, but also of the problems under consideration and the techniques available. Let us consider the latter first. From the point of view of the questions of "reduction" of biology to physics, and or prediction in science, the most important technique to consider is the application of mathematics in idealizations.'4 And it is always possible that (a) ways will 14. There are, of course, other roles for mathematics in science besides its role in what I have here called "idealizations." There is, first of all, the use of statistics, which has become so important in biology. Further, there is the use of mathematics which results from the borrowing of concepts from other domains. Such borrowing (like idealization in the sense outlined here) usually takes place in full knowledge that we are not describing the entities with which we are concerned, or their behavior. Nevertheless, we use those concepts from other domains because there are some very general "structural" features of the biological situation which are open to such treatment. Examples of such borrowing are the use of concepts and techniques taken from game theory, decision theory, information theory, and computer science. Still a further use of mathematics arises when, in the total situation which is of interest to the scientist, it is found that most of the factors in the situation are either (a) irrelevant to the problem of concern, or else (b) explainable (caused) by a relatively small number of factors, so that the whole process being considered is an explanatory or causal function of that small number of factors. And often this function can be expressed in mathematical terms, even though the actual calculations involved might necessitate the employment of approximation techniques. Such simplification has been achieved in many cases in physics; and one might interpret the goal of

11

DUDLEY SHAPERE

be found of applying presently known mathematical techniques to construct idealizations of presently known biological subject matter, or (b) that new factual discoveries in biology will make possible the application of presently available techniques for the purpose of idealization, or (c) that new mathematical techniques will become available which will make idealizations in biology possible and useful. In the present paper, only alternative (a) will be considered-not because (b) and (c) are less interesting or important, but rather because they raise complex issues which are largely irrelevant to our present concerns; and, in any case, many of the major points of their relevance to present concerns will be covered in our discussion of alternative (a), to which we now turn. The formulation of useful mathematical idealizations which, however ingeniously applied, would consist essentially of applications of presently available mathematical techniques to presently known biological subject matter has been the program of a number of workers. In such efforts, Rashevsky was a pioneer. His attitude-at least at the outset of his worktoward the construction of idealizations in biology is reflected in the following passage, which closely parallels the discussion of "idealization" given above: In our study we should first start with the fundamental living unit, the cell. Following the fundamental method of physicomathematical sciences, we do not attempt a mathematical description of a concrete cell, in all its complexity. We start with a study of highly idealized systems, which at first may not have any counterpart in real nature. This point must be particularly emphasized. The objection may be raised against such an approach, because such idealized systems cannot be applied to real ones. Yet this is exactly what has been, and always is, done in physics. The physicist goes on studying mathematically, in detail, such nonreal things as "material points," "absolutely rigid bodies," "ideal fluids," and so on. There are no such things as those in nature. Yet the physicist not only studies them but applies his conclusions to real things. And behold. Such least within an application leads to practical results-at certain limits. This is because within these limits the real things have common properties with the fictitious idealized ones. Only a superman could grasp mathematically at once all the complexity of a real thing. We ordinary mortals must the DNA theory along these lines: to discover exactly how the activities of the cell, and perhaps the organism as a whole, can be understood in terms of the structure and behavior of DNA.

12

Biology and the Unity of Science be more modest and approach reality asymptotically, by gradual approximation.15 Rashevsky attempted to construct just such "idealizations" for mathematical use in biology, on the analogy of their employment in physics. Still, many biologists have felt uneasy about the results he achieved, and often for reasons which, though occasionally expressed in misleading ways, are reminiscent of the difficulties discussed above concerning the fruitful application of idealizations in biology. As one writer has commented on Rashevsky's work, "the biologist who has devoted great effort to examining some aspect of nature in all its richness and fullness often feels uncomfortable with the idealized system which fails to embody the details to which he has devoted so much effort." 16OFrom the standpoint of the analysis given above in the present paper, the terms "richness" and "fullness" are perhaps not the most appropriate to use here, and indeed are characteristic of the sort of misleading impression that can be conveyed: it is not simply the multiplicity and detail of the factors involved in biology, but rather that those factors play too strong and too relevant a part in the process under consideration, so that it is difficult to "abstract" from them (to "idealize" them) without omitting something that affects the situation in significant ways. And this is perhaps one source -not the only one-of the feelings of discomfort at Rashevsky's efforts to construct mathematical idealizations in biology. Similar remarks can be made regarding Waddington's attempts to formulate a mathematical concept of a "chreod." 17 This concept has, so far at least, been more suggestive than actually helpful in biology, either theoretical or practical. Once more, however, it is necessary to remind ourselves that the search for such idealizations is not a self-contradictory enterprise. It is made difficult, not "impossible," by the facts; but it also depends on the available tools of idealization (in the cases considered, mathematics) and, of course, by the ingenuity with which these are applied. It is important to understand this clearly, since critics of idealization in biology have often tended to misinterpret the legitimate point of their worries-to such an extent, even, that the not always admitted 15. N. Rashevsky, Mathematical Biophysics (Chicago: University of Chicago Press, 1938); quoted in H. J. Morowitz, "The Historical Background," in T. H. Waterman and Morowitz, eds., Theoretical and Mathematical Biology (New York, Blaisdell, 1965), pp. 30-31. 16. Morowitz (fn. 15 above), p. 31. 17. Waddington, "The Basic Ideas of Biology," in Waddington, Towards a Theoretical Biology, pp. 1-32; cf. also the comments by Ren6 Thom, ibid., pp. 32-41.

13

DUDLEY SHAPERE

implication of their view would be that the formation of general class-concepts is or should be alien to biology. For example, we saw above the worry that idealization omits the "richness" and "fullness" with which the biologist is concerned; and other writers, among them Professor Simpson, elevate this worry into a criticism of a failure to come to grips with the "uniqueness" of individuals. And then this concern with "uniqueness" is built into the very nature, the methodology, of biology, and is made the basis of a contrast between that subject and the physical sciences. Thus Professor Ernst Mayr has said: In the uniqueness of biological entities and phenomena lies one of the major differences between biology and the physical sciences . . . If a physicist says "ice floats on water," his statement is true for any piece of ice and any body of water. The members of a class usually lack the individuality that is so characteristic of the organic world, where all individuals are unique.18 The point here, again, is not simply that biology is concerned with "unique individuals" (which would suggest the absurdity that biology is not concerned with classification), but rather that, in being required to consider a multitude of strongly relevant factors, biology is generally constrained, by the facts with which it deals, to forego such techniques as "isolation" and "idealization" in the senses outlined above. And because the possibility of applying such techniques is dependent not only on those facts, but also on the state of development of the techniques available, it does not follow from this valid point that biology and the physical sciences must necessarily be distinguished by the former's concern with "uniqueness." These remarks also have ramifications for the notion that prediction is impossible in biology, or at least relatively insignificant. There is, as usual, an important element of truth in this notion, which has gone astray in the telling. A less misleading way of putting the point, perhaps, would be that, because of the number of factors involved in many problems of interest to biologists, and because of the complexity of their interactions, accurate prediction is difficult. If the problem at hand requires a considerable degree of predictive precision, such precision will in general not be attainable-but it will be unattainable only because of the complexity of the situation, not because (e.g.) of the "essentially historical" character of biological explanation.'l 18. E. Mayr, "Cause and Effect in Biology," in Waddington, ibid., p. 52. 19. Simpson's claims about the absence of laws and predictions in

14

Biology and the Unity of Science We have found, then, that, although the physical and biological sciences are distinguished from one another with regard to the extent of applicability of such techniques as idealization, nevertheless, this difference does not imply that biology cannot employ these techniques. For ingenuity may discover ways of applying presently available techniques to the construction of idealizations of biological subject matter. Consideration of possibilities deriving from new factual discoveries or mathematical developments would strengthen the same conclusion. But it will also be recalled that the possibility of employing such techniques as idealization is a function not only of the subject matter and the techniques available, but also of the problems under consideration. Often, in the history of science, we find that a major step is taken in the development of a subject when the problems themselves are refined so as to permit the application of precise techniques to their solution. This is the case with idealization: again, use of that technique may be made feasible by reformulation of biological problems in such a way that they may be conveniently and effectively approached via those techniques. III Proponents of the "autonomy" of biology have called attention to many features of their science which do exist, and which mark differences between it and the physical sciences. Thus, for example, biology does deal with a subject matter which is "complex"; and biology does characteristically deal with that subject matter in a more "concrete" way (in its "richness" and "fullness") than does physics. Yet these differences have tended to be misunderstood, and it has been the purpose of the preceding section to clarify and assess the import of those differences, which have lain at the basis of the controversies about the relations between the two subjects. I have argued that these differences arise fundamentally from the fact that biology deals with entities and interactions whose character makes biology, and the "historical" character of biological explanation, are based largely on a failure to observe the distinction between laws on the one hand, and initial and final conditions on the other. This failure is clearly evident in the following passage: "It is a law that states the relationship between the length of a pendulum-any pendulum-and its period. Such a law does not include the contingent circumstances, the configuration, necessary for the occurrence of a real event. . . . If laws thus exclude factors inextricably and significantly involved in real events, they cannot belong to real science" (Simpson, This View of Life, p. 128). When this distinction is taken into account, it is clear that physical laws can be (as they in fact are) applied to "historical" sequences of events, and in particular can be applied to sequences of biological events.

15

DUDLEY

SHAPERE

difficult (not "impossible") the application of techniques which are employed easily and fruitfully in physics. Employment of those techniques is, however, not logically precluded in biology, since ingenuity, factual discoveries, mathematical innovations, and refinement of problems can all make useful application of them feasible. Thus, when properly understood, these differences do not justify the extravagant conclusions that have often been drawn from them, regarding, for example, the "uniqueness" of biological subject matter, the "essentially historical" character of its explanations, and, in general, the irreducibility of biology to physics. However, there are two further important features of the traditional discussions of "reducibility" that have not yet been touched upon. First, are there not some general kinds of problems with which biology deals, and with which physics does not, and does not the existence of these special problems imply that biology cannot be completely reduced to physics? And, second, if biology does not utilize certain techniques important in physics, and if it need not utilize them, does this not imply a fundamental difference between biology and physics? The first of these questions has to do with Simpson's questions, "How come?" and "What for?" which, it is alleged, are asked in biology but not in physics. Whether or not they are asked in physics need not be determined here; for their occurrence in biology provides no ground for the assertion that biology and physics cannot be unified. Since we have already examined the question "How come?" 20 we may confine our attention to "the dreadful teleological question," "VVhatfor?" The occurrence of functional explanation in biology need not be any more mysterious than any other sort of question. The employment of "What for?" questions is a manifestation of the biologist's interest in certain sorts of facts about the entities and processes involved-namely, in their adaptation and survival, and in the mechanisms (e.g., of homeostasis) by which this is achieved. (Of course, there need not be only one criterion of adaptive success, and there may be alternative mechanisms by which such success is achieved.) Once this interest is declared, it is only natural to ask how the various constituents of those entities or systems are related to-contribute to-that adaptation. And, in one very ordinary sense of the word, the answers to such questions constitute "explanations." But the important problem from our point of view is, Does the existence of such questions, and 20. On the question, "How come?" in addition to the discussion

prediction in biology, found in sec. II of this paper, see also fn. 19.

16

of

Biology and the Unity of Science such explanations, imply some radical and unbridgeable gap between the biological and the physical sciences? Certainly not: the fact that a certain organ (for example) contributes to the survival of a biological entity (or species) no more precludes the giving of a physicochemical account of the properties and behavior of that organ, or of the organism as a whole, than the fact that oxygen contributes to our survival precludes our giving a physical account of the properties and behavior of oxygen. Indeed, the evidence that such "reductionist" explanations can be given is very strong, considering the amount that has been explained already in those terms. (And note, too, that that evidence is factual in character: belief in the unity, or unifiability, of science is not-and need not be at all-based merely on some metaphysical faith in the "simplicity" of nature.) The second question-whether, if biology does not, and need not, utilize certain important techniques of physics, the former is irreducible to the latter-may also be answered in the negative. The question of "reducibility," or, more generally, of the possibility of giving a unified treatment of physical and biological phenomena, has to do, in its standard current form, with the question of whether the character and behavior of biological entities can be accounted for in terms of the character and behavior of physical entities. That physicists (and chemists) typically deal with their entities not directly, but in terms of idealizations (and similar sorts of "constructs"), is irrelevant to the question of "reduction." For such techniques as idealization are utilized for other purposes than the representation of reality. If the physicist speaks of pointmasses, point-charges, perfectly rigid bodies, and so on, he is fully aware, on the basis of physical considerations, that reality does not consist of such entities.2' And, of course, "reduction" would have to do with establishing relationships between the character and behavior of the entities as they really are. Thus, even if such techniques as idealization were not, and did not need to be, employed in biology, that fact would not bear on the question of "reducibility"or "unifiability." But the point made here does say something about what should be involved in any adequate analysis of the concept of "reduction"; for if physics sometimes deals with its entities through idealizations, this shows that any reduction of biology to physics would-as we should have expected all along, despite what philosophers (and many scientists) have had to say -have to consist not simply of a direct, straightforward "de21. Cf. sec. U of the present paper.

17

DUDLEY SHAPERE

duction" of a description of the character and behavior of biological entities from one of physical entities, but will involve idealizations, approximations, simplifications-a host of indirect connections. Any full understanding of what is involved in "reduction" would have to include analyses of such techniques. And this brings us to one final point: Simpson's claim that "the life sciences . . . penetrate much farther into the exploration of the universe that is science than do the physical sciences," that the former are, "in a sense, even more" scientific than the latter. It is of course true, as we have seen, that biology generally deals with its subject matter more concretely -less in terms of idealizations, for example-than does physics. If this is a mark of superiority, it is gained at a high price; for the abandonment of idealization carries with it significant losses in power of treating a subject-matter. And on the other hand, just as biology is not logically prevented from recouping these losses through introduction of idealizational techniques, so, also, physics is not precluded from giving more concrete accounts of nature by introducing more and more realistic descriptions.

18

TheoriesandExplanationsin Biology KENNETH F. SCHAFFNER Department of Philosophy University of Chicago, Chicago, Illinois

INTRODUCTION No biologist today would dispute the claim that "molecular biology" has made significant contributions to our understanding of living organisms. The chemical structures of a number of "biological" molecules have been elucidated and used to explain their biological properties. The area of molecular genetics has developed a theory of heredity built around the WatsonCrick DNA structure and the "central dogma" of protein synthesis, which has led one distinguished molecular biologist to state that "There have been only two great theories in the history of biology that went more than a single step beyond the imediate interpretation of experimental results; these were organic evolution and the central dogma." 1 It can be said with respect to genetics that there are good reasons to believe that scientists are well on their way to developing a complete reduction of a biological discipline to chemistry and physics.2 It is not only in genetics that the molecular approach, with its use of physical and chemical theories, is fruitful; other areas of biology such as taxonomy and neurology have also found physical and chemical analyses and theories both useful and, in some cases, necessary. But biology has formulated nonchemical types of theories in its past and continues to do so in the present in many sectors of physiology, neurology, evolutionary theory, and even genetics. An interesting question 1. G. S. Stent, "That Was the Molecular Biology That Was," Science, 160 (1968), 390. 2. My paper, "The Watson-Crick Model and Reductionism," read to the 155th National Meeting of the American Chemical Society, April 1968, San Francisco, and which has been submitted for possible publication to the British Journal for the Philosophy of Science, outlines the history and logic of this reduction.

19

KENNETH F. SCHAFFNER

is accordinglyraised: what is the status of chemical and nonchemical theories in biology?B This question is closely related to another: what are the types of explanation which are given in biology? This paper will pursue aspects of these questions, and will attempt to show that it is a unification of the molecular and the "classical"approaches that is most fruitful from the viewpoint of theoryconstructionand explanationin biology. I shall understand the term "theory"to designate the conjunction of a set of sentences purportedly descriptive of the action of theoretical entities (or processes) and another set of sentences, sometimes called correspondence rules, which link the action of these entities with experimental-and, possibly, other theoretical-situations. These two sets of sentences, taken conjointly, explain these latter experimental (and/or theoretical) states of affairs. Typical examples of such theories would be Dalton's atomic theory, Maxwell's electromagnetic theory,quantummechanics, and statistical mechanics. This account roughly amounts to accepting the standard analysis of "theory"-but with certain important modifications, among them, different notions of the meaning of theoretical terms and correspondencerules.4 I shall understand by "explanation"either the deductive nomological type of explanation or the inductive (also deductive) statistical type of explanation.5 The former type, sometines called the "covering law" type, explains an event by showing that a sentence describing the event to be explained (the explanandum) is derivable from a confirmed general law(s) plus appropriateinitial conditions. The laws and initial conditions constitute the explanans. As an example of such explanation, consider the explanation of the motion of a pendulum by Newtonian mechanics. Clearly, this account can be generalizedto explain laws as well as events. The inductive statistical form of explanation makes an event probableby referring it to certain statistical 'laws" and initial 3. The word "chemical" is used here and elsewhere in this paper as shorthand for "chemical and physical." The expression "nonchemical" refers to theories, laws, or descriptive language which do not employ accepted chemical and/or physical terminology. 4. See E. Nagel's The Structure of Science (New York: Harcourt, Brace, & World, 1961), chaps. 5 and 6, for an account of the standard analysis. My paper, "Correspondence Rules," read to the 1st National Meeting of the Philosophy of Science Association, October 1968, outlines some of the ways in which my account differs from the standard analysis. 5. See C. G. Hempel's essay "Aspects of Scientific Explanation," in his book Aspects of Scientific Explanation (New York: Free Press, 1965) for a detailed discussion of these issues.

20

Theories and Explanations in Biology conditions. The explanation of an electron diffraction pattern by quantum theory would be such a case. These accounts of explanation are old work horses in the philosophy of science, and though they have been extensively criticized, as well as developed and defended, I find them quite clear and convincing on the whole and will be implicitly referring to them throughout the paper. It will turn out under this analysis that reduction can be considered as the explanation of one theory by a different theory, usually drawn from a different science. I shall begin the analysis of theories and explanations in biology by examining the relations between classical and molecular genetics and indicating some of the rather distinctive features of the reduction of biology to chemistry. I shall then consider the role of "purely biological" theories and language, and conclude with an application of various notions developed in this paper to an as yet unanalyzed area of biology. THE CHEMICALEXPLANATION OF HEREDITY An all too brief, partly historical, partly logical, reconstruction of the development of a chemical account of genetics will serve to introduce some general ideas about the role of physics and chemistry in biology.6 Genetics in the form given to it by Mendel and Morgan is a "biological" theory of heredity. It postulates the existence of hypothetical entities-genes-which are causally responsible for phenotypic characteristics. The chromosomal theory of inheritance makes an attempt to cytologically explain Mendelian and post-Mendelian genetics, and does not involve non-biological chemical terms and laws. Experiments on the transformation of virulence characteristics, first performed by Griffith in 1928, and later followed up and more carefully analyzed by Avery, MacLeod, and McCarty, and Hershey and Chase, indicated that deoxyribose nucleic acid (or DNA) was the transforming chemical.7 DNA, 6. A more complete account of the historical topics to be referred to here can be found in H. L. K. Whitehouse's Towards an Understanding of the Mechanism of Heredity (New York: St. Martin's Press, 1965. The paper referred to in fn. 2 above is also relevant here, as the account given in this paper is abstracted from the earlier paper. 7. F. Griffith, "The Significance of Pneumococcal Types," J. Hyg., 27 (1928), 113; 0. T. Avery, C. M. MacLeod, and M. McCarty, "Studies on the Chemical Nature of the Substance Inducing Transformation of Pneumococcal Types. .. ," J. Exp. Med., 79 (1944), 137; A. D. Hershey and M. Chase, "Independent Functions of Viral Protein and Nucleic Acid in Growth of Bacteriophage," J. gen Physiol., 36 (1952), 39.

21

KENNETH F. SCHAFFNER

accordingly, became an excellent candidate for the chemical basis of the gene. DNA was investigated from a number of points of view by biologists and chemists prior to the discovery of its chemical structure by J. D. Watson and F. H. C. Crick.8 Watson's and Crick's discovery was remarkable because of the biological which they themimplications of the structure-implications selves were quick to note. In a paper published very shortly after their classic paper announcing the structure of DNA, Watson and Crick indicated that their structure for DNA suggested "how it might carry out the essential operation required of a genetic material, that of exact self-duplication." 9 Their proposal, that the double helix uncoils and forces the synthesis of a complementary chain alongside of each separated helix, has been supported by a significant amount of experimental data. 10

In their paper, Watson and Crick also noted another most important consequence of their "model": "[Though] the sugar phosphate backbone of our model is completely regular . . . any sequence of the pairs of bases can fit into the structure. It follows that in a long molecule many different permutations are possible, and it therefore seems likely that the precise sequence of bases is the code which carries the genetical information." The same paper also contained the suggestion that "spontaneous mutation may be due to a base occasionally occurring in one of its less tautomeric forms." 11 A combination of experimental support and theoretical development of these implications has led to the "cracking" of the genetic code and an explication of the "mechanism" by which the information coded in the DNA directs protein synthesis. The auto- and hetero-catalytic properties of DNA now have broad experimental support, and the so-called "central dogma" of protein synthesis, according to which the DNA codes RNA to direct protein synthesis, is a well-confirmed hypothesis.12 8. For example, see E. Chargaff's paper, "Chemical Specificity of Nucleic Acids and Mechanism of their Enzymatic Degradation," Experentia, 6 (1950), 201; J. D. Watson and F. H. C. Crick, "Molecular Structure of Nucleic Acids: A Structure for Deoxyribose Nucleic Acid," Nature, 171 (1953), 737. 9. J. D. Watson and F. H. C. Crick, "Genetical Implications of the Structure of Deoxyribose Nucleic Acid," Nature 171 (1953), 964. 10. See Schaffner, "The Watson-Crick Model." 11. Watson and Crick, "Genetical Implications," p. 965. 12. Recently Dr. Barry Commoner cast doubt on the "central dogma" by marshaling experimental evidence which, among other things, suggested that DNA polymerase plays a role in providing or altering some information that is translated into protein sequences. Whether this effect

22

Theories and Explanations in Biology Progress toward reduction frequently necessitates changes in the reduced science, and sometimes in the reducing science as well.'3 In genetics, such changes were occasioned more by the increased precision of genetical methods, as Demerec, Pontecorvo, and Benzer developed a "fine structure genetics." 14 As a result of the increased "resolving power" of the genetical methods of counting and crossing, Benzer was led on the basis of experiments to introduce new "definitions" of the "gene"definitions which permitted the identification of the modified classical gene with a DNA sequence.'5 The gene, now understood (in one of its senses) as a functional unit or "cistron," came to be known to direct, via RNA, the synthesis of a polypeptide chain composed of linked amino acids. Such a polypeptide chain is a protein. About the same time, Francis Crick suggested that "the amino acid sequences of the proteins of an organism are the most delicate expression possible of the phenotype of an organism." 16 What had been termed "unit characters" by Mendel came now to be understood-in much more sophisticated ways to be sure-as proteins, with their primary, secondary, tertiary, and quaternary structures dictated by the DNA in the chromosomes. The colinearity, or one-to-one correspondence, of the amino acid sequences in proteins, the nucleotide sequence in the DNA, and the sequence in the fine-structure genetic maps, has been established by the work of Kaiser and Yanofsky.'7 There is little doubt today that we have a good account of the chemical basis of heredity and some important clues, based on the work of Jacob and Monod, as to the chemical mechanisms of growth and differentiation.'8 is "noise" or due to impurities in the chemicals, or whether it requires a new, more complex, "dogma" is not as yet clear. Commoner's paper was delivered at the symposium referred to in fn. 2 above. 13. See my "Approaches to Reduction," Phil. Sci., 34 (1967), 137, for examples of such changes. 14. See E. A. Carlson's account of this development in his book The Gene: A Critical History (Philadelphia: W. B. Saunders, 1966), chap. 22. 15. S. Benzer, "The Elementary Units of Hereditary," in The Chemical Basis of Hereditary, ed. W. D. McElroy and B. Glass, (Baltimore: John Hopkins Press, 1957), 70. 16. F. H. C. Crick, "On Protein Synthesis," Symp. Soc. Exp. Biol., 12 (1958), 138. 17. A. D. Kaiser, "The Production of Phage Chromosome Fragments and their Capacity for Genetic Transfer," J. Mol. Biol., 4 (1962), 287; C. Yanofsky, et al., "On the Colinearity of Gene Structure and Protein Structure," Proc. Nat. Acad. Sci. U.S.A., 51 (1964), 266. 18. See F. Jacob and J. Monod, "Genetic Regulatory Mechanisms in the Synthesis of Proteins," J. Mol. Biol., 3 (1961), 318; J. Bonner, The Molecular Biology of Development (New York & London: Oxford University Press, 1965).

23

KENNETH F. SCHAFFNER

THE LOGICOF REDUCTION IN BIOLOGY Attention to the logic and methodology of molecular genetics reveals some important implications for explanation and theory construction in biology. A number of authors, including myself, have written on logical problems of reduction in general;19 in this paper I want to concentrate on certain features of reduction which have significance for explanation in biology. It is well known that some connections must be made between the primitive terms of a reducing theory and those of the reduced theory.20 In the case of molecular genetics, when one attempts to construct what I have termed "reduction functions" for entity and predicate terms connecting biology and chemistry, it is important to note the place of "organization" in the explanans, or reducing theory. The chemical laws, chemical theories, and known (even including perhaps all empirically knowable) initial conditions stated in general chemical terms, do not suffice to chemically account for the existence of the chemically characterized referents of the biological entity terms. An example, even though it is somewhat artificial, will help to make the point clearer.21 Consider the biological entity term "gene," or "cistron," for, say, the brown pigment in eyes. The chemical description of this entity will be a DNA sequence of several hundred nucleotides. The important word here is sequence-for if the order of the nucleotides is disturbed or altered, sometimes by as little as one nucleotide, the DNA sequence may no longer be capable of directing the synthesis of the pigment in the brown eyes. Even if the DNA sequence were affected in such a way as to produce only a simple codon deletion of three nucleotides, such a 19. See E. Nagel's The Structure of Science, chaps 11 and 12; J. Kemeny and P. Oppenheim, "On Reduction," Phil. Studies, 7 (1956), 6; P. K. Feyerabend, "Explanation, Reduction, and Empiricism," in Minnesota Studies in the Philosophy of Science, 3, ed. H. Feigl and G. Maxwell (Minneapolis: University of Minnesota Press, 1962), p. 28; and Schaffner, "Approaches to Reduction." 20. Except in such cases as Nagel's "homogeneous reductions." See E. Nagel, The Structure of Science, p. 339. 21. The significance of "organization" in reductions of biology to chemistry has been a point of contention between myself and some other authors who have written on this topic. See my "Antireductionism and Molecular Biology," Science, 157 (1967), 644, and the exchange of letters in Science, 158 (1967), 857-862. Michael Polanyi has also treated this problem in his "Life Transcending Physics and Chemistry," Chem. Eng. News 45 (1967), 54, and more recently in "Life's Irreducible Structure," Science, 160 (1968), 1308. Polanyi has drawn radically different conclusions from mine from the fact that organization must be assumed in some molecular-type explanations.

24

Theories and Explanations in Biology single change in, for example, the amino acid sequence of the synthesized protein, could result in a radically different folding, or tertiary structure, with consequent loss of enzymatic properties. As Watson and Crick were quick to point out in their second paper on the structure and implications of DNA, "any sequence of the pairs of bases can fit into the structure [of the DNA]" (my italics). What determines such a specific sequence is a previous identical (or near-identical) sequence. (Of course in certain cases we want to allow for complementary sequences and also for the interchange of uracil and thymine in the RNA-DNA transcriptions.) Accordingly, in addition to specifying the chemical constituents or molecules in setting up connections between biological and chemical theories, the order of these constituents must be specified. Such specific primary structures as biologists work with are the outcome of long and empirically inaccessible evolutionary stories in which ultraviolet radiation, Brownian motion, other mutagenic agents, and natural selection have played important parts. Reconstructing in detail-in specific detail-the chemical account of the processes by which these structures developed seems to be a hopeless task, though interesting plausibility-type arguments such as are to be found in connection with developments of the Oparin-Haldane hypothesis about the origin of life and its early development can make such accounts more credible.22 The difficulty, then, is that the reductiandum in biology has, chemically speaking, an organization which is describable in chemical terms, but not explicable in chemical terms without making use of a hypothesis of historical evolution: of random processes and natural selection. This can be contrasted with, for example, the describable and explicable formation of the homopolar chemical bond, important in the reduction of chemistry to physics. For in this case, physics can indicate under which empirically accessible and testable conditions atoms will form molecules. Accordingly, the molecular biologist who would explain a brown pigmented eye will redescribe the brown pigment in terms of the protein sequence optically responsible for the brown color, and trace the formation of that protein through the complex interactions of enzymes and the RNA's in the ribosomes to the original information encoded in the primary sequence of the DNA. He must take this DNA sequence as 22. See A. I. Oparin's Life: Its Nature, Origin, and Development,

A. Synge (New York: Academic Press, 1964), chaps. 2, 3.

trans.

25

KENNETH

F. SCHAEFNER

given, as part of his initial conditions. His physical and chemical theories do not, without the evolutionary account referred to above, account for the organization he employs in his explanans or reductians, though of course the physical and chemical theories are used as part of the explanans. The other part of the explanans is a sentence describing the primary structure of the DNA. Incidentally, the role of organization of chemical molecules and systems in biological explanation can be underscored by reference to a very elementary observation, namely, that no physical or chemical law is violated as an organism dies and decomposes. The physical and chemical laws are the same in both living and dead biological organisms. Nevertheless, because organization has to be taken as given in such cases, this does not entail that the organism's organization is the result of occult forces or some Designer's influence. All positive evidence, especially that which is connected with developments of the Oparin-Haldane hypothesis and the study of spontaneous synthesis and polymerization of amino acids under primitive earth conditions, suggests that "random" molecular association may produce amino acid polymers that undergo coacervation, or a similar process. These primitive protein systems can be treated as evolvable systems that are influenced by natural selection, and which may well have produced the first "biological" systems. The fact that there is a historical aspect to a complete chemical explanation of a living organism does not warrant the assertion that molecular biological explanations which take a primary structure as given are not good explanations. Newtonian mechanics explains the behavior of the planets (if we neglect the small relativistic corrections). Employing initial conditions describing position and momenta of the planets and the sun, it predicts and postdicts. Nevertheless, it does not give an account of the origin of the solar system, without supplementing it with a historical story of a wandering star passing near the sun or a nebular evolutionary hypothesis. We must be careful to avoid what might be termed an "inverse genetic fallacy" in connection with biological explanations. The ordinary genetic fallacy mistakes an account of the origins of X for an explanation and justification of X; the inverse genetic fallacy assumes that without an account of the origin of X, X's present and future activity cannot legitimately be accounted for. Once the organization of an organism on the chemical level has been taken into account, there is no evidence that chemical laws and theories, supplemented with whatever physical the-

26

Theories and Explanations in Biology ories are required, such as thermodynamics and quantum mechanics, are not fully adequate to explain biological processes. This is not to say that there are no biological phenomena that are not now explained in chemical terms; it is to say that there is no positive evidence that chemical explanations will not be possible for all biological phenomena. As Stent .aoted recently, no paradoxes and no new laws of physics have arisen from the careful study of the (genetic) molecular processes of living organisms.23 From the above considerations it should be clear that the molecular biologist and the evolutionist are not providing competing explanations of biological organisms. Rather, my account suggests that the molecular biologist needs the evolutionary account to make his chemical approach complete. It will become clearer below that, in addition, the evolutionist also needs molecular biology, and that both the evolutionist and the molecular biologist need, though for more pragmatic reasons, the classical biologist. Some of the more organismically inclined biologists such as G. G. Simpson and E. Mayr24 have asserted similar theses as regards the importance of organization and history: for example, that the historical aspect is seen as the source of the organization. Where I differ with these claims is that I can see no arguments that imply that a chemical system could not have evolved. Accordingly my position is reductionistic, whereas the organismic biologists claim to be nonreductionistic or even antireductionistic. IMPLICATIONSOF THE REDUCTION OF GENETICS Mendelian and Morganian genetics are biological theories, in the classical tradition, employing nonchemical terminology and ostensibly nonchemical entities such as genes. The methodology is patently "biological": it involves taxonomical classification, breedings that allow cross-fertilizations, induced mutations, and the counting of numbers of progeny types. Approached in such a manner, changes in the phenotype of the organisms can be associated with survival value in a given environment and the results of evolutionary investigations brought to bear on the examined organisms and species. Likewise, the genetical theories can be used to enrich and deepen evolutionary explanations of species survival and modification. 23. Stent, Science, 160, 394. 24. G. G. Simpson, This View of Life (New York: Harcourt, Brace and World, 1964); E. Mayr, "Cause and Effect in Biology," Science, 134 (1961), 1501.

27

KENNETH

F. SCHAFFNER

It seems clear that prima facie "purely biological" laws and theories can be and have been discovered.25 Furthermore, rather than turning out to be rival accounts to chemical accounts of biological processes, the existence of these biological theories has been of significant value in the search for chemical explanations of biological processes. The transformation of biologically characterized properties by DNA, in the experiments referred to earlier, suggested that DNA might be the genetic material. The "one [biologically characterized] gene=one [chemically characterized] enzyme" hypothesis formulated in 1941 by Beadle and Tatum26 paved the way for the chemical theory of protein synthesis. Benzer's fine structure genetics, utilizing very sophisticated but still nonchemical methods of crossing and counting phages, assisted in the revision of genetic theory and the establishment of the colinearity hypothesis. The development of the chemical theory of inheritance indicates that biological theories suggested what to look for and where in the bewildering chemical complexity of the cell and organism. Biological descriptions and theories in one area also provide connections between chemical descriptions and processes in the same area and biological descriptions and processes in another sector of inquiry. How this works I shall attempt to show in the next section. At the moment I wish to support the contention that biological theories which select nonchemical features of living organisms and attempt to interrelate these nonchemical features and to provide explanations of some in terms of others (some of which may be 25. Though I have great sympathy with a position developed by J. J. C. Smart in his book, Philosophy and Scientific Realism (New York: Humanities Press, 1963), to the extent that his position is antireductionistic, I disagree with his contention that there are no biological laws. Smart proposes that although there are laws of physics and chemistry, the "laws" of biology are really "generalizations" much as one can find generalizations in engineering texts. But if it turns out that many physical and chemical laws are in a similar position to engineering "generalizations"_ and this indeed seems to be the case: consider Kirchhoff's laws in current the supposed difference theory, or chemical laws in a plasma world-then is a matter of degree. In the light of the widespread, perhaps universal, applicability of genetics and the genetic code, and in the absence of any persuasive extraterrestrial evidence in favor of Smart's thesis, I subscribe to the position that there are biological laws in the same sense that there are chemical laws, though both are reducible to physics. For a biologist's viewpoint on such biological laws, see D. E. Green and R. F. Goldberger, Molecular Insights into the Living PTocess (New York: Academic Press, 1967), especially the Preface and chaps. 15 and 16. 26. G. W. Beadle and E. L. Tatum, "Genetic Control of Biochemical Reactions in Neurospora," Proc. Nat. Acad. Sci. U.S.A., 27 (1941), 499.

28

Theories and Explanations in Biology biologically theoretical), are not only possible, but that they have existed and have been most useful in assisting scientists seeking out physicochemical accounts of organisms.27 Just how "biological" theories can pragmatically assist in the development of "chemical" accounts of biological processes, and why they are not rival accounts, can be seen in the history of genetics. The development of the chromosome theory as an explanation of the "mechanism" of the linked genes focused attention on the chromosome as the locus of the heredity determiner. Chemical analysis of the chromosomes led to the suspicion that it was the protein, or possibly the protein and the DNA, which transmitted heredity. But the transformation experiments noted above clearly indicated that it was the chromosomal DNA which was the genetic material. Thus the biological or cytological chromosomal theory of inheritance functions as an important stage in the development of a chemical theory of inheritance. The existence of biological theories does not mean that these theories are rival accounts to the chemical explanation of phenomena in the same domain. The logical status which these biological theories have is not different from that which certain "phenomenological"-type theories have in physics. But "phenomenological" is a misleading word here, a name and not a description, and it seems better to follow Einstein and Poincare and call such theories, as can function independently and yet be explained on a deeper level by underlying mechanisms, "theories of principle." 28 "Phenomenological" thermodynamics and Maxwell's theory of electromagnetism (as applied to macroscopic material bodies) play such roles with respect to statistical mechanics and the Lorentz electron theory (and later quantum electrodynamics), and they stand in a well-confirmed reduction-relationship to these "deeper" theories. Further analogies with physics are also present: the existence of such theories of principle assisted both Boltzmann and Lorentz in their search for their "deeper," or in Einstein's terminology, "constructive," theories. Furthermore, these constructive theories specify "mechanisms"-in a very broad sense of the term-which explain regularities that are expressed in the language of the theory of principle. Similarly, with respect to biological theories, physics and chemistry function 27. To a certain extent, classical biological language seems to function as the "observation language" of molecular biology. 28. A. Einstein, "What is the Theory of Relativity," in his The World As I See It (New York: Covici-Friede, 1934), pp. 74ff; H. Poincar6, The Foundations of Science, pt. II: The Value of Science (New York: Science Press, 1929), chap. 7.

29

KENNETH

F. SCHAFFNER

to provide the "constructive" theories, which in turn provide "mechanistic" explanations for biological phenomena (including regularities). An interesting question which is raised by the possibility of having these theories of principle in biology is what a reduction to the chemical would add to our knowledge. The question is similar to what the theory of statistical mechanics adds to our thermodynamical knowledge. Beyond introducing an aesthetic element which is bound up with the "unity of science," it seems that such reductions often do add signfficant information, though, to be sure, there are often problem areas in the reduced theory which are not much affected by a reduction. Just as telescope manufacture is largely unaffected by Maxwell's electromagnetic theory of light, as in the reduction of optics to Maxwell's theory, so some applications of genetical theories of plant-breeding may not be affected by knowledge of DNA's structure and the "central dogma." But in many areas of the sciences which are brought together, there are important consequences. It is likely that there will be a sharpening of the limits of applicability of the reduced theory, and an understanding of the "mechanisms" by which certain processes as characterized by the reduced theory are brought about. There is in many cases an increase in the precision of predictability, as many interfering factors can now be better understood, and there is, most likely as a consequence, the possibility of added control over natural processes. Reduction often permits the possibility of new experiments to test both reduced and reducing theories; for example, the Michelson-Morley interferometer experiment in optics tests any electromagnetic theory of moving bodies. TEMPERATUREADAPTATION OF ORGANISMS There are three general theses which I want to propose in connection with explanation and theory construction in biology: (1) "biological" language is useful, and in certain cases biological theories are constructable within, or with some new nonchemical additions to, this biological terminology; (2) these biological descriptions and theories are reducible to (or in some cases replaceable by) chemical theories if the "organization" of the organism's chemical constitution is added to the explanans; and (3) the biological laws and theories (comprising classical and evolutionary theories) and the chemical laws and theories mutually assist one another in biological explanation.

30

Theories and Explanations in Biology These theses can be explored by considering an area of biological research different from genetics, but in which some of the knowledge obtained from genetic and molecular genetic investigations can be applied. Considerable research has been done in the last decade on physiological adaptation of temperature-maintaining "mechanisms" in biological organisms. This sector of inquiry seems particularly suited for exploration in connection with problems raised in earlier discussion because (a) these biological investigations have been conducted on bacteria, insects, plants, and animals, thus covering large classes of living organisms; (b) measurable results have been obtained: specific temperature ranges have been determined and some molecular characteristics of the adapting organisms have been studied; and (c) there is the possibility that the logic of homeostatic processes developed in connection with temperature regulation by Ernest Nagel and others29 might admit of future application in this area. Much of the recent work in the area of adaptation of temperature regulating mechanisms was surveyed and summarized at a AAAS symposium in late 1965.30 C. Ladd Prosser, the editor of the volume, has included an article in which he summarizes the import of contributions by botanists, zoologists, bacteriologists, and biophysicists. From research in this area we know, on the basis of physiological and metabolic data, that the temperature-regulating mechanisms of organisms are affected by shifts in the environment's average and extreme temperatures. These mechanisms adapt to the shifts. That this adaptation is significant from the point of view of species evolution should be clear from the way in which natural selection operates. Since many biologists believe that organisms are nothing more than chemical systems, to investigate the chemical mechanisms that must underlie the adaptive physiology and species evolution is an important task. Prosser indicates what the over-all strategy of formulating explanations in this area is to be as follows: Since all of the genetically transmissible information is in certain macromolecules (semantides), it is necessary to trace the molecular connections or pathways (i) between an adaptive phenotypic character and the genes that control it; and (ii) between environmental stresses and the expres29. Nagel, The Structure of Science, pp. 411-418. 30. C. Ladd Prosser, ed. Molecular Mechanisms of Temperature Adaptation (Washington, D. C.: Amer. Assn. Adv. Sci., 1967).

31

KENNETH

F. SCHAFFNER

sion of the character. Adaptive variations at various levels of organizational complexity must ultimately be explained in molecular terms.31 Clearly there is a commitment to reductionism and to the importance of chemical explanations. Prosser, however, also notes that analysis of the protein structures of organisms in chemical terms is not sufficient. He asks: How can reactions to temperature help us to understand a molecular basis for speciation? The differences in proteins between different kinds of animals, whether they are seen in electrophoretic pattems, in amino acid analyses, or by immunological techniques, are not likely to give much information about speciation until they are related to adaptive function [my italics]. Analyses of base sequences in RNA and DNA are useful in tracing relationships, but they can be effective in speciation only by the end-products for which they are templates.32 I might add that descriptions of the "end-products" have to be in a language such that the end-products can be seen to be relevant to survival value in an environment, and that classical and evolutionary biological classification language seems useful and relevant in this regard. Description of these end-products in protein language is, however, important when a chemical mechanism of the survival feature is wanted. Prosser notes that "flagellar proteins of thermophilic bacteria are heat-stable, apparently because of stable hydrogen bonding, since they are also resistant to hydrogen bond breakers. Primary protein structure rich in aspartic and glutamic acids appears to provide the basis for natural selection in thermophilic bacteria." 33 The primary structure of the proteins is of course based on the sequential order of the nucleotides in the organism's DNA. This, as I noted above, must be taken as an initial condition in chemical explanations in biology. Prosser implicitly agrees with this, noting that "critical properties of proteins, such as temperature of denaturation melting points, and much of enzyme specificity, depend on the genetically determined primary structure." 34 CONCLUSION It seems that the above account of explanation-strategy in the area of temperature adaptation underscores many of the 31. Ibid., p. 353. 32. Ibid., pp. 369-370.

32

33. Ibid., pp. 357-358. 34. Ibid., p. 357.

Theories and Explanations in Biology points made earlier. First, it discloses the fruitful interaction of classical, evolutionary, and molecular approaches. Secondly, it indicates that biological characterizations are not rival accounts to chemical ones. Thirdly, it stresses the importance of the DNA sequence order in chemical explanations of biological organisms. One feature which this area does not seem to reveal, which genetics does, is the development of a biological (that is, nonchemical) theory which would explain temperature adaptation in nonchemical terms. What this shows is that one should not always expect systematization at the biological level, in the sense of being able to refer many phenomena to a small number of basic principles. The future development of laws and theories in biological terms in this area is, of course, not precluded by the fact that none have yet been discovered. The case does indicate that physical and chemical language and theories are extremely useful in biological investigations, and the physicochemical approach may well be the best approach in this area. In any event, ultimate reduction of the biological to the chemical seems to be a foregone non-controversial conclusion here as elsewhere.

33

The Basesof Conflictin BiologicalExplanation RICHARD C. LEWONTIN Department of Biology University of Chicago, Chicago, Ilinois

I believe that a large part of the problem of biological explanation is sociological rather than philosophical, for the renewed interest in the problem of explanation in biology is a result of the rise of a self-conscious and self-defined group of biologists who call themselves molecular biologists. This group feels, and explicitly states, that the reason biological explanation is in such a bad state is that biologists have by and large not asked the right kinds of questions and not used the right methods. There has been a setting up of a kind of dialectic of the molecular versus the evolutionary, the molecular versus the organismic, the reductionist as opposed to the synthetic theories of biology, a dialectic that is pretty much a false one. I would like to distinguish, from the standpoint of the biologists, wherein this conflict is a real one and wherein it is simply a sociological consequence of the excitement that some people have about what they do. There is first the false conflict, which is really a conflict of interest. The difficulty has arisen in the context of everyday language because the two kinds of biologists have asked what seemed to be the same questions in the same way, but have really been asking different questions. They have all used the word "how," but there is a confusion of final and efficient causation, a confusion that biologists must be permitted because most of them have never heard of Aristotle and those who have heard of him have never read him. The confusion between final and efficient causation among biologists is contained in the following exemplar: "Why is the sex ratio in Drosophila 1:1?" It is a question typical of the biologist, for it is really two different questions even though it is stated as one. "Why is the sex ratio 1: 1?" I can give a perfectly satisfactory answer to that question by discussing laws of genetics; by discussing chromosomal mechanics, which themselves are clearly reducible to physicochemical terms, although that has not yet been properly done; by reducing the explana-

35

RICHARD C. LEWONTIN

tion to a series of statements about the chemistry of cells and about the chemical nature of genes. That is, I can answer this question by talking about the formation of gametes, and how it is that sex is determined by chromosomes. For the molecular biologist, that is the answer of interest to the question: "Why is the sex ratio 1:1?" For the molecular biologist an answer which is not framed in those terms is not very interesting; it does not say what one ought to say about the world. There is a different answer to the question, "Why is the sex ratio 1 :1?", and that is to explain why it is 1:1 in Drosophila but 1: 2 in Habrobracon. We are not here interested in the question of the mechanism of the determination of the particular sex ratio in a particular organism, but rather in asking why there is a difference between two organisms in their sex ratio. We will not accept the answer that the mechanics of chromosomes are different in the two organisms; we will only accept an answer framed in terms of a functionalist or adaptive or evolutionary viewpoint. It will be an explanation that attempts to say that it is sensible that Habrobracon should have a 1:2 sex ratio while Drosophila should have a 1; 1 sex ratio. The various ways in which this answer is given will be discussed later in this paper. Here it should be emphasized that one of the problems in biological explanation arises from the fact that there are two groups of biologists asking a question like "why is the sex ratio 1:1?" and satisfied with two different kinds of answers because they are really interested in two different kinds of questions. Efraim Racker recently asked me what I was doing in my laboratory. I began to tell him, and as time went on it became clear to me that he really thought what I was doing was very dull. It was not that he thought it was not well done, but rather that I was not asking anything very interesting about nature. At the other extreme from the conflict of interest is what I will call the exclusive conflict between final and efficient cause, that is, the cases in which it is denied that an answer to the question "why?" in terms of final cause is even possible. Let me give an example of an evolutionary problem: R. A. Fisher asked, "Why is it that most mutations that occur in organisms are recessive?"-in other words, what is the origin of dominance of genes? Fisher gave an answer to this in terms of final causation. He talked about what happens in natural selection to new mutations as they arise and as they affect the fitness of organisms, and concluded by mathematical analysis that if mutations arose over and over again, then eventually

36

Bases of Conflict in Biological Explanation the genotype of organisms would evolve in such a way that any new mutation would be recessive in its phenotypic manifestations. What Fisher said was that dominance is a result of an evolutionary process in which the whole nature of the genome has been adjusted in order to hide the effects of new mutations. Sewall Wright opposed this notion, claiming that natural selection was too weak to accomplish the result predicted by Fisher, and that dominance is a physiological fact of life that does not need any explanation in terms of natural selection. Wright said that dominance arises because of the nature of the response-curve of enzymatic reactions. As is well known, when the concentration of enzyme is limiting, the amount of product is proportional to the amount of enzyme. As the amount of enzyme increases, the amount of product produced falls off and reaches an asymptote. Thus the phenomenon of dominance-because, for a doubling of the amount of enzyme, there is not a doubling of the amount of product. There will be complete dominance if in fact the amount of enzyme produced by one gene completely saturates the substrate. Notice that I am making a distinction between my first case, "why is the sex ratio 1: 1?", in which there exist alternative explanatory systems, each designed to answer essentially different questions framed in the same terms, and a second case in which the two forms of explanation commonly used in biology are at variance with each other. In the latter case, these two explanations are juxtaposed and one is rejected as being, in fact, incorrect. This conflict also occurs in molecular biology. As an example, let us ask why it is that the genetic code looks as it does. Why does a certain set of triplets of nucleotides and not another set of triplets specify phenylalanine? Is it just a result of chance? Would it make any difference in the world if instead of UUU specifying phenylalanine it were UGC? There are two schools of thought about this matter. One says that the code we have is an historical accident, that there could have been (and may have been at one time) other codes. One of the interesting discoveries of modern molecular genetics is that the code is universal. It appears that all living organisms at present share the same relationship between the set of triplets and the amino acid that is produced. On the theory that many codes are a priori equally likely, this discovery proves the monophyletic origin of life. However, it can be argued that the set of triplets which specify particular amino acids are not at random with respect to each other. If you really look at them hard, you notice that the redun-

37

RICHARD C. LEWONTIN

dancies are meaningful, that they are of such a form that the code is protected against any phenotypic effect of single-step mutations. It is adaptive to have UUG and UUU specify phenylalanine because a single step mutation in position three will not affect the amino acid specified. Among all the possible codes that could exist, evolution has, in fact, selected a code that has a set of redundancies with certain optimal properties. Competing with this, there is a third theory, which is that the code is not random but that it has not evolved either. Rather, it is postulated that there is a quantum-mechanical relationship between the code elements and the amino acids produced. On this view, even if we allow an evolutionary argument for redundancy, it is not chance that phenylalanine is specified by this set of triplets and not by an utterly different but equally redundant set of triplets, let's say AUG and AGG. There may be quantum-mechanical reasons that a certain triplet will most easily code for a given amino acid because of the way in which this information is encoded in the messenger RNA. Here we see three explanations. The first is that the actual realization out of all the possible ones is that which has arisen by a process of natural selection, and indeed that there were other groups of organisms with other kinds of coding but that these groups were not so successful because their coding systems were in some sense less perfect, less optimal, and so died out. The second is that there were other groups of organisms with other codings and these just happened to die out by chance, the one remaining being in no sense optimal. The third is that there has been one code because in the time of prebiotic evolution the particular coding arrangement that arose was that which had the highest probability under certain rules of quantum chemistry. A third conflict of explanation occurs in what may be called the problem of exclusive versus complete explanation. I am not going to say very much about this because it is fairly trivial and obvious to everyone, but it should at least be mentioned. Theoretical biology has as its goal the placing of bounds on the possible outcomes of given input conditions. That is to say, what theoretical biology is engaged in doing is taking a certain set of contingent statements about the world and constructing a machine, enabling one to say that if the inputs to this machine are of such and such a nature, the outputs must fall in a certain restricted subset. Theoretical biology is in fact nothing but a biology of exclusion. It says: if mutation rates are of the following nature, if the differences in organisms in rates of reproduction are such and such, then it

38

Bases of Conflict in Biological Explanation is not possible that outcome X will occur or the probability of outcome Y is less than 10-6. A good deal of theoretical biology is misunderstood by molecular biologists, who by and large call themselves nontheoretical biologists and think of themselves as the archetype of the experimental biologist because they are concerned not with exclusion but with complete specification. If we were to agree that because somebody has worked out the complete base sequence of one of the transfer RNA's, we will not allow anyone to have any more money to work out any of the others, because it is just more of the same, molecular biologists would revolt. They would say that we have destroyed one of the most important tasks of the biologist, which is to specify completely the set of biological mechanisms. The theoretical biologists will say, on the other hand: "We set up a theory which predicted certain characteristics; we have allowed you to go off and spend a lot of money making one molecule to prove that it really works. Now go away and do something else useful." A conflict arises, then, between those who say that we do not explain an observation until we have filled in the complete picture of all the mechanical levers and gears, and those who are satisfied with an explanation which only says that the levers and gears must look "something like the following," must fall within a certain set. This is a problem that arises not only in the conflict between theoretical biology and experimental molecular biology, but arises even within these disciplines. Related to it and of much more interest is what one might call the conflict between sufficient and exact explanation. Most of evolutionary theory has the following structure: there is some general framework of laws, and we know, from observations about nature, something about the values of the parameters that we put into the calculating engine. Thus we produce an allowable set of outcomes. We know something about mutation rates; we know about the differential rates of fertility of organisms. Given such bounds, the evolutionary biologist can then predict a possible set of outcomes. If he is really very clever, he can even set up a distribution function of the probabilities of the outcomes. The actual outcome is then observed, and we ask (a) whether it falls inside the set of allowable outcomes, and (b) if it does, what was the probability of that outcome. If the outcome is one of the highly probable ones, then we are happy, we have a sufficient explanation of the phenomenon and the evolutionary biologist goes no further. There is a very strong resemblance, then, between explanatory mechanisms in

39

RICHARD C. LEWONTIN

evolutionary biology and the standard theory of hypothesis testing in statistics. What one does is to make deductions about the set of probabilities of possible outcomes, to look at a given event and ask whether that event falls within a critical set (which is constructed from the a priori distribution of the possible outcomes); if it does, we are happy and claim a sufficient explanation. Most evolutionary biologists are satisfied with this kind of explanation. Let us take a classical case. There was a tremendous flowering of the architecture and ornamentation of the heads of the ceratopsian dinosaurs, a group of rhinoceros-like herbivorous dinosaurs. The question arises in evolutionary theory: why were there one-homed and three-homed ceratopsians; why is it that they all did not have the same ornamentation? There are two explanations for such a manifestation. One says that each one of the specific ornamentations was specially selected in an environment that was different from others and therefore was favored by natural selection. The assumption is that if a one-homed and a three-horned ceratopsian dinosaur were living in the same environment, three horns would be better than one and therefore three homs evolved. The alternative explanation is that these organisms evolved from somewhat different original pools of genes; that they evolved in isolation from each other, both temporally and spatially; that there was a tremendous time-scale involved, and that each of these outcomes represented a sufficient adaptive outcome of the evolutionary process. The question of comparing them or of demanding that there be specific environmental reasons for the different types is asking too much. In the multidimensional space of gene frequency, natural selection moved the population to one of a set of alternative singularities. Where it moved depended on its initial state and certain random elements. It is not even sensible to ask the question: is it better to be at one than at the other, or more adaptive? These are alternative outcomes of the evolutionary process, alternative probable realizations. It might be possible to compare these realizations if the organisms then become sympatric, but that is not at issue. The second explanation follows from the existence of alternative stable states of a given dynamical system. This is a very important point which is not always understood. It is not true that, in a deterministic dynamical system in which everything is fixed except the initial starting values, the system will go to one and only one point. There exist, in any dynamical system of sufficient complexity, many, many alternative

40

Bases of Conflict in Biological Explanation stable states to which the system may go, and the determination of which state the system evolves to is a function of variation in the starting point. For example, if you are a mountain climber with a compulsion to go upward, then in the Rocky Mountains there are over fifty peaks above 14,000 feet, any of which will make you happy. It only depends on where you start. It is not true, therefore, that all the results of evolution are different from each other because the forces of natural selection were different. Many are different from each other because the starting conditions were different. This second explanation is sufficient; it says that what happened could happen. A difficulty arises when an event occurs which ought to have happened with a very, very low probability. Even in such cases, of course, since the probability-density function of all outcomes of evolution is almost flat anyway, there is a serious problem of deciding whether a thing could have happened or not. This is a serious methodological problem for the evolutionists. The evolutionist who wants to decide among explanations by using the principle of maximum likelihood is in serious trouble because the likelihood surface is so flat. For that reason, evolutionary explanatory hypotheses generally are not very specific. Rather, a set whose total likelihood is high is accepted. I would like at this point to destroy what I think is an error in the writings about evolutionary theory, an error at least from our modern standpoint. That is the problem of whether or not the principle of natural selection, or survival of the fittest, is an explanation of evolution. I maintain that it is not an explanation of evolution and it should not be intended as such, whatever the intent of Charles Darwin and his disciples may have been. Evolution is the necessary consequence of three observations about the world, observations which are at this moment unchallengeable. They are: (1) There is phenotypic variation; the members of a species do not all look and act alike. (2) There is a correlation between parents and offspring. This statement says nothing about Mendelism, nothing about genetics. It just says that if the two parents are shorter than the average, their children will on the average be shorter. That is an observation which is true for every character to a different degree, but for every character that can evolve, it must be true. (3) Different phenotypes leave different numbers of offspring in remote generations. (If they have more grandchildren, that's good enough). If these three propositions are true, there will be an unavoidable evolutionary change in the population. Does that

41

RICHARD C. LEWONTIN

mean, then, that the principle of natural selection is an explanation of evolution? These are three contingent statements, all of which are true about at least some part of the biological world, and it therefore follows that evolution must occur. The reason I list them is because there is a lot of confusion about the so-called "tautological" nature of the explanation of fitness and natural selection. There is nothing tautological here. These are three true statements about the world, and it follows from them that changes must occur in the population and that these changes can be predicted. If you like, these three statements are the explanation of evolution. They do not explain the origin of any particular event in evolution; they explain why there is a change, in general, in the forms of organisms in time. Finally, I come to my main question: What do we mean by the historical elements in evolutionary explanation and in biological explanation? Let us consider first a generalized dynamical process in time, operating in some space. We can regard a system as being in some state E at time t and passing into some other state at E' at time t + 1, continuously or discontinuously. The system can be described as a point in a multidimensional state-space, and through time the system sweeps out a trajectory in the space. In this space there exist singularities. Think of the space as a vector field, a field showing the motion of the particle at every point. These singularities may be point singularities or they may be limit cycles, but more often in evolutionary context, they are not cycles but points. Arrows may be drawn showing the directions and the speeds with which the particle will move within the space at various points. Stable equilibria are points toward which all the vectors in a neighborhood point. If the system arrives at such a point, it will not come out again. It is characteristic of a dynamical system in time that if the system is at such a point or in such a closed circle, there is no information about how it got there. That is, at equilibrium, there is no historical information. Of course it is possible to subdivide the space into regions from which the system would have fallen into one or the other of alternative equilibria. So it has some information about the set of possible starting points, but only that very weak kind of information. So, any explanatory hypothesis in biology which has explicit in it that the system is in an equilibrium state is ahistorical because it says that the biological world is in a stationary state in the dynamical sense. If you deliberately choose to concen-

42

Bases of Conflict in Biological Explanation trate your attention on the equilibrium aspect of biological systems, you are making a deliberate choice to ignore their historical aspect. In choosing to study only the equilibrium you destroy any possibility of information about history. You can, on the other hand, choose to study the trajectories themselves. Since these trajectories can be described in general by some differential equations, if you have information about a number of points through which the trajectory passes, you can reconstruct the differential equations. That is to say, you then have historical information which allows many more deductions. For example, you can say where the system is going and where it has been. Thus we must distinguish in a dynamical system two sets of points that make up the whole space. One set of points is a small set in which all historical information has been destroyed, the set of all singularities. The rest of the points lie in a set in which one can reconstruct some historical information. The argument among biologists as to the relevance of history in explanatory hypothesis revolves around the division among biologists into that camp which says it wishes to study the equilibrium states in biology and the camp which wishes to study the dynamical states in time. This division exists among evolutionary biologists and especially among ecologists. There are now two schools of ecologists: first, those who say that what is interesting is only the ergodic properties of biological systems-the properties of the ensembles which are invariant in time and about which no historical statements can therefore be made; and second, those who want to study properties which are not invariant in time but which are in fact unique in time and space. I will give an example from modern ecology where these two alternative explanatory schemes are most in conflict. Let us ask the question: "WVhyis it that in a copse in England there are the following five species of birds-Species A, B, C, D, E-in the proportions PA' P B' PC> PD' and P., whereas in a copse in the United States which has life forms very similar to those in England, there are only four species of birds, F, C, H, and I, and in quite different proportions. This is a specific example of the whole problem of zoogeography and ecology; it becomes in fact the leading problem of ecological, evolutionary studies. There are two answers to this question. One is given by David Lack and represents the English school of explanatory hypothesis about zoogeography. The particular environmental situation is clearly different in England from that in America, and therefore there will be a different number of species and the identity of the

43

RICHARD C. LEWONTIN

birds will be different. That is, American birds occupy their copse and English birds theirs. They are different sets of birds because the copses are different. This is equivalent to the argument that one-horned dinosaurs and three-homed ones were one-horned and three-horned because the environment was different. Under this hypothesis, if we want to understand why a particular set of bird species occupies a particular place in the world, we must know almost everything there is to know about the environmental circumstances in that place and about all the interactions between the birds. Opposed to this view is the one typified by MacArthur and Wilson and the American school and given cogent form in their recent book, Equilibrium Theory of Insular Zoogeography. The authors suggest that we should forget about the names and identities of the species and ask why there are four or five species and not twenty-five species in this environment and whether we would predict more species in the English copse than in the American, on the theory that all the species are alike, or interchangeable. The equilibrium aspects of the occupation of woods by birds are such that, given a long time, such a copse will settle down to about four or five birds in frequencies predicted from a few variables. If all the birds of the species now occupying the copses became extinct, some other four or five species would get in and stabilize. In the case of islands, if one simply considers the size of the island, the distance from the nearest island, the distance from the mainland, and the number of species in the pool on the mainland, one can in fact give a fairly good description of how many species will be on each island. Moreover, one can predict the difference in the distribution of the number of species on islands that are close to the mainland and far out, and one can actually draw a graph of the relation between distance of the island to the mainland and the number of species. But in this equilibrium theory, the identity of the species is destroyed. Thus, two alternative explanatory devices exist in zoogeography. One claims that the particular five species in a copse are only a sample of all the propagules that arrived there, and the interesting question is really how many there are and not what their particular names are. The alternative says that what is important is not how many there are but which particular ones there are, and that no other combination of species would be stable. I presume the real answer is-if I take all samples of five bird species that could have been in a copse -that they will be graded according to the probability with

44

Bases of Conflict in Biological Explanation which any particular group of five would be successful, or the length of time they would be resident before being replaced by another group of five. I do not know what the distribution of this probability looks like; I rather suspect that it is strongly modal-that is, that there are a few combinations which have very high probabilities because of the nature of the interaction in the species, and lots of combinations which have very low probabilities. For the most part I have been concerned throughout this discussion with the sociology rather than with the philosophy of science. The two explanatory hypotheses, one of which includes an historical and specific element, and the other which includes a statistical and equilibrium element, have become for biologists mutually exclusive by the nature of the conflict between biologists as people. There is here a problem of historiography. Many of the very positive statements of reputable and intelligent scientists will be colored by the tendency of biologists, like everyone else, to polarize in some dialectical fashion. Historians have been taught from the cradle to be wary of what they read, not to confuse the record with what was. This caveat applies in science as in any other social endeavor. Each one of us has his preferred explanatory system and we tend to see the other explanatory systems as excluded by our own.

45

Explanationin Biology BENTLEY GLASS State University of New York at Stony Brook

It does not seem profitable to regard scientific description as constituting a form of explanation. By explanation we are to understand an interpretation of phenomena involving "cause" in a scientific sense. Yet cause is not to be conceived in a simplistic, unitary sense. As Bertrand Russell has pointed out, nothing less than the total prior state of the universe is causal for any particular event.' Nevertheless, in making scientific hypotheses upon which predictions may be based and subjected to test, certain recurrent, regularly associated relationships are usually sought out, on the basis of which one may predict from the occurrence of the earlier one that the other will follow with a certain definite probability. This relationship, to be sure, does not distinguish between "cause" and "correlated event"-between the phenomena in direct sequential relationship and those on different paths branching from the same initial event. I have elsewhere discussed this ambiguity and criticized the identification of "necessary connexion" (that is, cause-according to Hume, Mach, Pearson, Braithwaite, and others) with "constant conjunction." 2 Unless no interval of time at all intervenes between A (cause) and B (effect), an assumption which Russell points out is untenable, and unless therefore no intermediate state could possibly exist, correlated events may exhibit perfect conjunction and temporal sequence, yet belong to separate branches of the nexus of events. In the simplistic view assumed by those who have followed Hume in adopting constant connection as the criterion of cause, only the simple chain reaction, with one event 1. Bertrand Russell, "On the Notion of Cause, with Applications to the Free-Will Problem," in Readings in the Philosophy of Science, ed. H. Feigl and May Brodbeck (New York: Appleton-Century-Crofts, 1953), p. 392. 2. Bentley Glass, "The Relation of the Physical Sciences to BiologyIndeterminacy and Causality," in Philosophy of Science. The Delaware Seminar, vol. I, 1961-1962, ed. B. Baunmrin (New York and London: John Wiley and Sons, 1963), pp. 223-257; esp. pp. 233-239.

47

BENTLEY

GLASS

at each interval of time, is allowed. This view may appeal to the physical scientist, but the biologist and biochemist are too familiar with branching step reactions to accept such a view. Verbal explanations have pervaded biology in the past, to its real detriment. Thus, the vague and untestable concept of protoplasm as the material substance of life long prevented a critical analysis of living systems. The theory of the gene, so fruitful in its early history, became a deadening influence because of general lack of sufficient appreciation of the distinction between the operational modes (mutation, recombination, function) whereby a given gene becomes known. The conclusion by Louis J. Stadler in 1954 that geneticists must clarify their concepts through sharper use of operationally defined concepts,3 and the somewhat later brilliant work of Seymour Benzer4 carrying out that injunction, did much to pave the way for the great revolution in genetic thinking that took place in the 1950's and marked the advent of the era of the genetic code. Biology is the science of life, but what "life" itself is we cannot define in precise operational terms. As William S. Beck has stated, "Life and nonlife . . . are ends of a spectrum whose graded quantity is complexity: life is on the complex end, nonlife on the simple. Between the two is a middle ground which is neither one nor the other. It is, one might say, what it is." 6 That view is further emphasized in the title of an essay written by N. W. Pirie, "The Meaninglessness of the Terms 'Life' and 'Living'."6 Granting the foregoing, explanation in biology must clearly grade into the modes of scientific explanation used in the physical sciences. The question remains: are there other modes of explanation, either current or outmoded, which are peculiar to biology, either because of its history as a science or because of the nature of the phenomena and interrelationships at the complex end of the spectrum of nature? I believe that there are. We may well start with Aristotle's doctrine of the Four Causes, since Aristotle was the greatest biologist of the ancient world and rather surely formed his 3. L. 3. Stadler, "The Gene," Science, 120 (1954), 811-819. 4. Seymour Benzer, "The Elementary Units of Heredity," in A Symposium on the Chemical Basis of Heredity, ed. W. D. McElroy and B. Glass (Baltimore: The Johns Hopkins Press, 1957), pp. 70-93. 5. W. S. Beck, Modern Science and the Nature of Life (New York: Harcourt, Brace, 1957), p. 186. 6. Published in Perspectives in Biochemistry, ed. J. Needham and D. E. Green (Cambridge [Eng.] University Press, 1939), pp. 11-22.

48

Explanation in Biology epistemology on the basis of his studies of living creatures and their development. Aristotle distinguished four categories of cause: the Material Cause, or matter, out of which the thing is made; the Motive, or Efficient, Cause, which sets the process going and is that by which the thing is made; the Formal Cause, or Form, which is responsible for the character of the course which the process follows, and is the special character of the development that the process pursues; and the Final Cause, the End or Object toward which a formative process advances, its logos, its rational purpose, the complete and mature final stage. These definitions are taken from the excellent discussion by A. L. Peck in the Introduction to his translation of Aristotle's Generation of Animals.7 Note further that, in terms of the development of a chicken or a dog, the Material Cause was conceived by Aristotle to be supplied by the female parent, the Motive or Efficient Cause by the male parent. The Formal Cause consisted of the sequence of stages through which the embryo and young chick or puppy pass during the course of development, and the Final Cause is represented by the mature bird or dog, the goal of the developmental process. The modern biologist has translated these varieties of cause into his explanations of biological phenomena. The Material Cause has become the molecular explanation of the biochemist and geneticist. In terms of metabolic maps, each substrate of a step in a chain of reactions is the "material cause" of the product or products. Thus, in the step reaction A->B-+C, A is the material cause of B; B the material cause of C. When, more often, two or more substrates react and enter into the formation of one or more products, each one must be regarded as a "material cause." Not only is the immediate substrate in such relation to the product, but any material substance at an earlier step which becomes transformed into the product is so to be regarded. Thus, in the well-known steps of glycolysis, glucose, glucose-6-phosphate, fructose-6-phosphate, fructose-1, 6-diphosphate, glyceraldehyde-3-phosphate, 1,3-diphosphoglyceric acid, 3-phosphoglyceric acid, 2-phosphoglyceric acid, and phosphoenolpyruvic acid are all precursors and principal substrates for the formation of pyruvic acid, in the order 7. Aristotle, Generation of Animals, trans. A. L. Peck (Cambridge, Mass.: Loeb Classical Library, Harvard University Press, 1943), pp. xxxviii-lxiv. At the Asilomar Conference my attention was directed to a similar identification of Aristotle's Four Causes with modern biological conceptions of cause. Unknown to me, the same idea had previously been set forth by John Herman Randall, Jr., in Aristotle (New York: Columbia University Press, 1960).

49

BENTLEY

GLASS

named. Some of the steps simply involve rearrangements of the substrate molecule; others require the addition or removal of phosphate groups, water or hydrogen atoms. The small chemical groups (such as the phosphate group or water) may be regarded as material causes no less than the great ones. Many biological explanations stop at this molecular level. It is purely descriptive except for the introduction of the dimension of time. The Efficient Cause, in modern terms, relates to the energetics of the biological process: thermodynamics, the kinetics of reactions, and the effects of catalytic agents, especially enzymes, upon the required activation energies and upon the steric placement of substrate molecules that so greatly enhances reactivity. Since most energy is supplied in biological systems in the form of chemical energy, the boundary between this category of causes and the first is none too sharp. Nevertheless, I would place here those biological explanations that relate to the genetic control of protein synthesis, the enzymatic control of metabolic reactions, the capture of solar energy through photosynthesis, chemosynthetic mechanisms, and all positive and negative feedback regulatory mechanisms that promote homeostasis. Life has been characterized as an eddy in the downward flow of energy to the sea of entropy, a good analogy calling attention to the evanescent storage of energy and the reversal of the Second Law of Thermodynamics that take place in a living organism.8 An explanation of a sort, perhapsl Certainly many biologists are satisfied to leave their explanations of the phenomena they study at this level. Trace the energy to ATP, the reaction to the presence or absence of an enzyme, the developing characteristic to a gene-and they feel the job is sufficiently done. The Formal Cause in modern biology takes the shape of explanations that relate form to function. The complementarity of form and function is a basic biological conception with little counterpart in the physical sciences, where the idea of function-the performance of action directed toward an appropriate adjustment to the environment-is essentially lacking. The idea of function is, in fact, a teleological notion unless construed in the light of evolutionary adaptation as a 8. The idea of life as a whirlpool ("tourbillon continuel") of matter entering and leaving the organism was a favorite of Cuvier, cited and quoted by Lawrence K. Henderson in The Fitness of the Environment (New York. Macmillan, 1927, pp. 23-34). The subsequent relation to the Second Law of Thermodynamics was pointed out by Henderson himself, as well as by many others; see in particular Homer W. Smith, Kamongo (New York: Viking Press, 1932, pp. 147-154).

50

Explanation in Biology product of natural selection. The "Wisdom of Divine Providence," seen in the eighteenth and nineteenth centuries to be manifestly revealed in the relation of structure to function in the human hand or in the mutual suitability of bees and flowers for pollination, is witness to the essentially teleological bias relating the Formal Cause to the Final Cause. Darwin freed us from this constraint by showing clearly just how natural processes can produce "appropriate" structure and function without divine intervention. The same arguments apply, furthermore, to the appropriateness of successive stages in the development of an organism. Natural selection has worked on all stages of the life cycle, not on the adult stages alone. The apparent foresight with which each stage of development prepares for the advent of the next one and looks forward to the goal of a mature, reproducing adult is actually a relation of form to function that looks back to the programmed instructions for the developmental process, instructions located in the genotype of the individual. The genotype is well-adapted to the usual environment in which development takes place because of the natural selection of genotypes that occurs in every generation-that perpetuation of the better-adapted and elimination of the less fit in a world in which there is differential mortality and fertility of genotypes. The Final Cause, the teleological aspect of Aristotle's view of nature, is replaced in modern biology by explanations of a teleonomic sort, to use the term usefully introduced by C. S. Pittendrigh.9 The teleonomic explanation explains adaptations of form and function, as well as of genotype and final adult characteristics, on the basis of natural selection. On this view it is the environment that serves as the screening element in the perpetuation of whatever mutant forms of genes compete in any given population. The environment in this sense is to be understood in the widest possible sense. It operates at different levels of the living world. At the molecular level the enzyme or the messenger RNA may be regarded as the environmental pattern that affects the reactivity of substrates or the synthesis of proteins. The special structures such as mitochondria in which enzymes of the respiratory chain are arrayed in assembly lines provide a further example of the ordered molecular environment whereby reactions can be carried out locally that would never be perceptible in a randomized system composed of the same constituents. The environment of the 9. C. S. Pittendrigh, "Adaptation, Natural Selection, and Behavior," in Behavior and Evolution, ed. A. Roe and G. G. Simpson (New Haven: Yale University Press, 1958), p. 394.

51

BENTLEY

GLASS

individual living cell in a multicellular organism is comprised of the other cells of the tissue and of the intercellular fluids. Natural selection must have served also to evolve the mutual fitness of cells, both those of the same sort in a tissue and those of different tissues making up a single organ and performing a joint function. To natural selection is also due the remarkable fitness and constancy of the internal milieu at which physiologists since Claude Bernard have consistently marveled, and upon which the homeostasis of the organism and its regulation by neural and hormonal messages largely depend. The individual is of course molded both temporarily and genetically by the pressures of the external environment, including especially the influences of other beings of the same population and species, as well as members of other living species in the community of which the individual is a part. At all these levels of interplay between living system and environment the teleonomic, evolutionary explanation is central. Life on this earth has had a history, to be understood only in terms of the evolutionary process. The fitness of living attribute to the forms-all their remarkable adaptations-we repeated action of natural selection through times past on individuals, their characteristics, their genotypes, and ultimately their individual genes. Many times in this history the thing that actually did happen was but one of many possibilities. The entire history thus adds up to a generation of highly improbable creatures in the multitude of possible types never realized, at least on this planet. There is thus a uniqueness about the evolutionary history of life that escapes prediction although it is fully explicable on the basis of the current theory of mutation, natural selection, and isolating barriers as the essential factors in the origin of species. The types of biological explanation chiefly employed at the present stage of the science may be summarized in the form of a series of questions that relate one aspect of life to another. Another biologist might suggest others. The following list is my own: 1. How is a specific character related to genotype? a. How is a specific protein determined by the genetic code? b. How is a specific character related to the function of a particular protein? 2. How is one stage of development and life cycle related to those that precede and those that succeed it? a. How is form generated? 52

Explanation in Biology

3.

4. 5. 6. 7. 8.

b. How are the definitive features of a species replicated in successive generations? How is one level of organization related to those below it and above it? a. What special laws obtain at particular levels? Are they reducible? How is the living system related to its environment, at each level of organization? How is form related to function, in each specific case? How are regulation and homeostasis achieved, at each level from molecular to biome? How is specific response related to specific stimulus, over short and long intervals of time? How do living organisms acquire adaptations through the evolutionary process?

Each of these questions might be explicated on a simple descriptive basis; each one can also be pursued through all the levels of biological organization. The molecular biologist, the ecologist, the developmental biologist, and the evolutionist will give different, but complementary, answers to the same question. Only the total matrix of biological explanation can fully suffice.

53

Hugode Vriesand the Receptionof the "MutationTheory"* GARLAND E. ALLEN Department of Biology, Washington University, St. Louis, Missouri

INTRODUCTION In 1905 the American Society of Naturalists devoted their Philadelphia meeting to a series of papers on the mutation theory by the Dutch plant physiologist, Hugo de Vries. Originally published in two massive volumes between 1901 and 1903, this theory was an attempt to explain the origin and inheritance of species differences in terms of large-scale changes between one generation and the next.' From the papers presented it is clear that workers from all the major fields of biology-cytology, embryology, plant and animal breeding, and even ethology (the study of animal behavior)-felt the relevance and importance of de Vries' work. Essentially a concept of heredity and evolution, the mutation theory gained widespread popularity at a time when both the Darwinian and Mendelian theories were before the biological community. It is true that in 1903 Mendel's laws had not been given a general enough applicability to win a wide following. But the Darwinian theory had commanded considerable attention for almost two generations. The willingness of many biologists to accept de Vries' mutation theory as an alternative to Darwinian natural selection raises some interesting questions. Why was de Vries' theory so attractive in the decade between 1903 and 1913? What inherent problems with Darwinian explanations led many biologists into the de Vriesian camp? Was the difference which many biologists saw between de Vries and Darwin substantive, methodological, or both? Why did so many workers accept de Vries' theory and reject Dar* Preparation of this paper has been aided by a grant from National Science Foundation (GS 1832). 1. Hugo de Vries, Die Mutationstheorie (Leipzig; Von Veit and Co., 1901-1903), 2 vols.

55

GARLAND E. ALLEN

win's? The answer to this last question provides an insight into the then current criteria for acceptable biological explanations, with particular reference to evolutionary theory. REACTIONS TO DARWINIAN THEORY AT THE TURN OF THE CENTURY In 1885, on occasion of the dedication of the Darwin memorial to the British Museum, Darwin's long time friend and defender, T. H. Huxley, spoke optimistically of the triumph which the theory of natural selection had achieved over its opposition: It is curious now to remember how largely, at first, the objectors predominated; but considering the usual fate of new views, it is still more curious to consider for how short a time the phase of vehement opposition lasted. Before twenty years had passed, not only had the importance of Mr. Darwin's work been fully recognized, but the world had discerned the simple, earnest, generous character of the man, that shone through every page of his writings.2 Yet Darwin's opposition had never been completely routed, and by the turn of the century had reappeared, like the indestructible Hydra, in a variety of new forms. Orthogenesis, neo-Lamarckism, metakinesis, heterogenesis, convergence, emergence, and a host of other processes or factors were elaborated as theories and put forward as alternatives or substitutes for Darwinian natural selection. So widespread was the disaffection with the Darwinian theory that in 1903 the German botanist, Eberhart Dennert, could write: "We are now standing at the death bed of Darwinism, and making ready to send the friends of the patient a little money to insure a decent burial of the remains." 3 Although polemical and inaccurate, Dennert's book represented one extreme of a continuous spectrum of biological treatises which in varying degree opposed Darwin around the turn of the century.4 The situation was summarized by the Stanford entomologist, V. L. Kellogg, in 1907 when he said: Although it may be stated with full regard to the facts that a major part of the current published output of general 2. T. H. Huxley, "The Darwin Memorial," reprinted in T. H. Huxley, Darwiniana: Essays (New York: Appleton, 1915), pp. 249-250. 3. Eberhart Dennert, At the Deathbed of Darwinism, trans. E. B. O'Hara and John H. Peschges (Burlington, Iowa: German Literary Board, 1904), p. 4. 4. V. L. Kellogg, Darwinism Today (New York: Henry Holt, 1907), p. 7n.

56

Hugo de Vries and the "Mutation Theory" biological discussions, theoretical treatises, addresses, and brochures dealing with the great evolutionary problems, is distinctly anti-Darwinian in character . . . the fair truth is that the Darwinian selection theory, considered with regard to its claimed capacity to be an independently sufficient mechanical explanation of descent, stands today seriously discredited in the biological world.5 This evaluation has been confirmed more recently by Sewell Wright and Loren Eisley.6 The opposition to Darwinian theory was not limited either to specific nationalities or to workers in specific fields of biology. Aside from Germany, other countries represented were Holland, Russia, Italy, and France.7 Americans, too, got into the act. Especially among such paleontologists as Alphaeus Hyatt, E. D. Cope, A. S. Packard, W. H. Dall, and H. F. Osborne, a dissatisfaction with Darwinian theory was evidenced by a strong advocacy of neo-Lamarckism.8 Within biology itself, embryologists such as Hans Driesch, Hans Spemann, and T. H. Morgan, anatomists such as Thomas Dwight, animal psychologists such as Jacques Loeb, geneticists such as Wilhelm Johannsen, cytologists such as E. G. Conklin and E. B. Wilson, and even long-standing morphologists such as W. K. Brooks, all found the Darwinian theory lacking in one or more crucial points. However, the climate of opinion was by no means wholly anti-Darwinian in character. Among the strongest supporters of natural selection were three of the most influential names in biology at the turn of the century: A. R. Wallace, John George Romanes, and August Weismann. In addition, Darwin found support from a large number of naturalists, among them, David Starr Jordan, J. T. Gulick, 0. C. Marsh, V. L. Kellogg and many others.9 While these men loyally and per5. Ibid., pp. 4-5. 6. Sewall Wright, "Genetics and the Twentieth-Century Darwinism," American Journal of Human Genetics, 12 (1960), p. 367; L. C. Eiseley, Darwin's Century (New York: Anchor Books, 1961), p. 229. 7. Among the German anti-Darwinians were Wiegand, Haacke, von Sachs, Goette, Korschinsky, Haberlandt, Steinmann, Eimer, von Kolliker, Nageli, Kerner, Fleischmann, and 0. Hertwig. See Kellogg, Darwinism, p. 5.

8. For a discussion of the developments of neo-Lamarckism in America see Edward J. Pfeifer, "The Genesis of American neo-Lamarckism," Isis, 56 (1965), 156-167. 9. See D. S. Jorgan, Standeth God in the Shadows (New York: Thos. Crowell, 1901); J. T. Gulick, Evolution, Racial & Habitual (Carnegie Institute of Washington, 1905); Karl Jordan, "Reproductive Divergence: A Factor in Evolution," Nat. Sci., 11 (1897), 317-320; on Marsh see

57

GARLAND E. ALLEN

sistently defended Darwinian theory, the fact remains that many biologists of note argued just as strongly against it for reasons we will explain in the following pages. In this paper the terms "Darwinism" and "neo-Darwinism" will be used synonymously. It must be recognized, however, that the particular brand of neo-Darwinism current around the turn of the century was not the unadulterated thought of Darwin himself. Rather, it emphasized several points in particular: the all-importance of natural selection as the agent producing evolution, and the action of selection on minute, individual variations rather than on large-scale monstrosities, or sports.10 Neo-Darwinism also deemphasized the inheritance of acquired characters and the effects of use and disuse of environment on the origin of variations. The opponents of Darwin were fighting principally the neo-Darwinians, not Darwin, and they recognized this by paying special tribute, time and again, to the pioneering efforts and laudable objectivity of the English naturalist. THE MUTATION THEORY Although he began his career as a plant physiologist, Hugo de Vries (1848-1935) became best known for his work in heredity and plant breeding. In the 1890's, during a field excursion near the town of Hilversum, just outside Amsterdam, de Vries noted what appeared to be several species of the evening primrose, Oenothera, growing side by side. One seemed to be a parental strain which had given rise to two offspring types differing in enough characters to be regarded as separate species. De Vries transplanted all three types of plants back to his laboratory garden and grew them to maturity. He recognized in these plants the opportunity for laboratory and experimental study of evolution in a way that was impossible in the field. First published in two volumes in 1901 and 1903, de Vries' mutation theory is essentially "the proposition that attributes of organisms consist of distinct, separate and independent units." 11 It is a particulate concept of heredity, developing from de Vries' own theory of "intra-cellular pangenesis," origC. Schuchert and E. M. Levene, 0. C. Marsh, Pioneer in Paleontology (New Haven, Yale University Press, 1940); V. L. Kellogg, Darwinism Today. 10. V. L. Kellogg, Darwinism Today, pp. 13, 15. 11. The Mutation Theory, trans. J. C. Farmer and A. D. Darbishire (Chicago: Open Court, 1910), p. 3.

58

Hugo de Vries and the "Mutation Theory" inally published in a volume of that title in 1899.12 Although de Vries' hereditary particles bore the name of "pangenes," he had something more in mind than Darwin's units of the same name. To de Vries, pangenes were "in many respects analogous to the molecules of the chemist," though having a much more complicated structure and arising in a historical way. However, de Vries recognized that hereditary units, unlike the chemist's molecules, could not be isolated and subjected to experimental analysis. The evolutionist had to take the less direct route of controlled breeding experiments, from which could be deduced the exact hereditary condition of the parents.13 De Vries' commitment to a theory of pangenesis prepared him to interpret his Oenothera discoveries as examples of the obvious discontinuity which he felt must exist in nature. This meant that related species, or even varieties in nature, existed as discrete, definable types between which no intergradation could be found. Such discontinuity in the field implied a discontinuity in the origin of the types themselves. Thus, according to de Vries, new species arose all at once, with no preparation and with no obvious transitional forms between one and the other. De Vries' theory was in opposition to that of Darwin, which emphasized the origin of species by the action of selection on minute, individual, or, as they were called, "fluctuating" variations. De Vries recognized his break from Darwin and pointed it out explicitly in the preface to The Mutation Theory:14 The origin of species has so far been the object of comparative study only. It is generally believed that this highly important phenomenon does not lend itself to direct observation, and, much less, to experimental investigation. This belief has its root in the prevalent form of the conception of species and in the opinion that the species of animals and plants have originated by imperceptible gradations. These changes are indeed believed to be so slow that the life of a man is not long enough to enable him to witness the origin of a new form. The object of the present book is to show that species 12. Intracellulare Pangenesis (Jena: Gustav Fischer, 1899). English translation by C. S. Jaeger as Intracellular Pangenesis (Chicago: Open Court, 1910). 13. Mutation Theory, II, 567-568, 570. 14. I, viii. Throughout the remainder of this paper references to this work will be taken from the English edition cited here, rather than the original German version of 1901-1903.

59

GARLAND E. ALLEN

arise by saltation and that the individual saltations are occurrences which can be observed like any other physiological process. Forms which arise by a single saltation are distinguishable from one another as sharply and in as many ways as most of the so-called small species and as many of the closely related species of the best histemotists, including Linnaeus himself. In this way we may hope to realize the possibility of elucidating, by experiment, the laws to which the origin of new species conform. The results of these studies can then be compared with those which have been obtained with systematic, biological and particularly with paleontological data. A most remarkable agreement will be found to exist between these and my new results. De Vries made a point of distinguishing mutations from individual, fluctuating variations. The latter were subject to statistical study and could be regarded merely as temporary adaptations to the environment, having nothing to do with the origin of species. Selection could not make permanent any changes which depended upon this type of variation.15 Mutations, on the other hand, occurred less frequently than individual variations and could not be regarded merely as extremes of some spread around a mean established for the species as a whole. Mutant forms had a newly established mean of their own, which represented a distinct break from the mean of the parent species. A consequence of this large-scale difference of the mutant forms was that the offspring were infertile with their parents. To account for the range of variation observed within the natural world, de Vries distinguished between three types of mutations: progressive, retrogressive, and degressive. Progressive mutations were the sudden changes which produced a new character. The new character appeared to arrive completely de novo, for de Vries stated that in two races, of which one has arisen from the other by the formation of a new character, that new race has one more hereditary factor than the other.'6 A progressive mutation need not be visible as soon as it occurs in the organism. According to de Vries, it could remain latent for a period of time though it eventually would express itself as a visible mutant character. Retrogressive mutations involve the disappearance of a character which previously existed: appearance of a colorless variety of flower 15. Mutation Theory, I, 5-6. 16. Ibid., II, 573.

60

Hugo de Vries and the "Mutation Theory" in one generation, the loss of variegation in the leaf, and so on. Such mutations were, however, less important than progressive mutations in the over-all evolutionary changes giving rise to new species. Finally, degressive mutations involved the activation of a latent character which previously existed in the history of the species but had been inactive for many generations: that is, true atavism. De Vries cited the occasional appearance of crumpled leaves in Oenothera laevifolia as examples of degressive mutations in that species."7 To de Vries there was a qualitative difference between progressive mutations on the one hand and retrogressive and degressive mutations on the other. Only progressive mutations really involved formation of new characters. The other two involved the loss or gain, respectively, of a character whose potentiality was already present in the germ cells of the organism. Furthermore, progressive mutations did not follow Mendelian laws whereas retrogressive and degressive mutations did. As one of the rediscoverers of Mendel's paper of 1900, de Vries was explicit in showing how his mutations related to the laws of segregation and random assortment. "As far as our present experience goes, the retrogressive changes follow the Mendelian laws, but the progressive do not . . . The laws which govern the splittings (of hybrids) have been laid down by Mendel, but the behavior of the progressive differences is still awaiting extended investigation." 18 Mendel's laws applied to a specific set of problems, those involved in hybridization. But they had no relation to the larger question of the origin of new characters or species. For de Vries, who was imbued with his discontinuous view of heredity, small Mendelian variations could not have any significant role in evolution. His distinction between retrogressive and progressive changes may have been one of the factors responsible for the delay in applying Mendelian principles to an understanding of evolution. Although he made a clear distinction between progressive, retrogressive and degressive mutations, de Vries maintained that it was actually possible for new species to arise in any one of the three ways."' According to de Vries, mutations arose one at a time, but he was never clear whether only one of many individual mutations were necessary to produce a form representing a different species from its ancestors. It is ap17. Ibid., II, 70-72; I, 311. 18. "Luther Burbank's Ideas on Scientific Horticulture," Century Magazine, 73 (1907), 674-6S1. 19. Mutation Theory, II, 71-72.

61

GARLAND E. ALLEN

parent that one progressive mutation, involving a single character, could produce a new species. A single retrogressive or degressive mutation, however, was not enough to produce a completely new species. It only added variety within the main lines of development established by progressive mutations. As he wrote, "Progress on the main line of descent results from the production of new characters; but the extraordinary variety of forms results from the occasional disappearance of characters already existing, or from the activation of latent ones (retrogression, degression)." 20 That new species can arise from several degressive or retrogressive mutations de Vries makes clear in his discussion of hybridization. By recombination of enough such characters a form could result which would be infertile with its parents that is, a wholly distinct species. The distinction between the origin of species by hybridization (of degressive and retrogressive mutations), and the origin of specific characters (by progressive mutations) allowed de Vries to see evolution as resulting from two fundamentally different processes: the appearance of totally new characters and the reshuffling of already existing characters or potentialities through crossbreeding.21 In de Vries' system both were means, though by different mechanisms, for producing new species in sudden jumps. The pattern in which various mutations occurred was of some concern to de Vries. First, retrogressive and degressive mutations were predictable. The essential conditions for the production of these mutations were always present-they were responses on the part of the organism to stimuli from the environment. A latent character could be called forth, or a demonstrable one made latent by specffic external factors. Thus, by knowing the conditions of the environment and the nature of the plant, one could know what kinds of mutations to expect. Second, retrogressive and degressive mutations could occur in only one direction: the character could either appear or disappear, but it could not take off in a wholly new direction. Progressive mutations, on the other hand, were predictable neither in the frequency nor in the direction in which they occurred. They were produced by some internal stimulus, 20. Ibid., II, 75. 21. Ibid., II, 72. As de Vries wrote: "Species can arise from hybrids but specific character cannot arise by means of hybridization." (p. 72). He went on to say that the number of species originating through hybridization was much smaller than that originating through progressive mutations. See also, de Vries, "Luther Burbank's Ideas of Scientific Horticulture," p. 677.

62

Hugo de Vries and the "Mutation Theory" had no relation to environmental stimuli, and could occur in any number of directions. Thus, the nature of a progressive mutation could never be predicted before it occurred. Furthermore, progressive mutations tended to occur, according to de Vries, during "mutation periods" rather than purely at random.22 During mutating periods, the rate of mutation in a population of organisms was considerably higher than in normal periods. This meant that many different progressive mutations were occurring, but it also meant that the same mutations might occur in more than one individual at a time. Thus de Vries could explain how a new mutant form, unable to breed with its parents, could still be crossbred with another individual of similar hereditary makeup. De Vries defined a species as "a form which owes its origin to the production of a new internal factor." A variety, on the other hand, was "a form which owes its peculiarity merely to the change in condition of a factor already present." 23 The term species itself, de Vries felt, was often used in two different senses: one was the systematists' species, the other the "real"species." 24 To de Vries, systematists' species were arbitrary units created to bring some order into our view of the plant and animal world. The distinctions which systematists made between species was based, according to de Vries, for the most part on trivial or non-adaptive characters. Thus, species boundaries defined taxonomically had no reality in nature. In contrast, "real" species did exist, and were distinguished from one another on the basis of marked and macroscopic character differences. To these groups de Vries gave the name "elementary species." It was the elementary species that formed the smallest real, distinct group which underwent evolution. But how was this concept of mutation related to the actual creation of species in nature? In Volume II of The Mutation Theory, de Vries tried to get at this point. Progress in organic nature, he maintained, consisted essentially of an increase in differentiation. The specific characters which make up the species become more numerous and, consequently, every more highly organized form has a greater number of characters than its ancestors had. Thus, the number of elementary characters, that is, those produced by progressive mutations, must increase with increasing differentia22. Mutation Theory, II, 74. 23. Ibid., II, 579. 24. H. de Vries, Species and Varieties

(Chicago:

Open Court, 1905),

p. 10.

63

GARLAND E. ALLEN

tion. If it were possible to count the characters which a species possessed, biologists would have a measure of the degree of differentiation or total differences between two groups. What, then, about the origin of one species from another by the addition of a new character (progressive mutation)? De Vries wrote: Obviously the individual steps are only small ones, at the present time at least; any single one of them can hardly effect a noticeable increase in differentiation. At any rate we have at present no means of so exactly measuring the degree of differentiation, since we cannot estimate the possible influence of a single unit on a complex build-up of thousands of them. Only groups of units produce clear and obvious differences in the degree of organization; but within the limits of a small genus or of a multiform collective species the several types seem to us to be almost always equivalent.25 The above statement implies that no two species can be differentiated from one another on the basis of a single mutational difference. However, in the "Introduction" to Volume I of the same work, de Vries implied something quite different when he defined mutation: "Each new unit, forming a fresh step in this process, sharply and completely separates the new form as an independent species from that from which it sprang. The new species appears all at once; it originates from the parent species without any visible preparation, and without any obvious series of transitional form." 26 That de Vries was unable to make a clear and unequivocal statement about the actual differences between species illustrates the highly theoretical nature of the relationship he saw between mutations and the origin of species. De Vries compounded the confusion when he attempted to distinguish between species and varieties. For the most part, he made variety synonymous with the systematists' species.27 He recognized, however, that the term variety was often loosely used; quite frequently what was called a variety was indeed a good elementary species. True varieties were, he maintained, derived from and subordinate to the higher category of species. De Vries was no systematist, nor was he particularly well trained in field botany. As an experimentalist having little familiarity with natural populations, he could not fully appreciate or understand the enormous complexities involved 25. Mutation Theory, II, 570. 26. Ibid., I, 3. 27. Ibid., II, 58.

64

Hugo de Vries and the "Mutation Theory" in arriving at a broadly applicable species concept. The definitions which he did develop unfortunately had little relevance to species in nature or to the problem of the origin of variations. In hitting upon mutations, de Vries thought he had given an answer to a crucial problem of the time: a process of heredity which could also explain evolution.28 That he had not been able to show mutating forms in many genera other than Oenothera was no serious problem to him. He was confident that such forms could be found if plants were grown under proper experimental conditions. He was equally confident that mutations could be demonstrated in animals, a task to which several workers turned their attention between 1901 and 1910. THE RECEPTION OF THE MUTATION THEORY De Vries wrote in an age dominated by the idea-at least in name-of particulate inheritance or its logical counterpart, discontinuous variation. His own earlier theory of intracellular pangenesis talked of hereditary units or particles which operated within the physiology of the cell. His later theory of mutation was couched in the analogous vein of discontinuous inheritance. Discontinuity in the inheritance of individual characters was for de Vries the logical counterpart of a particulate theory of inheritance, since hereditary particles were themselves discontinuous entities. The idea of discontinuity had gained great support from a voluminous study by the British embryologist and naturalist, William Bateson in 1894.29 De Vries referred to Bateson's Materials for the Study of Varieties as an important stimulus to his own thinking. Bateson surveyed many natural populations and concluded that since species in nature were discontinuous, the variations giving rise to them were also discontinuous. In addition, theories of discontinuous inheritance had been advanced before this time by von Kolliker (1864), W. H. Dali (1877), and H. Korschinsky (1899).30 But of all these it 28. See L. C. Dunn, A Short History of Genetics (New York; McGrawHill, 1965), p. 56. 29. William Bateson, Materials for the Study of Variation: Treated with Especial Regard to Discontinuity in the Origin of Species (London: Macmillan, 1894). I am grateful to Dr. William Coleman for providing me with sales records of Bateson's Materials. Of the original 1500 copies printed, 306 were sold in the first year (1894), 79 the next year, 23 in 1896, and 14 in 1897. A cut in the prices increased the sales to 378 in 1898. While Bateson's influence was probably not extensive it was keenly felt by certain workers concerned with heredity and evolution, such as de Vries. A number of workers referred to Bateson, though the extent of their debt is difficult to ascertain. 30. For example, see V. L. Kellogg, Darwinism Today, pp. 327-333.

65

GARLAND E. ALLEN

was certainly de Vries' work of 1901 which aroused the most attention. The response of the biological community to the mutation theory was on the whole highly favorable. A large number of workers from many fields supported the theory with varying degrees of enthusiasm.8' Wherever anti-Darwinism was particularly strong, the mutation theory enjoyed its greatest popularity. It was obvious from the beginning that de Vries' work threw a new light on the problem which Darwinian selection had left unclear in many minds: the origin of hereditary variations. As Kellogg indicated, "On the whole the theory has been warmly welcomed as the most promising way yet presented out of the difficulties into which biologists had fallen in their attempts to explain satisfactorily the phenomena of the origin of species through Darwinian selection." 82 An even more enthusiastic worker claimed that "no work since the publication of Darwin's Origin of Species has produced such a profound sensation in the biological world as . . . Die Mutationstheorie by Hugo de Vries." 88 Similar accolades were given by R. R. Gates and by C. D. Davenport.34 One indication of the interest raised by de Vries' theory was the attempt, in the early years of the twentieth century, to find mutations in a variety of organisms other than Oenothera. For example, in 1905 and 1906 C. B. Davenport set about to find examples of animal species which differed by discontinuous variation that would fit under de Vries' concept of mutation.35 Davenport could easily write: As to the correctness of de Vries' conclusions the future alone can give the final decision-doubtless in some points of detail they will have to be modified. The main truth of 31. These included C. B. Davenport, W. E. D. Scott, Thomas L. Casey, D. T. MacDougal, Thomas Dwight, E. G. Conklin, Charles A. White, Jacques Loeb, Frank Shull, T. H. Morgan, F. E. Lutz, R. R. Gates, J. A. Harris, and R. H. Lock. It is curious to note that the last named individual was a Mendelian, and one of the first to apply Mendel's laws to the problem of natural selection. As de Vries himself showed, the mutation and Mendelian theories were not mutually exclusive, and it was quite possible to be a strong adherent of both simultaneously. However, between 1900 and 1910 most workers held one or the other theory most strongly. 32. V. L. Kellogg, Darwinism Today, p. 348. 33. F. C. Baker, "Application of de Vries' Mutation Theory to the Mollusca." Amer. Naturalist, 40 (1906), 327-334. 34. R. R. Gates, "Review of the English translation of Die Mutationstheorie," Amer. Naturalist, 45 (1911), 254-256; C. B. Davenport, "Species and Varieties, Their Origin by Mutation, by Hugo de Vries" [Review], Science, 22 (1905), 369-372. 35. "The Mutation Theory in Animal Evolution," Science, 24 (1906), 556-558.

66

Hugo de Vries and the "Mutation Theory" the vast importance of mutations in the origin of species can no longer be questioned. The reviewer is convinced that as good an argument might be made from the zoological side as de Vries has made from the botanical. Undoubtedly many, if not most of the characteristics of the races of domesticated animals and probably of feral species have arisen by mutation.3ff T. H. Morgan and Jacques Loeb, though independently, were both interested in testing de Vriesian mutations in the fruit fly, Drosophila. D. T. MacDougal and his colleagues at the New York Botanic Garden tried to repeat de Vries' production of mutations in Oenothera itself, as well as in several other varieties of plants. Thomas L. Casey applied the mutation theory to explain the sudden appearance of certain genera of mollusks in early Eocene strata and in certain lower Oligocene rocks, while Schaffner found a variation of Verbena which he considered to be a mutant form living along side the parent species.37 Yet de Vries had opponents from the very first. A number of distinguished biologists both in America and Europe found the mutation theory a naive and insufficiently documented concept to add anything serious to evolutionary thought. Among 36. C. B. Davenport, "Species and Varieties, Their Origin by Mutation, by Hugo de Vries" [Review], p. 372 37. D. T. MacDougal, "Mutation in Plants," Amer. Naturalist, 37 (1903), 737-770; D. T. MacDougal and A. M. Vail, Mutants and Hybrids of the Oenotheras (Publ. no. 24, Carnegie Institute of Washington, 1905); Thomas L. Casey, "The Mutation Theory," Science, 22 (1905), 307-309; John H. Schaffner, "A Successful Mutant of Verbena Without External Isolation," Ohio Naturalist, 7 (1906), 31-34. The list of those who supported the mutation theory could be extended considerably. For example, William E. D. Scott of the Worthington Society for the Investigation of Bird Life, tried to explain the origin of nine obscure species of North American birds on the basis of the mutation theory, though his interpretation was challenged by J. A. Allen. [See W. E. D. Scott, "On the Probable Origin of Certain Birds," Science, 22 (1905), 271-282; and J. A. Allen, "The Probable Origin of Certain Birds," Science, 22 (1905), 431-434]. Even theologians, it was pointed out, could accept the mutation theory with clear conscience and thus avoid the complications of reconciling the biblical account of the origin of animals and plants with the Darwinian: "It [de Vries' theory] should be especially acceptable to the theologians, also, as they maintain the spiritual and undying nature of man. If we can see that man originated abruptly by some unaccountable molecular change in the ovum producing the original twins, Adam and Eve, there can be no doubt of the time when man became thus immortal, whereas there would be necessarily much uncertainty as to the time when this occurred among the successive infinitesimal increments of brain development necessitated by the Darwinian theory" [Thomas L. Casey, "The Mutation Theory," Science, 22 (1905), 308-309].

67

GARLAND E. ALLEN

these were Ludwig Plate and August Weismann in Germany, William Bateson in England, and E. Hart Merriam, C. 0. Whitman, 0. F. Cook, and H. L. Bolley in America. Merriam, for example, made a survey of over 1000 species and subspecies of North American mammals and birds to look for differences which might have originated by de Vriesian mutation. His conclusion was succinct and definite: "My own conviction is that the origin of species by mutation among both animals and plants is so uncommon that as a factor in evolution it may be regarded as trivial." Merriam, in 1906 a member of the U.S. Biological Survey, was disturbed that many of his colleagues rushed so quickly to a theory based on relatively slight evidence. He accused them of being fickle: "Are we, because of the discovery of a case in which a species appears to have arisen in a slightly different way [from the Darwinian theory]-for after all the difference is only one in degree-to lose faith in the stability of knowledge and rush panic-stricken into the sea of unbelief, unmindful of the cumulative observations and conclusions of zoologists and botanists?" 38 Plant and animal breeders, by and large, were opposed to the mutation theory, and in England opposition was strong from the biometricians, who as followers of Darwin tended to emphasize evolution by continuous variation.39 Curiously enough, one writer, 0. F. Cook of Washington, D.C., opposed the mutation theory on the grounds that it attempted to introduce particles into biology analogous to atoms and molecules in chemistry. Cook claimed that de Vries was like many mechanistically oriented biologists of the day who felt they could reduce complex matters such as evolution or species to precise terminology, or as he put it, "to the laws of thermodynamics." 40 Although fewer in number than the enthusiasts, the opponents of de Vries' mutation theory were to have their heyday starting shortly after 1910. In a series of brilliant papers between 1910 and 1912, Bradley Davis showed that the so-called mutations of Oenothera itself were actually the result of an unusual hereditary pattern which basically followed Mendelian laws. In 1914 Renner showed specifically that Oenothera was 38. C. Hart Merriam, "Is Mutation the Factor in the Evolution of the Higher Vertebrates?," Science, 23 (1906), 241-257. 39. See H. L. Bolley, "Observations Regarding the Constancy of Mutants and Questions Regarding the Origin of Disease Resistance in Plants," Amer. Naturalist, 42 (1908), 171-183; T. H. Morgan, A Critique of the Theory of Evolution (Princeton, N.J.: Princeton University Press, 1916). 40. 0. F. Cook, "Physical Analogies of Biological Processes," Amer. Naturalist, 46 (1912), 493-498.

68

Hugo de Vries and the "Mutation Theory" a permanent heterozygote between two complexes, the pure homozygotes failing to survive.4' And in work extending from 1923 to 1950, R. E. Cleland of Indiana University conducted a thorough investigation into the cytology of Oenothera, showing that chromosomal behavior during gametogenesis could explain de Vries' peculiar variations, as well as the persistent recovery of parental types from hybrid crosses.42 Thus the multitude of variant forms of Oenothera were not new species at all, but rather complex recombinations. De Vries kept defending the generality of his theory until his death in 1935, but by 1915 the mutation idea, especially as it was put forth in its original form, had passed out of the serious biological community. When Morgan adopted the term to refer to the white-eyed variant of Drosophila in 1910, he was speaking in a sense much closer to our modern usage then to the largescale jumps of de Vries. DE VRIES AS AN ALTERNATIVE TO DARWIN Those who took up the mutation theory as an alternative to the Darwinian theory of natural selection did so for a variety of explicit and quite specific reasons. It seemed to answer many of the long-standing arguments brought to bear against Darwin's work: the nature of variation, the role of selection, the utility of incipient stages of new characters, the evolutionary time scale, peculiarities of the geological record, the questionable role of isolation, and the rigor of selection. There were also certain underlying and general features of the mutation theory which made it especially attractive to many biologists. I would like to examine first the specific reasons which many biologists cited in their support of de Vries, and then turn to an analysis of the deeper, more general influences which may have operated to mold opinion regarding the role of mutation in evolution. Two issues which were constantly intertwined during the post-Darwinian period were the nature and origin of variation, 41. See A. H. Sturtevant, A History of Genetics (New York: Harper & Row, 1965), pp. 63-64. 42. It would be impossible to describe in detail here the complex chromosomal patterns in Oenothera which Cleland's careful work has uncovered. For further references see: R. E. Cleland, "Chromosome Arrangements During Meiosis in Certain Oenotheras," Amer. Naturalist, 57 (1923), 562-566; R. E. Cleland and A. F. Blakeslee, "Segmental Interchange, the Basis of Chromosomal Attachments in Oenothera," Cytologica, 2 (1931), 175-233; and R. E. Cleland, "Some Aspects of the Cytogenetics of OenotheTa," Botan. Rev., 2 (1936), 316-348.

69

GARLAND E. ALLEN

and the role of selection in producing new species. The first issue resolved itself into two questions: what kind of variations were heritable, and what factors caused variations in the first place? The second issue also reduced to two questions: is selection only a negative factor, or does it have some positive, creative role? And on what kind of variations (continuous or discontinuous) does selection act? Because these two issues are not readily separable, and because much of the evidence adduced for one bore directly on the other, I will discuss them together. Although the question of how variations originated and were inherited was not a new problem, many workers around the turn of the century felt that it still represented the most serious objection to Darwin's theory. In particular, the Darwinian theory seemed unacceptable because there was no evidence to indicate that individual (continuous) variations were inherited.43 Darwin had, of course, assumed that they were, but the lack of any clear evidence had complicated the issue and focused attention on discontinuous variations as the major source of evolutionary material. De Vries' mutations were one such form of discontinuous variation. By the evidence which he presented, de Vries appeared to have shown that large-scale variations were inherited and could produce the kind of differences necessary for the evolution of new species. To de Vries, small individual variations were probably environmentally controlled and thus not directly heritable. By relating degressive and retrogressive mutations to the Mendelian theory, de Vries had shown that these discrete variations were subject to the laws of hybridization. More than any other worker since Darwin, including Mendel, de Vries had addressed himself to the relationship between the types of variations which can occur, and the origin of new species. In the post-Darwinian period a great deal of attention was focused on the issue of how variations originate. In particular, a prominent school of neo-Lamarckians arose between 1880 and 1900, appealing to the principle of inheritance of acquired characters. This is not surprising, considering the fact that Darwin himself, in the Origin of Species, continually referred to adaptations arising through the possible effect of use and disuse, or the direct action of the environment. But because the evidence supporting this principle was sparse, many biologists placed little value in it. Neo-Lamarckians talked enough 43. See, for example, Raymond Pearl, "Some Evolutionary Housecleaning," The Dial, 48 (1910), p. 13; V. L. Kellogg, Darwinism Today, pp. 30-31.

70

Hugo de Vries and the "Mutation Theory" about their theory, but others observed that they were slow in producing experimental support. The weary polemics of August Weismann and Herbert Spencer in the pages of The Contemporary Review between 1893 and 1895 were typical examples of the vacuous debates then current over the origin of variations.44 It was thus with a particularly hearty appetite that many biologists approached de Vries' concrete and expeririental evidence on the origin of variation. De Vries' evidence seemed objective, repeatable, and positive. To many who were skeptical of Lamarckian principles, the mutation theory was particularly welcome since it emphasized the internal rather than external origin of variation. In other words, variations originated from internal physiological causes rather than as a response of the organism to specific environmental conditions. One of the major objections brought against Darwin by his contemporaries arose directly out of the question of variations and their inheritance. According to the prevalent theories of heredity in the 1860's, any new variation was expected to be "swamped" or blended in with the normal traits by crossbreeding, and thus would be passed on in ever-diminishing force to succeeding generations.45 This criticism had forced Darwin and his followers to assume that for any variation to persist, it must occur in two or more individuals at a time. The chances of this were thought to be infinitesimal. Thus, until Mendel's theory of heredity was fully established (and this did not come until after 1910), understanding evolution depended upon understanding how variations could avoid being swamped. It is obvious that the mutation theory solved the difficulty of "swamping" in one stroke. Not only was a mutant form infertile with its parents, thereby preventing its being swamped 44. See for example, Herbert Spencer, "The Inadequacy of Natural Selection," Contemporary Review, 63 (1893), 157; August Weismann, "The All Sufficiency of Natural Selection; a Reply to Herbert Spencer," Ibid., 64 (1893), 609; H. Spencer, "Weismannism Once More," Ibid., 66 (1894), 599; A. Weismann, "Heredity Once More," Ibid., 68 (1895), 435. Such debates contributed little to solving the question of whether acquired characters could be inherited or not. The arguments themselves often degenerated into meaningless debates about why lap dogs do not bite their mistresses, and to what extent this tendency is an acquired or inherited trait. 45. This criticism had first been voiced by Fleeming Jenkin in an article, "The Origin of Species," North British Review, 46 (1897), 149-171. Darwin was deeply troubled by Jenkin's criticism and had no satisfactory answer to give in his own behalf. We know today, of course, that heritable variations are not swamped, since even recessive traits reappear when combined with other recessive traits in the homozygous condition.

71

GARLAND E. ALLEN

through back-crossing, but also during the so-called "periods of mutation" the same mutation would very likely occur several times. Thus a mutation would not be lost to the next generation by being unable to cross-fertilize with another like individual in the population. This advantage of the mutation theory was picked up by a number of writers around the turn of the century.46 It served as an important alternative to the problem which Darwin raised, but had been unable to solve. In emphasizing mutations as the large-scale, discrete steps by which evolution occurred, de Vries placed his work in direct opposition to Darwin and neo-Darwinians. He noted this explicitly in the Introduction to The Mutation Theory when he wrote: The Mutation theory is opposed to that conception of the theory of selection which is now prevalent. According to the latter view the material for the origin of new species is afforded by ordinary or so-called individual variation. According to the Mutation theory individual variation has nothing to do with the origin of species. This form of variation, as I hope to show, cannot even by the most rigid and sustained selection lead to a genuine overstepping of the limits of the species and still less to the origin of new and constant characters.47 The opposition which de Vries saw between his theory and that of Darwin was concerned primarily with the kind of variations on which selection acted. Because mutations were heritable and fluctuating variations were not (or so it seemed), de Vries and his followers felt that the former created one new species difference on which selection could act. At the same time the mutation theory raised the whole question of what role selection itself played in the origin of species. A principal attack on Darwinian theory in the late nineteenth century had focused on the negativity of selection. According to some, selection acted as a sieve which weeded out the unfit, but in and of itself could not create the fit. Selection could not be the direct cause of those alternatives 46. See, for example, E. D. Cope, "The Energy of Evolution," Amer. Naturalist, 28 (1894), 205; C. B. Davenport, "The Mutation Theory in Animal Evolution," Science, 24 (1906), 556-558; and an unidentified reviewer, H.M.R., in Amer. Naturalist, 39 (1905), 747-748. 47. Mutation Theory, I, 4.

72

Hugo de Vries and the "Mutation Theory" from which it selects.48 Or, as one writer put it in a vain attempt to be humorous: "Natural selection may explain the survival of the fittest, but it cannot explain the arrival of the fittest." 49 The argument of negativity seems to have stemmed explicitly from Carl von Nageli, who compared natural selection to a gardener pruning various bushes and trees, without actually creating a new variety.50 A number of other writers pointed out the apparent truth of this argument; for example, E. D. Cope wrote: It has seemed to the author so clear from the first as to require no demonstration, that natural selection includes no actively progressive principle whatsoever; that it must first wait for the development of variation, and then, after securing the survival of the best, wait again for the best to project its own variations for selection.51 The same point was made by Yves Delage in 1903: "The conclusion of this critique is that selection is unable to form species. Its role, nevertheless, is not nul. But it is limited to the suppression of extremely bad variations, and to the maintenance of the species in its normal character. It is less an instrument for the evolution of species than a means of guaranteeing their fixity." 52 Like Delage, Cope, and others, de Vries had seen that selection by itself could not produce evolution. It demanded the origin of new, heritable variations as a starting place. NeoDarwinians had claimed an Allmacht for selection which de Vries and many others could not accept. To de Vries, selection was an important factor; it eliminated the unfit or preserved the fit. But it was new variations-in de Vries' case mutations -which were the creative aspect of the evolutionary process. In this regard de Vries' theory received considerable support from the pure-line experiments of the Danish botanist Wilhelm Johannsen (1857-1927) in 1903. Johannsen had shown that selection of fluctuating variations only separated 48. E. D. Cope, "The Energy of Evolution," p. 205. In other words, selection as a process could not give rise to the initial variations on which it acted. 49. Quoted in V. L. Kellogg, Darwinism Today, p. 89. 50. T. H. Morgan, Critique of the Theory of Evolution, p. 35. The reference is from Nageli's Mechanischphysiologische Theorie der Abstammungslegre (Leipzig, 1884). 51. Origin of the Fittest, p. 175. 52. L'Heredite, 2nd ed. (Paris: Reinwald and Co., 1903), pp. 419-420. Referred to in Kellogg, Darwinism Today, p. 95.

73

GARLAND E. ALLEN

out the pure-lines already existing in a heterogeneous population of organisms. He emphasized that there was a limit to which selection could produce change in a population of organisms. The longer selection was practiced, the less progress could be observed within any one of the lines. He also noted that if selection were relaxed, that is, if the pure line were allowed to hybridize again, the differences between them would soon disappear. Selection seemed to produce changes in a population which persisted only so long as the selection process was rigorously maintained. It could not cause the species to transcend that threshold level of variation between one species and another.53 Johannsen's conclusion opposed Darwinian theory on two accounts. First, it showed that selection of fluctuating variations was not able to produce a new species; second, it showed that the effects of selection by itself were strictly negative. The pure-line experiments provided some important experimental evidence confirming the anti-Darwinian prejudices already in existence. Thus, Johannsen's work contributed significantly to the acceptance of de Vries' mutation theory. A number of biologists saw in the combined arguments the best evidence to date against Darwinian selection.54 The idea that selection could not produce new species was strengthened by the assertion that there was a fault in Darwin's equating of artificial and natural selection. Several writers in the later 19th century had pointed out that even the most rigorous artificial selection (as in animal and plant breeding) had never produced a single case of a wholly new species.55 53. Wilhelm Johannsen, Ueber Erblichkeit in Populationen und in reinen Linien (Jena: Gustav Fischer, 1903). 54. See, for example, T. H. Morgan, Evolution and Adaptation (New York: Macmillan, 1903), p. 298; Thomas L. Casey, "The Mutation Theory," Science, 22 (1905), 307-309; E. G. Conklin, "Problems of Evolution and Present Methods of Attacking Them," Amer. Naturalist, 46 (1912), 121128; C. B. Davenport, "Species and Varieties, Their Origin by Mutation by Hugo de Vries [Review]," Science, 22 (1905), 369-372; D. T. MacDougal, "Discontinuous Variation in the Origin of Species," Science, 21 (1905), 540-543. 55. For example, see Ludwig Plate, Ueber die Bedeutung des Darwin's schen Lehre, 2nd ed. (Leipzig: W. Engelmann, 1903); Morgan, Evolution and Adaptation; Georg Pfeffer, Die Unwandlung der Arten (1894), pp. 19-20. Pfeffer's argument was an interesting one. He claimed that the difference between natural and artificial selection was that the latter was always carried out on the basis of one or two characters, whereas natural selection always operated on the total fitness of the organism. For this reason artificial selection could never produce a wholly new species. Pfeffer's conclusion is wrong, but his recognition that selection acts on total fitness was somewhat singular among German anti-Darwinians.

74

Hugo de Vries and the "Mutation Theory" De Vries made use of this point in maintaining that the forms produced in artificial selection are not as permanent as those produced in nature. Acting on slight individual variations, artificial selection produced only temporary types which, if allowed to hybridize, would quickly revert to their original form. To de Vries the creation of the fit was a result of new mutation. The species differences created, as de Vries thought, by mutation were more in agreement with those differences found among species in nature than with those forms produced through artificial selection.56 For Darwin selection was essential in the creation of new species. For de Vries species arose spontaneously (by mutation); selection merely preserved the favorable and eliminated the unfavorable. Although de Vries thought of his work as an alternative to Darwin on the question of discontinuity, it also suggested a wholly new concept of the temporal relation between the process of selection and the origin of species. Another kind of objection brought against Darwinian theory was that voiced originally by St. Georges Mivart in 1871. As Mivart's argument ran, an adaptive character such as the vertebrate eye could only function when it was fully developed. If it were to have arisen by slight individual variations, the earliest stages would have been nonfunctional and therefore useless to the organism. E. D. Cope also voiced the same objection when he wrote that natural selection could effect the survival of a character only after the character had attained some functional value. Presumably the most rudimentary stages of an eye could not have had any function whatsoever. The mutation theory answered this objection by claiming that a new character arose all at once by a single, progressive mutation. The character would not have to pass through incipient and presumably useless stages in order to arrive at its fullborn form. Mivart's and Cope's objections are of course the result of a failure to understand either the nature of adaptation, or the ranges of selective value which a character could possess.57 One of the most difficult criticisms which Darwin had faced was that posed by William Thompson, Lord Kelvin, concerning the age of the earth. According to Kelvin's calculations, the age of the earth did not allow enough time for selection acting on minute, individual differences to produce the wide variety 56. H. de Vries, "Altere and neuere selektions methode," Biol. Centralbl., 26 (1906), 385-395; Mutation Theory, I, 7. 57. St. Georges Mivart, The Genesis of Species (New York: Macmillan, 1871), esp. chaps. 4 to 9; E. D. Cope, "The Energy of Evolution," Amer. Naturalist, 28 (1894), 205.

75

GARLAND E. ALLEN

of animal and plant forms known to exist.58 Although we know today that Kelvin's figures were off by an order of magnitude, at the turn of the century biologists were forced to reconcile their views of evolution with the then established time scale of the physicists. The mutation theory, on the other hand, fitted in very well with the time scale allotted by Kelvin. By producing a new species from an old one in one generation, mutation significantly reduced the time required for evolution.59 The American biologist Jacques Loeb (1859-1924) was a convert to the mutation theory largely on these grounds. Loeb pointed out another aspect of the time problem which de Vries had contributed to solving. One of the biggest problems with the Darwinian theory and its draft of limitless time was the assertion that evolution occurred so slowly that it could never be observed in a human's lifetime. Loeb found that de Vries' theory was especially attractive because it showed that evolution was observable; it could be studied in the laboratory, and it was subject to experiment. Evolution was not just a black box which could never be investigated directly.60 Another problem on which de Vries threw some light was related to the geological and palaeontological record, which presented two distinct problems: one, the appearance in successive strata of a fossil sequence showing apparently procame to be gressive developments in a straight line-what called orthogenesis; and, two, that there existed distinct, largescale differences in the fossil forms found in two adjacent strata. Orthogenesis was a widespread and popular belief around the turn of the century. According to this idea, there tend to occur is a kind of direction in evolution-variations along one line, as if once started in a given direction, some internal force kept the characteristic varying in that same direction in the future. The initial stages of development along a certain line were not under the influence of selection, but were predetermined by internal causes. Selection came into play only after the character had developed far enough along the line to have either positive or negative selection value. As Kellogg summarized the understanding of orthogenesis in 1907: 58. William Thompson (Lord Kevin), "The Doctrine of Uniformity in Geology Briefly Refuted," in Popular Lecture and Addresses (London, 1894). 59. L. Eiseley, Darwin's Century (New York: Anchor Books, 1961), pp. 229-230. 60. J. Loeb, "The Recent Development of Biology," Science, 20 (1904), 781.

76

Hugo de Vries and the "Mutation Theory" In true orthogenesis the variation, and hence the lines of modification, are predetermined. It seems obvious, however, to any believer in natural selection that sooner or later the fate of these lines of development will come into the hands of selection. And most orthogenesists do indeed admit this. But it is precisely in the making of a start in modification that orthogenesis filled a long-felt want, and if capable of proof should be gladly received by Darwinians as an important auxiliary theory in the explanation of modification, species-forming, and descent.6' Basically teleological, the concept of orthogenesis had its mystical element. Although many orthogenesists were striving to phrase their explanations in mechanical (and therefore presumably non-mystical) forms, it is impossible to divest the theory of the old Aristotelian view of internal drive and striving. The idea of progress inherent in any description of orthogenesis was explicit in the work of Korschinsky, who wrote in 1899: "In order to explain the origin of higher forms out of lower it is necessary to assume in the organism a special tendency towards progress." 62 It was this purposefulness or direction in orthogenesis which filled the "need" Darwinians felt for explaining the fossil record, and for explaining how complex characters got their start. In one form or another many biologists had accepted a theory of orthogenesis by 1900. Among those explicitly advocating such ideas were F. B. Loomis, S. Korschinsky, Carl von Nageli, 0. F. Cook, George Pfeffer, Bashford Dean, Ludwig Plate, T. H. Eimer and C. 0. Whitman. As Kellogg points out, orthogenesis received its greatest support from the American paleontologists: "Cope was a paleontologist, and his belief in the necessity of some factor or factors besides that of natural selection to explain evolution lines as revealed by paleontological study is shared by a large majority of the recognized American paleontologists." 63 Kelogg went on to cite among the paleontologists Henry Fairfield Osborn, Alphaeus Hyatt, and A. S. Packard as strong orthogenesists. To those biologists who held a belief in some sort of directionality to evolution, de Vries offered a handy explanation which did not resort to the teleological implications of orthogenesis. In Darwinian terms, minute individual differences, occurring by chance, might never become established because 61. Kellogg, Darwinism Today, p. 276. 62. S. Korschinsky, "Heterogenesis and Evolution," Naturwiss Wochenschrift, 14 (1899), 273-278. 63. Kellogg, Darwinism Today, pp. 287-288.

77

GARLAND E. ALLEN

of swamping or because they were not initially useful. De Vriesian mutations, on the other hand, were definite, and directional from the start, providing a foundation on which further mutations could build. Followers of de Vries, such as Maynard Metcalf and T. H. Morgan, tried to show in detail how the mutation theory could support the idea of directional evolution, without appealing to the principle of orthogenesis.64 A mutation in one direction, which happened to be favorable, would serve as a foundation on which further mutations in that same direction could occur. Morgan suggested, in addition, that germinal material which was susceptible to variation in a particular direction at one time might be susceptible to the same variation at a later time. The geological problem which encouraged the development of orthogenetic theories of evolution was also conducive to popular acceptance of the mutation theory. Still another problem which biologists faced was determining the role of isolation in the formation of species. Moritz Wagner in the 1870's and 1880's had emphasized the importance of geographic isolation far more than Darwin himself, and this had been taken up by J. T. Gulick and D. S. Jordan around the turn of the century. There were varied schools of thought on the matter: those who claimed that no speciation could occur without isolation, those who held that isolation was a factor in speciation but not a necessary one, and those who claimed that isolation had no role whatsoever in speciation. In his paper of 1905, D. S. Jordan had come closest to our modern thinking in pointing out the importance of isolation in the formation of species.65 64. Maynard Metcalf, "Determinist Mutation," Science 21 (1905), 355356; T. H. Morgan, "Chance or Purpose in the Origin and Evolution of Adaptation," Ibid., 31 (1910), 201-210. 65. Moritz Wagner, The Darwinian Theory and the Law of Migration of Organisms (London: Edward Stanford, 1873); John T. Gulick, "Divergent Evolution through Cumulative Segregation, Journal Land Society of Geology, 20 (1888), 189-274; "Isolation and Selection in the Evolution of Species," Amer. Naturalist, 42 (1908), 48-57; and D. S. Jordan, "The Origin of Species through Isolation." Science, 22 (1905), 545-562. But even Jordan had made an equivocal statement: "Adaptation is the work of natural selection; the division of forms into species is the result of existence under new and diverse conditions" (p. 558). It is difficult to interpret Jordan's meaning here. Does the distinction he makes between "natural selection" and "the division of forms into species" refer to the distinction we would make today between monophyletic speciation and the division of one species into two or more? Or, does it mean that Jordan actually thought that isolation, by itself and without selection, could actually produce species differences which might or might not be adaptive. With the former interpretation Jordan stands as a far-reaching thinker;

78

Hugo de Vries and the "Mutation Theory" With no concept of population genetics, it was difficult either to prove or to disprove the importance of isolating factors in speciation. The mutation theory, coming as it did at the turn of the century, made the factor of isolation virtually unnecessary. With mutant forms being infertile with their parents or normal sibling types, sympatric speciation (to use modern terminology) was just as possible as allopatric speciation. In this case, as in others, de Vries again provided a ready answer to an apparently existing biological stalemate. Behind all of these specific objections to Darwinian theory were two basic and rather general issues: the question of what a species was and how it was defined, and the question of the appropriate methodology for biological research. Let us consider how both of these issues contributed to the popularity of de Vries' mutation theory. Probably the single most general problem shared by most if not all anti-Darwinians at the turn of the century was that of understanding the nature of species. The species concept itself was in a state of considerable confusion at this time.66 Although some far-reaching thinkers such as Karl Jordan and W. M. Wheeler were striving toward a comprehensive and, by today's standards, modern, species concept, the majority of biologists were often confused about one or another aspect of the problem. Nothing better illustrates the extent of such confusion than E. D. Cope's assertion that an organism could change genus without changing speciesl 67 The contemporary situation was ably summarized by the agricultural botanist, L. H. Bailey in 1905, when he wrote: "It would be profitless at this time to enter into a disquisition as to what a species is. Many discussions of this subject are so many admissions that no one knows . . . Our formal nomenclature in practice recognizes only two grades-'species' and 'variety,' with no two persons agreeing which is one or the other." 68 De Vries' work had attempted to clarify the issue by proposing that there were several kinds of species: elementary species and systematists' species. Bailey, for one, felt that with the latter, he shows himself to be much more a man of his own time, such as Gulick, who saw isolation as a factor in its own right, contributing something creative for speciation. A fuller study of Jordan's work would be necessary to establish more clearly what he actually thought about the factors involved in species formation. 66. A. L. and A. C. Hagedoorn, The Relative Value of the Processes Causing Evolution (The Hague: Martinus Nijhoff, 1921), pp. 189-190. 67. Edward J. Pfeifer, "The Genesis of American neo-Lamarckism," Isis, 56 (1965), 156-167; p. 159. 68. "Systematic Works and Evolution," Science, 21 (1905), 532-533.

79

GARLAND E. ALLEN

de Vries had made an important contribution in this regard: "The only point I care now to make is that we all recognize the fact that the single word 'species' covers groups of widely different grades of a value, of differentiations, and of evolutional development. This fact has been brought forcibly to our attention again by the stimulating work of de Vries. There are collective species, elementary species and other grades." 69 Following de Vries, Bailey recognized that species could no longer be defined on purely morphological or anatomical grounds. He held, as did Karl Jordan, something of a dynamic species concept, and was therefore unwilling to accept the old systematic view: These varying grades of species and varieties are all the results of processes of evolution, and some, if not all, of these processes are still in operation. Therefore, the new definition of species concepts must rest on physiological or functional grounds, not merely on morphological and anatomical grounds. Many of us feel that the present methods of nomenclature and description will be outgrown for these methods are made for the herbarium and the museum rather than for the field.70 The changing criteria of defining a species was an indication of the inability of biologists at the time to come to any agreement about either the reality or unreality of species, or the methods by which species groups may have originated. Behind all of this were two views of species characterized by Ernst Mayr as the typological and the nominalist. The former held that each species had its own characters, its own essential type which was a fixed entity. Variations from the type were somehow regarded as abnormal, as differences to be tolerated but as basically unimportant for understanding the true nature of the species. As one writer put it: The process of selective elimination is most severe with extremely variable individuals, no matter in which direction the variations may occur. It is quite as dangerous to be conspicuously above a certain standard of excellence as it is to be conspicuously below the standard. It is the type that nature favors.71 69. Ibid., p. 533. 70. Ibid. 71. H. C. Bumpus, "The Elimination of the Unfit as Illustrated by the Introduced Sparrow, Passer domesticus," in Biological Lectures, Woods Hole Laboratory for 1898 (Boston; Ginn & Co., 1899), p. 219.

80

Hugo de Vries and the "Mutation Theory" The latter view, by claiming that species were arbitrary units, was essentially concerned with the individual in nature. Nominalists pointed out that taxonomy separated one species from another on the basis of characters which were often so trivial as to be of no adaptive significance. Thus the species designated by systematists were not the groups which naturalists saw evolving as organisms with a common set of adaptations. Both typologists and nominalists had some difficulty understanding certain aspects of Darwinian evolution, though the latter did better than the former. To the typologist the idea that the essential character of a species could undergo change by gradual modification rubbed against the grain. He could not get around his conviction that natural selection would work to preserve the type specimen at the expense of the variant. Thus selection, to him, seemed more of a retarder than an accelerator of species change.72 Nominalists had difficulty of another sort. They could understand evolution of individuals by natural selection, but they could not understand the divergence of one population into two groups with species-specific differences. Because they considered the systematists' species as unreal, as having no adaptive features in common, they could not see how natural selection was supposed to account for the origin of such groups.73 It was thus of little wonder that both typologists and nominalists found in de Vries' work a solution to their difficulties. Essentially a typologist in his thinking, de Vries pointed out that mutations produced a complete and sudden change in the type of the organism. The new species type was created in one jump and represented a new center of stability around which individual variations could fluctuate. The cause was internal, and thus seemed to appeal to typologists, who were somehow concerned that any significant change in a species involved the very core of the organism's being. To the American botanist D. T. MacDougal, de Vries' mutation theory was particularly attractive because it produced "new, distinct types," and whether they were called a species or a variety was of little consequence: "That is the main thesis of the mutation theory-the saltatory movements of characters, regardless of the taxonomic value of the resultant forms. That the derivatives might be considered as species by one systematist, and 72. In addition to the article by Bumpus, see Otto Amon, "Der Abanderungsspielraum," Naturwiss. Wochenschrift, 11 (1894), 137-143, 149-155, 161-166. 73. A good example of confusion of this sort can be found in the evolutionary writings of T. H. Morgan, especially Evolution and Adaptation.

81

GARLAND E. ALLEN

varieties by another, is quite incidental and of very little importance." 74

The typologist's view was put even more explicitly by Thomas Dwight, Parkman Professor of Anatomy at Harvard Medical School. In contrasting evolution by natural selection with evolution by mutation, he wrote: "The radical difference between the two theories is this: Darwinism pure and simple is essentially fortuitous; it aims in no particular direction, there is no goal; while mutation by producing suddenly a new species or at least a sub-species, implies the existence of a type and of a law which under certain conditions becomes operative." 75 With de Vries' theory, the type concept could be preserved, since the result of a progressive mutation, was indeed a new type. The old ambiguity of such terms as variety, subspecies, and so on, could thus be dispensed with in favor of real units: an essential organic form qualitatively as well as quantitatively different from all other forms. To the nominalist, mutations produced changes which might or might not be adaptive. As de Vries' own examples of Oenothera had shown, many changes had occurred, particularly in leaf shape, which had no apparent significance to the life of the organism. Thus, the mutation theory, with its lack of concern for adaptation, could explain more easily than the Darwinian theory, with its emphasis on adaptation, the origin of the characters frequently used by systematists to distinguish between species. By appearing to dismiss the confusing terminology and petty concerns of the professional taxonomist, de Vries appealed to many biologists who saw systematics as only a kind of housekeeping activity. As a typologist, de Vries offered a positive alternative to those dissatisfied with the attempt of professional systematists to understand the nature of species. De Vries, however, was not without his critics. A number of naturalists, including C. Hart Merriam, and W. M. Wheeler, saw in his work some obvious errors of taxonomic consideration.76 Both Merriam and Wheeler were seeking a broadened definition of species and thus saw de Vries' work as avoiding, though admittedly not concealing, the issue. But Wheeler, Merriam, Karl Jordan, and a few others were by and large lone voices. The failure of a large segment of the biological community to really understand the species problem was one 74. D. T. MacDougal, "Discontinuous Variation in the Origin of Species," Science, 21 (1905), 540-541. 75. Thomas White, "Mutations," Science, 21 (1905), 529. 76. C. Hart Merriam, "Is Mutation a Factor in the Evolution of the Higher Vertebrates?," Science, 23 (1906), 241-257; W. M. Wheeler, "Ethology and the Mutation Theory," Science, 21 (1905), 535-540.

82

Hugo de Vries and the "Mutation Theory" of the chief reasons for the consistent misunderstanding of Darwinian theory and for the corresponding popularity of de Vries. A final feature of de Vries' theory which attracted a number of workers was that it offered a new methodology for approaching the old problems of evolution. De Vries had emphasized that his work on evolution was experimental, as opposed to the speculative approach characteristic of many neo-Darwinians.77 Strong selectionists such as August Weismann and Ernst Haeckel, who had dominated much of the evolutionary thinking of the later nineteenth century, had indulged in broad speculative theories. In reaction to these methods many biologists, especially among the Americans, had turned in dismay from the Darwinian theory and all of the disciplines associated with it. As E. G. Conklin put it: "In all these speculations fancy occupies so prominent a place and facts are so scarce that it is no wonder that the whole 'phylogeny business' has come into disrepute." 78 It was as if a fresh breeze had blown over stagnant water when the methods of de Vries' mutation theory appeared on the horizon as a substitute for the methods of neo-Darwinism. Having studied with the great plant physiologist Julius Sachs at Wurzburg, de Vries came from an important experimental tradition. Indeed, his own early work had involved studies on the effect of temperature and osmotic concentrations on cell growth, the permeability of plant cell membranes, and biological factors contributing to water pollution.79 These early experiments were beautiful in design and execution, showing that de Vries was an experimenter of considerable ability. The application of experimental methods to a previously non-experimental subject such as evolution was to de Vries an important aspect of his own work on mutations. As he wrote in the preface to Volume I of The Mutation Theory: The origin of species has so far been the object of comparative study only. It is generally believed that this highly important phenomenon does not lend itself to direct observation, and much less, to experimental investigation . . . The object of the present book is to show that species arise 77. Mutation Theory, I, viii-ix. 78. E. G. Conklin, "The Mutation Theory from the Standpoint of Cytology," Science, 21 (1905), 525-529. T. H. Morgan, Conklin's contemporary, was vehement in his opposition to the speculative theories of Weismann and Haeckel. Among botanists, D. T. MacDougal voiced the same objection: see Ibid., 540-543, esp. p. 540. 79. See, for example, Hugo de Vries, Opera e Periodicis Collata (Utrecht: A. Oosthoek, 1918) vol. I, where many of these early papers are reprinted.

83

GARLAND E. ALLEN

by saltations and that the individual saltations are occurrences that can be observed like any other physiological process . . . In this way we may hope to realize the possibility of elucidating, by experiment, the laws to which the origin of new species conform.80 Although de Vries' emphasis on experimentation was in part a reaction to the endless speculation of many evolutionists, it also reflected the tremendous advances in chemistry and physics, and therefore in physiology, of the later nineteenth century. By employing the methods of laboratory analysis developed by the chemists and physicists, first physiologists, and later embryologists, had begun to show how biology could ask precise questions and obtain precise and rigorous data with which to formulate an answer. Wilhelm Roux's program for developmental mechanics ("Entwickelungsmechanik") served as a model for many early twentieth-century biologists who wanted to apply experimental and mathematical analyses to other areas of biology. Jacques Loeb, in his Mechanistic Conception of Life (1911), had further emphasized the experimental nature of biology when he attempted to analyze by physicochemical methods such complex phenomena as fertilization, development, and animal behavior. Enthusiastic about the mutation theory, Loeb saw in de Vries' work the application to evolutionary theory of the same criteria of rigorous experimentalism that he himself had been making to the study of behavior.8' This enthusiasm is not unexpected. When he was working on brain physiology at Wurzburg in the 1880's, Loeb, like de Vries, had been strongly influenced by Julius Sachs. Loeb explicitly took from Sachs the concept of tropism in plants and applied it to the study of animal behavior. But he also imbibed Sachs' animus toward experimentalism as a means of studying complex biological phenomena. Beside Loeb, a large number of early twentieth-century biologists praised de Vries for at last bringing evolutionary studies within the realm of experimental biology. In a review of de Vries' book Species and Varieties, in 1905, C. B. Davenport wrote: De Vries's great work "Die Mutationstheorie" marks an epoch in biology as truly as did Darwin's "Origin of Species." The revolution that it is working is less complete, perhaps, because there has remained no such important doctrine 80. Mutation Theory, I, viii. 81. See Jacques Loeb, The Dynamics of Living Matter (New York: Macmillan, 1906); also Donald Fleming's "Introduction" to the republication of Loeb's Mechanistic Conception of Life (Cambridge, Mass.: Harvard University Press, 1964).

84

Hugo de Vries and the "Mutation Theory" as that of continuity to be established. But there was need of a revolution in our method of attacking the problems of evolution. Ever since Darwin's time most biologists have been content to discuss and argue on the modus operandi of evolution. The data collected by Darwin have been quoted like scriptural texts to prove the truth of the most opposed doctrines. We have seen biologists divided into opposing camps in defense of various isms, but of the collection of new data, and above all, of experimentation we have had little. The great service of de Vries' work is that, being founded on experimentation, it challenges to experimentatation as the only judge of its merits. It will attain its highest usefulness only if it creates a widespread stimulus to the experimental investigation of evolution.82 MacDougal made the same point: For it is now thoroughly realized that the main questions of descent and heredity and of evolution in general are essentially physiological, and as such their solution is to be sought in experiences with living organisms and not by deductions from illusory "prima facie" evidence, which has been so much in vogue in evolutionary polemics, nor by "interpretations of the face of nature" with the accompanying inexact methods and superficial considerations." 83 Speculative methods were both non-testable and inexact. Consequently, no precise or rigorous results could ever be expected. Yet other fields of biology such as embryology had become quantitative during the past several decades; it was the belief of many that de Vries had shown how another area-the study of evolution-could profit from adopting the same methods.84 CONCLUSION De Vries' mutation theory has not stood the test of time. The supposed mutations of Oenothera were in reality complex 82. C. B. Davenport, "Species and Varieties, Their Origin by Mutation, by Hugo de Vries [review]," Science, 22 (1905), 369. In addition, D. T. MacDougal, T. H. Morgan, and J. A. Harris all echoed Davenport's adulation. To many, it was less important whether de Vries' ideas turned out to be correct than that his method marked a revolutionary approach to evolutionary problems. 83. D. T. MacDougal, "Discontinuous Variation and the Origin of Species," Science, 21 (1905), 540. 84. See Kellogg, Darwinism Today, p. 19, where the author states: "Experiment and statistics are capable of mathematical treatment; biology may become an exact science instead of one solely of observation and induction."

85

GARLAND E. ALLEN

recombination phenomena, ultimately explicable in Mendelian terms, while instances of large-scale mutations were found wanting in other species. By 1915 the mutation theory had begun to lose its grip on the biological community; by de Vries' death in 1935 it was almost completely abandoned. Yet, as we have seen, during the first decade of the present century it achieved an enormous popularity. As this paper has tried to suggest, one of the principal reasons for this was that de Vries' theory served as a banner around which a whole crowd of disaffected Darwinians or anti-Darwinians could rally. However, not all of those who favored de Vries did so for quite the same reasons. Underlying the multitude of views ran several common threads: a dissatisfaction with current Darwinian theory born out of misunderstanding natural selection, a general misunderstanding of the nature of species, and a prejudice against speculative, nontestable theories in biology. Supporters of de Vries were not the only opponents of Darwinism, nor was the mutation theory the only alternative to natural selection. In the early twentieth century a number of theories had been proposed to explain away the problems which Darwin had left unsolved. There was the idea of orthogenesis, championed by the American paleontologists Cope, Osborn and others; organic selection (or orthoplasy) was championed by M. M. Baldwin and C. Lloyd Morgan; there were the concepts of convergent evolution proposed by Hermann Friedmann, the theory of physiological selection by John George Romanes, and the concepts of reproductive divergence by H. M. Vernon. Virtually none of these men either accepted or were strong supporters of the de Vriesian theory, for each had his own particular "ism"to advocate as the major factor in evolution. The existence of a large number of such theories, each purporting to be the explanation, was characteristic of evolutionary theory at the turn of the century. It is to a large extent the emphasis on such fragmentary concepts that retarded development of the comprehensive theory of evolution which emerged in the 1920's and 1930's. For the historian, however, a study of these alternative theories is instructive in trying to understand the inherent difficulties which Darwinian theory posed to biologists at the time. De Vries' mutation theory serves historically as a mirror to reflect the critical mood of a generation hostile to the theory of natural selection. It has often been claimed that it was impossible to understand the mechanism of natural selection until it could be placed in genetic and mathematical terms. It is certainly true that great strides have been made in population genetics and

86

Hugo de Vries and the "Mutation Theory" the treatment of evolutionary concepts with mathematical tools in the last forty years. But the very people who developed the genetical and mathematical approach to evolution were already convinced of the essential correctness of Darwinian theory before they started. Advances in an understanding of Mendelian heredity aided greatly in solving one important issue for evolutionists: the origin of variations. And the rigor with which selection acted could best be studied by observing changes in gene frequencies (calculated mathematically) over a number of generations. But as this paper has shown, two of the basic problems which biologists faced in evaluating Darwinian theory at the turn of the century-the nature of species, and the criteria of what constituted an acceptable not be answered diexplanation in biological science-could rectly by mathematics. What mathematical and genetical theory did do was to help convince the skeptics of the validity of the Darwinian proposition. The change in explanatory criteria which many hailed as de Vries' most important contribution to evolutionary theory seems to have been part of a general emergence of twentiethcentury biology from the domination of theorizers in the nineteenth. It also marked the emergence of America from the domination of biological, and particularly evolutionary, influence of Europeans. The change occurred in three areas: in the kinds of questions asked: testable versus non-testable; in the kind of data sought: quantitative versus qualitative; and in the kinds of theories proposed: analytical and reductive the attempt to see complex processes in terms of simpler components-as opposed to synthetic and speculative. Although ultimately wrong in his idea, de Vries and his theories rode high on the wave of "experimentalism" which was the harbinger of a new era in evolutionary theory.

87

Joseph Barcroftand the Fixity of the Internal Environment' FREDERIC L. HOLMES Department of History of Science and Medicine Yale University, New Haven, Connecticut

In 1931 the distinguished British physiologist Joseph Barcroft published in Biological Reviews a long article whose title was a striking sentence written fifty-three years earlier by Claude Bernard: "La fixit6 du milieu interieur est la condition de la vie libre." In opening his discussion Barcroft asserted, "Of the principles which govern the physiological processes of the human body, that of the fixity of its internal environment has been as thoroughly established as any." 2 Barcroft's assessment reflected the fact that during the previous decade the ideas associated with Bernard's vivid phrase "milieu interieur" had attracted greater attention among physiologists than at any other time since Bernard himself had made it a central theme in his teaching. Yet it is difficult to discover what meaning can be attached to Barcroft's statement that the concept had been "thoroughly established." Except for one specialized, though notable exception, the validity of the concept had never been established by experiments designed to that end, nor by assembling all of the data which might argue for it. Few who discussed it, including Bernard, had felt a need to demonstrate empirically its correctness. Neither had it become "thoroughly established" in the sense that physiologists usually referred to it when discussing the 1. After I had read this paper at the CUEBS conference on biological explanation, Professor Donald H. Barron, who collaborated with Barcroft on the investigations of foetal physiology mentioned in the last section, very kindly read the manuscript and discussed with me his recollections of Barcroft's personality and work. To utilize adequately Dr. Barron's very valuable insights would require a more comprehensive study of Barcroft than I have undertaken here, but I have incorporated into footnotes several of the comments which were particularly pertinent. 2. Joseph Barcroft, "'La Fixit6 du milieu int6rieur est la condition de la vie libre' (Claude Bernard)," Biol. Rev., 7 (1932), 24.

89

FREDERIC L. HOLMES

many specific physiological phenomena to which in principle it applied, for most of these were and had long been investigated independently of it within other frames of reference. The concept was, however, becoming established in the sense that in the preceding years a number of leading physiologists had recognized in it a significant point of view around which to organize a broad range of otherwise unconnected phenomena. Ernest Starling, John Scott Haldane, Lawrence J. Henderson, Walter Cannon, and Barcroft were only the most notable of those who had found that the concept brought a new perspective to the collective results of investigations which they and their contemporaries had carried out. Did the milieu interieur, however, merely provide a way for those who took a broad view of physiology to reflect retrospectively upon empirical findings obtained in the pursuit of more specific problems, or did it exert in return some sort of guiding influence upon the choice or course of experimental investigations themselves? By raising this question I hope to illustrate a general historical problem, that of defining the extent and nature of the interactions which have taken place between broad general ideas concerning biological phenomena and the highly specialized activities of experimental biologists. One should be cautious about reaching for definitive answers to such problems, however, for the relationships are not only elusive but dependent upon the personal modes of thought of individual investigators. For this brief discussion I believe I can best illustrate the complexity of the situation by focusing upon the influence of the concept of the internal environment in the work of one man, Joseph Barcroft. Neither Barcroft nor others who utilized the idea of the milieu interieur in the twentieth century seem to have explored the richness of Claude Bernard's thought on the topic as it is presented in numerous writings from 1854 until his death in 1878. Most modem physiologists have relied primarily on the last statement of it he gave in his posthumously published Ph6nomenes de la vie communs aux animaux et aux v6getaux. Though this final statement does not reveal the considerations which earlier led Bernard to formulate the basic idea of the internal environment,3 its prominence as a source of his later influence justifies making it the starting point for the present discussion. 3. I have discussed several aspects of this prior development in "Origins of the Concept of the Milieu Int6rieur," in Claude Bernard and Experimental Medicine, ed. F. Grande and M. Visscher (Cambridge, Mass.: Schenkman, 1967), pp. 179-191; "Claude Bernard and the Milieu Inttrieur," Archives internationales d'histoire des sciences, 16 (1963), 369376; and "The Milieu Inttrieur and the Cell Theory," Bull. Hist. Med., 37

90

Barcroft and the Internal Environment The vital elements composing an organism can only be active, Bernard maintained, when they are in contact with an environment containing the proper physical and chemical conditions. Yet the systems, organs, and tissues of higher animals function in a sensibly uniform manner, unaffected by considerable variations in the external surroundings. This situation is only possible, he believed, because the elementary units of the tissues do not live in the external environment, but within a liquid internal environment formed by the lymph of the circulating blood and the interstitial spaces. The maintenance of a constant internal environment providing the invariant conditions essential to the tissue elements presupposes "such a perfecting of the organism that external variations would be at each instant counterbalanced and equilibrated." This continual, delicate compensation is regulated by the nervous system.4 Bernard discussed four conditions whose preservation in the milieu interieur in proper amounts is essential to the tissue elements: temperature, water, oxygen, and chemical reserves. In summarizing the way each is maintained, he delineated several basic patterns of regulation, though he did not make a special point of these serving as illustrative types. Animals keep a determinate quantitative proportion of water by balancing through nervous control the intake and losses occurring along various pathways. The sensation of thirst drives the animal to make up for losses, whereas urinary excretions act as an overflow to eliminate surpluses. Variations in extemal temperature are equilibrated by the actions of antagonistic nerves which can increase or decrease the amount of heat produced by the chemical reactions in the tissues. Vasomotor nerves also control heat losses by directing more of the circulating blood either toward the periphery or the central organs of the body. The amount of oxygen distributed to the internal environment is limited primarily by its pressure in the surrounding air, but to a certain degree the animal can compensate when this supply diminishes by varying the amount of (1963), 315-335. Dr. Mirko Drazen Grmek has traced the stages of development of Bernard's thought in detail from both published and unpublished writings in "Evolution des conceptions de Claude Bernard sur le milieu interieur," in Philosophie et methodologie scientifiques de Claude Bernard (Paris: Masson et Cie., 1967), pp. 117-150. Joseph Schiller has given a very interesting interpretation of certain aspects of the concept in Claude Bernard et les probl6ms scientifiques de son temps (Paris: Editions du CUdre, 1967), pp. 172-200. 4. Claude Bernard, Legons sur les ph6nom2nes de la vie communs aux animaux et aux vWg&taux(Paris: Bailliere, 1878), I, 112-114.

91

FREDERIC

L. HOLMES

hemoglobin in the blood which absorbs oxygen and by increasing the respiratory movements. "These examples," he concluded, "which we could multiply, prove to us that all of the vital mechanisms, however varied they may be, have always just one objective, that of maintaining the unity of the conditions of life in the internal environment." 6 Besides these mechanisms counteracting the continual variations in the exteral environment, Bernard recognized another kind of control necessary to maintain the chemical composition of the internal environment. Since the nourishment of an animal is intermittent and varying in quality, the materials of the blood could not be formed directly from aliments, but must be produced indirectly from reserve stores into which nutrient matters first go. There they are elaborated and poured into the blood as necessary to keep its constitution fixed. This view Bernard reached by generalizing from his earlier demonstration that sugar is deposited in the liver in the reserve form of glycogen.6 Because of the protection these arrangements afford to the vital tissue elements, Bernard said, the lives of man and higher animals are not bound by changes in the external conditions. "The fixity of the internal environment is the condition of the free, independent life," 7 he added in the phrase which long afterward struck Joseph Barcroft so forcibly. Though Barcroft complained that the meaning of the phrase "free life" was incongruously vague compared to the first part of the statement,8 it is clear what Bernard meant if one compares the preceding situation with what he termed "latent life" and "oscillating life." In the former category he included organisms whose vital activities are completely suspended when any of the normal conditions of temperature, humidity, and oxygenation are absent from their external environments. He showed experimentally that seeds did not germinate when he suppressed any of these conditions. Numerous animals, such as paramecia, rotifers, wheat worms, and certain small arachnids, also ceased temporarily all manifestations of life when they were dried. In these cases there is no internal environment to protect the organisms, so that latency is their only means to avoid destruction by changes in the external environment. "Oscillating life" applied to higher plants and to animals which have internal environments not protected from altera5. 6. 7. 8.

Ibid., pp. 114-122. Ibid., pp. 122-124. Ibid., p. 113. Barcroft, Biol. Rev., 7 (1932),

92

24.

Barcroft and the Internal Environment tions. Here he focused on invertebrates and cold-blooded vertebrates, which can only be fully active when the external temperature is within a certain range and become torpid in colder or warmer conditions. By contrast with these types of life, it is obvious what Bemard considered the fixity of the internal environment to provide for higher animals; a "constant life," in which the animal can exert its maximum vital energy regardless of the normal external fluctuations.9 Joseph Barcroft had ample opportunity early in his career to encounter Claude Bernard's thought. The director and founder of the laboratory of physiology at Cambridge, where Barcroft began, was Michael Foster, a strong admirer of Bernard. Shortly after Barcroft's arrival in 1897 Foster was at work on a biography of Bernard, and the first research Barcroft undertook on the submaxillary gland concerned a problem which he was aware was related to Bernard's research on that organ.10 The teaching in the department of physiology, Foster had said in 1878, was largely based on Bernard's conception of the internal environment." Each of the major lines of investigation Barcroft pursued over the next three decades can be considered to fall within the framework of the concept of the milieu interieur as Bernard had expressed it in 1878. The work of his first ten years Barcroft summarized in 1908 as concerned with "the gaseous exchange between the organs and the medium with which they normally come into contact-the blood." 12 During much of the following fifteen years he was concerned with the special properties of hemoglobin which make it so efficient in the transport of oxygen that the circulatory system is able to deliver that vital element rapidly and in sufficient amounts to the tissues. Taking a synthetic view of his study of respiration in 1913, he wrote that the subject should be considered first of all from the standpoint of the "call" for oxygen of the cells in the active tissues, then of the mechanisms by which the organism responds to that call, rather than by placing pulmonary respiration in the foreground in the more cus9. Bernard, PhWnomAnes, pp. 67-112. 10. Michael Foster, Claude Bernard (London: T. Fisher Unwin, 1899); Joseph L. Barcroft, "The Oxygen Tension in the Submaxillary Glands and Other Tissues," Biochem. J., 1 (1906), 7; Claude Bernard, Le!ons sur les propri6ths physiologiques et les alterations pathologiques des liquides de l'organisme (Paris: J. B. Bailli6re, 1859), I, 299-307. 11. Michael Foster, "Claude Bernard," Brnt. Med. J., 1 (1878), 560. According to Dr. Barron, however, Barcroft was not greatly influenced by Michael Foster himself. 12. Joseph Barcroft, "Zur Lehre vom Blutgaswechsel in den verschiedenen Organen," Ergebnisse Physiol., 7 (1908), 702.

93

FREDERIC L. HOLMES

tomary way.-8 This approach fits very well within Bernard's expressed goal of studying the ways in which the internal environment provides the elements necessary for the life of the tissues. Beginning with an expedition to Teneriffe in 1910, Barcroft investigated for more than a decade the means by which the human respiratory system compensates at high altitudes for the diminished pressure of oxygen in the surrounding atmosphere so that it may maintain as nearly as possible the normal quantity of oxygen in the blood. Between 1923 and 1928 he showed that the spleen acts as a reservoir of blood and of concentrated hemoglobin. It contracts in response to a loss of blood, or to exercise and other conditions which would be expected to increase the volume of the vascular beds in some organs. He interpreted the function of the spleen, therefore, as being to adjust the volume of the circulating blood to the needs of the animal by supplying extra fluids as required from its reserve store. He saw such regulation as helping to prevent injury to vital organs which might otherwise suffer from a lack of blood supply under conditions of stress.'4 Despite the evident relevance of Barcroft's work for understanding the maintenance of the milieu int6rieur, he did not, as far as I have found, ever mention the concept in all the journal articles, lectures, and books which he wrote between 1900 and 1929, nor did he seem in any visible way influenced by it. This silence is especially paradoxical in view of the great importance he attributed to the internal environment in 1932 and afterward. To explain the situation we must consider briefly the position of the concept in physiological thought at the end of the nineteenth century, when Barcroft acquired his initial professional orientation. In his celebrated article on homeostasis in 1929, Walter Cannon reviewed previous ideas concerning the ability of living beings to maintain their own stability by means of selfregulating arrangements which respond to environmental influences tending to alter them. "To Claude Bernard (1878)," he said, "belongs the credit of first giving to these general ideas a more precise analysis." 15 If Cannon's remark were historically accurate one might certainly expect that in the 13. Joseph Barcroft, The Respiratory Function of the Blood (Cambridge [Eng.] University Press, 1914), p. 73. 14. See especially J. Barcroft et al. "A Contribution to the Physiology of the Spleen," J. Physiol., 60 (1925), 452; J. Barcroft and J. G. Stephens, "Observations upon the Size of the Spleen," ibid., 64 (1927), 17-21. 15. Walter B. Cannon, "Organization for Physiological Homeostasis," Physiol. Rev., 9 (1929), 399.

94

Barcroft and the Intemal Environment late nineteenth century, when physiologists were very interested in regulation, they would have traced the source of the problem to Bernard's writing. Bernard was not the first, however, to realize that such functions exist, nor to analyze them and to investigate them experimentally. Only in his later discussions, after 1870, did he come to view regulation as the central problem in the maintenance of the internal environment. Before then, other physiologists had begun to investigate the mechanisms which keep the blood constant, independently of the notion that the blood serves as an internal environment. In 1862, for example, Isidore Rosenthal proposed, on the basis of experiments on the breathing rates of animals placed in varied atmospheres, that respiration is regulated so as to maintain a constant proportion of oxygen in the blood. Other men who pursued this technically complex problem, as well as the regulation of temperature and of properties of blood such as its salt content, had as clear a conception of the features required of such a regulatory mechanism as did Bernard. His own later views on the compensatory regulation of the blood were undoubtedly symptomatic of the growing awareness among physiologists of the importance of such systems, rather than the source of the ideas of others on the topic. It is, therefore, not suprising that studies of the maintenance of the properties of the blood continued in the following decades without reference to the milieu inte'rieur. The significance and novelty of Bernard's view were not in the idea of regulation itself, but that such regulation protects the vital tissues and enables higher animals to achieve relative independence from the external environment.168 Bernard himself seldom undertook experimental investigations planned especially to demonstrate or develop his views about the internal environment, but used the concept primarily to organize and interpret information he had obtained by research on more specific problems. Those who mentioned his idea in the late nineteenth century usually treated it in the same way. There was one important exception. In 1882 Leon Fredericq raised the question of whether the internal environments of higher aquatic animals were more independent of the composition of the surrounding water than those of lower animals. Fredericq's efforts to answer this question made one aspect of the concept of the milieu interieur the direct subject of experimental research. Near the end of the century the topic began to attract expanding interest, and eventually helped 16. The topic of this paragraph is discussed in greater detail in F. L. Holmes, Claude Bernard's Concept of the Milieu Intrieur, unpubl. diss., Harvard University, Cambridge, Mass., 1962.

95

FREDERIC L. HOLMES

considerably, I believe, to spread recognition of Bernard's concept among physiologists.17 At the time Barcroft was entering physiology, however, these studies probably still appeared to those not involved in marine biology as a rather narrowly specialized problem. The limited application of Bernard's generalization in the late nineteenth century reflected also the fact that by the end of his life French physiology had been overshadowed by the burgeoning of research in Germany. Though Bernard attained an enduring international reputation through his experimental discoveries, his general approach to physiological problems, which he had conceived partly in reaction against German trends, did not deeply penetrate their work. I have found few German physiologists of the two decades following Bernard's death who showed any awareness of his concept of the internal environment. In England, where experimental physiology began to develop rapidly only after the founding of laboratories in Oxford and Cambridge during the 1870's, Bernard was honored as an outstanding experimentalist and teacher, as the previous description of Michael Foster's attitude indicates. English physiologists, however, were most likely to choose the well-equipped German laboratories if they went abroad for advanced training and to derive their research problems from leads supplied in the German physiological literature. It is reasonable to suppose from the foregoing that when Barcroft began his experimental research in the Cambridge laboratory he had at least heard of Bernard's statement about the milieu interieur, but it is doubtful that he heard the concept discussed extensively as an essential foundation for the consideration of the properties of the blood, of their maintenance, or of other problems which in principle it encompasses. We cannot, therefore, reach a definite conclusion as to why he did not mention the concept in the early, formative years of his career. Perhaps he had not paid enough attention to it to notice its relevance to his work, or it may be that he was quite familiar with it and still saw no interaction worth mentioning. In order to understand what role the fixity of the internal environment played in his work later, when he did declare its importance, we must nevertheless deal with his research from the beginning in order to see what kinds of considerations 17. F. L. Holmes, "Contributions of Marine Biology to the Development of the Concept of the Milieu Int6rieur," "Colloque International sur 1'Histoire de la Biologie Marine": Vie et milieu, suppl. No. 19 (1965), 321-335.

96

Barcroft and the Internal Environment he habitually relied on in the selection of problems to study and in the evaluation of their significance. If the milieu inteieur was not evident as a principle guiding him in his research choices, one may ask whether some other general biological conception filled that role. Or can his progress be understood primarily in terms of solving more narrowly conceived challenges posed to him? Did he foresee that each of his individual research projects formed a step along the way toward the resolution of some broadly conceived physiological question, or did he merely tackle each immediate problem as it came up, without any long term plan for reaching a future goal? Since I have used only Barcroft's published writings, I can give only a partial answer to such questions, for scientists often do not express in the papers recording their conclusions the hopes and aims with which they began their investigations. Barcroft, however, took time to reassess his work in longer monographs and in lectures where he could be more integrative and less formal in his approach than in his research papers, and less restrained in discussing possibilities not yet demonstrated. Together his writings make it possible at least to retrace some of the steps in the trail he followed. To a large extent the paths of research Barcroft followed can be understood as responses to encounters with immediate, limited problems. As he worked each one out, the solutions he reached, the methods he contrived, or difficulties he encountered opened up new opportunities which often led him in somewhat different directions from that in which he had seemed before to be heading. Sometimes he would also continue with the previous work, so that instead of a single trail his patterns of research resembled a branched one with sevral lines radiating out from key points. He was flexible in defining the over-all problem on which he was working, so that as he exploited new possibilities old work would come to have a somewhat different significance from what it had originally. Most of his effort was fixed on technical means of analysis, but he was alert for outcomes which might hint at some broader physiological significance. There was a clear logic in the relation of each stage in his research to that preceding it, but not necessarily a preconceived goal relating each step to what was to come later, or to any single synthetic, guiding idea. When an individual enters an organized experimental science his first research problem is most likely to be one handed down to him because it is a subsidiary problem within the research interests of those in charge of the laboratory. Bar-

97

FREDERIC L. HOLMES

croft's first investigation was suggested to him this way in 1897 by John Newport Langley, who was then Michael Foster's assistant. For many years Langley had been investigating the process of glandular secretion. Ever since Claude Bernard had investigated the functions of the submaxillary salivary gland this organ had been a favored subject for such work because of the accessibility of its duct, its vascular supply, and its nerve supply. The gland secreted when either the chorda tympani, a branch of a cerebral nerve, or the cervical sympathetic nerve was stimulated, but in the former case the saliva was thin and copious, and the blood vessels in the organ were dilated. In the latter case a smaller volume of more viscous saliva appeared, while the blood supply became constricted. Rudolph Heidenhahn had theorized that there are two types of nerves supplying glands: "trophic" nerves which stimulate the formation of the secretions and "secretory" glands which cause a flow of fluid from the gland. Langley became convinced by his own research that there was no evidence for more than one kind of secretory nerve fiber and that the differences in the character of salivary secretions which had led Heidenhahn to his view could be explained by vasomotor effects. When Barcroft entered the laboratory, Langley encouraged him to try to measure the effects of stimulating the submaxillary gland upon its metabolism, in order to throw light on this and related issues concerning its innervation.'8 Presumably, what Langley had in mind was that if stimulating one of the nerves caused the oxygen consumption and carbonic acid production of the gland to increase, that would indicate that it had brought about an active cellular process of secretion; whereas if the gaseous exchange remained the same any increased salivary flow would be due to passive changes. After Barcroft took up the problem, however, Langley's objective seemed to fade into the background, for the accurate measurement of the changes in the amounts of gases exchanged in the organ became in itself an absorbing problem on which he worked for several years. Several physiologists, including Carl Ludwig and his students, had previously com18. C. S. Sherrington, "Langley, John Newport," The Dictionary of National Biography, 1922-1930 (Oxford University Press, 1937), p. 478; J. N. Langley, "On the Physiology of the Salivary Secretions," J. Physiol., 9 (1888), 55-64; J. Barcroft, Ergebnisse Physiol., 1 (1908), 732; "The Gaseous Metabolism of the Submaxillary Gland, Part III," J. Physiol., 27 (1901), 31-47.

98

Barcroft and the Internal Environment pared the metabolism of individual organs functioning and at rest, but technical difficulties had prevented them from obtaining conclusive results. Barcroft had to refine the methods for extracting blood gases so that he could analyse quickly and accurately much smaller samples of blood than were customarily used. He based his first method on ordinary extraction with a mercury vacuum pump, adding special multiple receivers for the blood arranged for rapid collection and analysis, and eliminating sources of possible leaks in the system.1l In 1901, with John Scott Haldane, he developed a new method by which he could measure the gases contained in one cubic centimeter of blood with nearly the same precision usually achieved with much larger quantities. Placing the blood in a collection bottle attached to a small pressure gauge made of fine-bore tubing and filled with colored water, he displaced its oxygen by adding ferricyanide, and measured the pressure change when the volume was kept normal. He treated the blood next with tartaric acid to displace its carbonic acid and measured the latter gas in the same way. Enabled by such means to work with the small quantities of blood he could collect in a short time from the vessels entering and leaving the gland, he found that the physiological problem was also more complex than previously realized. It was not enough merely to measure the quantities of gases in equal volumes of arterial and venous blood and the amount of blood which flows out of the gland in a given time in order to calculate the rate of exchange, for he found that the hemoglobin in the venous blood was more concentrated because some of the water of the blood had been lost as salivary fluid. After he had taken such changes into account in his calculations he was able to show consistently that stimulation of the chorda tympani increased the oxygen uptake and the carbonic acid output to at least three or four times that of the resting state.20 Though he later discussed the bearing of his results on the issues of salivary gland innervation for which Langley had initially proposed the investigation, Barcroft turned his principal attention in the following years to other explorations 19. J. Barcroft, "An Apparatus for Estimating the Gases of Small Quantities of Blood," J. Physiol., 23 (1898), suppl., p. 64; "The Gaseous Metabolism of the Submaxillary Gland, Part I," J. Physiol., 25 (1900), 265-282. 20. J. Barcroft and J. S. Haldane, "A Method of Estimating the Oxygen and Carbonic Acid in Small Quantities of Blood," J. Physiol., 28 (1902), 232-240; J. Barcroft, "The Gaseous Metabolism of the Submaxillary Gland, Part II," J. Physiol., 25 (1900), 479-486; "Part III," ibid., 27 (1901), 31-47.

99

FREDERIC L. HOLMES

suggested by the existence of the methods he had worked out.21 His techniques for analyzing the gases of small volumes of blood were probably more valuable than the solution of the problem for which he had created them, since they were applicable to many other investigations. He continued in later years to perfect these methods and to adapt them to even smaller quantities of blood.22 Afterward he wrote that the shift from the belief that "the greater the quantity of a substance which could be obtained for analysis the more certain the analysis" to the goal of working with the minimum quantity that could be handled, was one of the most crucial recent changes in respiratory physiology. He found also that with some modification his manometric method enabled him to measure the amount of another substance, urea, in small samples of blood.23 The most obvious way for Barcroft to exploit his success with the submaxillary gland was to apply the same approach to other organs. This was the main direction of his research over the years 1902 to 1908. Taking advantage of the discovery in 1902 by William Bayliss and Ernest Starling that a humoral substance, secretin, stimulates the production of pancreatic juice, Barcroft showed in collaboration with Starling in 1904 that the pancreas also consumes more oxygen and releases more carbonic acid when it is secreting than when it is not. Next he turned to the kidney and to the heart, finding in each case a correlation between increased or decreased activity and the rate of metabolism of the organ. With each organ he encountered special technical problems and the question of defining what was the nature of the work the organ did.24 By 1914, however, he could sum up the outcome of this line of research with the generalization that "ia no organ excited by any form of stimulus can it be shown 21. Barcroft, Ergebnisse Physiol., 7 (1908), 738-741. It is not surprising that Barcroft did not continue to focus on the problem from the point of view that had interested Langley, in light of Dr. Barron's comment that Langley and Barcroft had remarkably little to do with each other during the long time in which both worked at the Cambridge physiology laboratory. 22. J. Barcroft and F. F. Roberts, "Improvements in the Technique of Blood-Gas Analysis," J. Physiol., 39 (1909), 429-437. 23. J. Barcroft, Respiratory Function, pp. 194-196; "The Estimation of Urea in Blood," J. Physiol., 29 (1903), 181-187. 24. J. Barcroft and E. H. Starling, "The Oxygen Exchange of the Pancreas," J. Physiol., 31 (1904), 491-496; J. Barcroft and T. G. Brodie, "The Gaseous Metabolism of the Kidney," ibid., 32 (1904), 18-27; ibid., 33 (1905), 52-68; J. Barcroft and W. E. Dixon, "The Gaseous Metabolism of the Mammalian Heart, Part I," J. Physiol., 35 (1907), 182-204.

100

Barcroft and the Internal Environment that positive work is done without the blood supply having to respond to a call for oxygen." 25 Between 1906 and 1909 the main focus of Barcroft's research shifted from the problem of the metabolism of individual organs to that of the oxygen dissociation curve of blood. The transition emerged in stages, through the convergence of several factors which were perhaps inherent in the logic of the problems with which he was dealing, but whose influence upon the direction of his investigations was due also to the partly fortuitous order in which he encountered them. Because he had been, in his early work, interested primarily in the quantities of oxygen and carbon dioxide exchanged by the organs, Barcroft had in his blood-gas analyses dealt only with the amounts of these substances in the arterial and venous blood samples. In 1905, however, Christian Bohr, then a dominant figure in the field of respiratory physiology, made use of Barcroft's experiments on the submaxillary gland in a way that Barcroft himself probably had not foreseen. Bohr and others had been investigating the relation between the percentage saturation of the blood with oxygen and the "tension" of the oxygen in the blood, the latter being proportional to the pressure of oxygen in the atmosphere with which the blood sample was in equilibrium. They were attempting to construct characteristic curves with which they could calculate either quantity when the other was known. In his discussions of the gaseous exchanges between the blood and tissues, Bohr emphasized that the quantity of oxygen immediately available to the cells does not depend on a simple, direct relation to the quantity of oxygen in the blood passing through a given tissue. The proportion of the hemoglobin in the blood saturated with oxygen only indirectly determines the amount which reaches the cells by its influence upon the oxygen tension in the blood plasma, for the quantity of oxygen in the intercellular fluids is proportional to the oxygen tension in the venous blood. These relationships, Bohr wrote in an article for Nagel's Handbuch der Physiologie, should be "taken to heart." Accordingly, when he used Barcroft's experiments as evidence of the means by which the oxygen supply to an organ increases in conformity to increased needs, he calculated from Barcroft's figures for the quantities of oxygen the tension of the gas in the venous blood plasma. This showed a 48 percent increase after the gland was stimulated. The result not only explained the mechanism by which oxygen is delivered at a greater rate to the cells from the blood through the intercellular 25. Respiratory Function, p. 105.

101

FREDERIC L. HOLMES

fluids, but also, Bohr suggested, made understandable the fact that oxygen had been found in the secreted saliva itself.26 Undoubtedly impressed by the application Bohr had made of his "own figures," Barcroft returned in December 1905 to the submaxillary gland to take up the problem inherent in Bohr's approach: how oxygen is transferred from the blood to the secretory cells and the saliva. Barcroft noticed, however, a striking anomaly; the available analyses of saliva showed that it contains a larger percentage of oxygen than the blood plasma does according to Bohr's calculations. It was hard to imagine how this concentration could occur. Furthermore, his own calculations of the rate of oxygen consumption required that, if oxygen is delivered to the cells simply by means of its dissolution in the intercellular fluid, as Bohr envisioned, the oxygen tension would have to be much higher than Bohr's calculations indicated. The only provisional way out of the dilemma Barcroft could see was that "there is something active, some oxygen secreting quality in the capillary wall." He raised the alternative possibility, however, that the oxygen dissociation curves might have been incorrect and have given a calculated oxygen tension in the blood plasma which was too low. He also wondered whether there was some other constituent of the blood which "played the role of tuming the oxygen out of the haemoglobin in the capillaries." The problems Barcroft raised in this paper, apparently in response to Bohr's discussion, became in the following years prominent themes in his work.27 It quite likely was his conjecture that the impasse he had discerned might be resolved if the oxygen dissociation curves were wrong that stimulated Barcroft to examine more closely the previous literature on the subject. He found that the results given by different investigators and by the same investigators in different cases were widely discordant. He decided, therefore, to check the curves for himself, encouraged in the undertaking by his recently improved methods of blood-gas analysis which would enable him to make "in a few days analyses that would previously have taken many weeks." He anticipated that he could assess the accuracy of the curves with little difficulty.28 26. Christian Bohr, "Blutgase und respiratorische Gaswechsel," in Handbuch der Physiologie des Menschen, ed. W. Nagel (Braunschweig: Vieweg, 1909), I, 84-93, 197-201. The volume I used must have been a later printing, although there is no such indication. Barcroft referred to Bohr's article in a paper of 1906. 27. J. Barcroft, "The Oxygen Tension in the Submaxillary Glands and Certain Other Tissues," Biochem. J., 1 (1906), 1-10. 28. Respiratory Function, pp. 41-42.

102

Barcroft and the Intemal Environment The outcome, however, was otherwise. At first he and his research associate, Dr. Mario Camis, seemed to obtain the uniformity of results which had eluded their predecessors. Blood of different kinds of animals, which before had given divergent results, came out the same for them. But when they tried human blood they could not obtain a similar concordance. As Barcroft later related it: '"e had found our way into the morass in which our predecessors had already floundered so hopelessly, and our newer and more certain methods instead of saving us from their embarrassments had only made the uncertainty of our position more certain . . . We then entered upon six months of research which became weekly more depressing." 29 Success finally came when Barcroft and Camis discovered that a variation in the salts present in a solution of hemoglobin altered its dissociation curve for oxygen. Systematically examining the effects of other changes in the salts, they found that they could reproduce curves consistently whenever they kept the composition of the solutions constant. By imitating the salts present inside the red cells they could duplicate with the hemoglobin solutions the curve for intact blood. The differences others had found in the curves for the blood of different animals then became explainable in terms of known differences in the salts contained in their red cells.30 Since he had brought order into a confused situation after a long, troublesome effort, it is not surprising that what he had begun in order to clear up a technical difficulty in the pursuit of another objective Barcroft now turned into the principal subject of his ensuing research. He followed two distinct, though interlocking lines of investigation. First, primarily in collaboration with A. V. Hill, he tried to characterize the oxygen dissociation curves mathematically and to derive from the characteristics of the reaction a theory of the mechanism of the interaction between hemoglobin and oxygen.31 Second, he continued to study the conditions which modify the curve. Bohr had already identified one of the principal factors affecting the dissociation of oxygen. Carbonic acid in the blood displaces the curve in such a way that at a given oxygen tension less oxygen remains combined with the hemoglobin than otherwise; or with a given concentration of oxygen the 29. Ibid., p. 42. 30. J. Barcroft and Mario Camis, "The Dissociation Curve of Haemoglobin," J. Physiol., 39 (1909), 118-142. 31. J. Barcroft and F. F. Roberts, "The Dissociation Curve of Haemoglobin," J. Physiol., 39 (1909), 143-148; J. Barcroft and A. V. Hill, "The Nature of Oxyhaemoglobin, with a Note on its Molecular Weight," ibid., 39 (1909), 411-428.

103

FREDERIC L. HOLMES

tension is raised. He had discussed the biological significance of this property for regulating the delivery of oxygen to the cells. Carbonic acid resulting from the metabolism of the tissues enters the blood in the capillaries, and by the modification it produces in the affinity of hemoglobin for oxygen it helps to displace the latter into the plasma and so to the cells. Since an increased need for oxygen in the cells ordinarily occurs together with increase in the production of carbonic acid, the effect becomes larger as the need becomes greater.32 Barcroft followed Bohr's example in looking for adaptive advantages in the influences of various conditions on the dissociation curve. A rise in temperature from 360 to 410 C, he found, shifts the curve so that at a given tension the blood is less saturated with oxygen. He concluded that this effect could aid in meeting the greater demand for discharging oxygen in muscles during violent activity, for the heat produced in very active muscles may cause a local rise in temperature. He investigated similarly the influence of lactic acid added to the blood, to see whether the remarkable action of carbonic acid was specific to it or due to a general ionic effect. The lactic acid modified the curves in the same way carbonic acid had, and Barcroft ascribed to it an equivalent physiological value; for when an animal is under stress in exercise its muscles discharge lactic acid into the blood.88 As Barcroft continued these investigations, the general question that had confronted him in 1906, how to account for the rapid transfer of oxygen from the blood to the tissues necessary to meet the enormously enlarged demands of active organs, became a persistent element in his thought. In his first paper on the dissociation curve Barcroft considered some implications of its properties for conditions at high altitudes. It had recently been discovered that at great heights the pressure of the carbonic acid in the lung alveoli of a man, and presumably therefore in his blood, was substantially decreased. Barcroft speculated that, in accordance with the known effect of carbonic acid on the dissociation curve, the latter would shift so that at a given oxygen tension the blood would take up more oxygen. This change would help the blood to acquire oxygen in the lungs in the face of the drop in oxygen pressure, although he acknowledged that it would be correspondingly more difficult for the blood to dis32. Bohr, "Blutgase," pp. 93, 196-199. 33. J. Barcroft and W. 0. R. King, "The Effect of Temperature on the Dissociation Curve of Blood," J. Physiol., 39 (1909), 374-384; J. Barcroft and L. Orbeli, "The Influence of Lactic Acid upon the Dissociation Curve of Blood," ibid., 41 (1910), 355-367.

104

Barcroft and the Internal Environment charge its oxygen in the tissues.34 It is hard to tell whether these ideas would have led Barcroft to make high-altitude physiology one of his major research interests if it had not been that, about the time he was thinking about this problem, Professor Nathan Zuntz invited him to join an expedition to the Island of Teneriffe. The group, whose purpose was to examine "the biochemical effects of high climates and solar radiation," operated from three stations on the island, at sea level, at 7000 feet, and at 11,000 feet.35 Contrary to his expectation, when Barcroft measured the dissociation curve of the blood at high altitude it was the same as that at sea level. If, on the other hand, he analyzed the blood sample taken at high altitude after putting it in equilibrium with a carbonic acid pressure equivalent to the alveolar pressure at sea level, the curve was displaced. Consequently the blood itself must have changed in such a way that its dissociation curve at the lower pressure duplicated that of normal blood at a higher pressure. He thought that probably some substance such as lactic acid was taken up in the blood in place of the decreased carbonic acid. "This change," he concluded, "compensated the change in alveolar CO. tension to such an extent that the actual dissociation curve under the conditions locally established did not alter." He considered it probable that this was one of the mechanisms produced in the process of evolution that enabled "the respiratory process to adapt itself to various unusual conditions." 33 After returning from Teneriffe, Barcroft continued with the two basic lines of research that he had established. He investigated the metabolism of the submaxillary gland, the kidneys, and the liver, becoming more deeply involved in the special problems each organ presented for measuring the gaseous exchange, the oxygen consumption, the work performed, and the relation between circulatory changes and activity.37 At the same time he penetrated further into the physical chemistry of hemoglobin. In 1910 A. V. Hill developed a mathematical equation from which he could derive 34. Barcroft and Camis, J. Physiol., 39 (1909), 132-133. 35. J. Barcroft, "The Effect of Altitude on the Dissociation Curve of Blood," J. Physiol., 42 (1911), 44. 36. Ibid., 44-63. 37. J. Barcroft and Franz Muller, "The Relation of Blood-flow to Metabolism in the Submaxillary Gland," J. Physiol., 44 (1912), 256-264; J. Barcroft and H. Piper, "The Gaseous Metabolism of the Submaxillary Gland with Reference Especially to the Effect of Adrenalin and the Time Relation of the Stimulus to the Oxidation Process," ibid., 359-373; J. Barcroft and L. E. Shore, "The Gaseous Metabolism of the Liver," ibid., 45 (1912), 296-306.

105

FREDERIC L. HOLMES

the oxygen dissociation curve, and from which he could make deductions concerning properties such as the state of aggregation of the hemoglobin molecules. The equation also enabled Barcroft to treat more precisely and quantitatively the alterations of the dissociation curve with altitude and with other changes in conditions.38 Thus his research still seemed to be following a pattern of natural evolution in which the resolution of one problem opened up related problems or opportunities, and to be leading him in more than one direction at the same time. In 1913, however, he wrote a book entitled The Respiratory Function of the Blood, in which most of his investigations of the preceding sixteen years came to appear as facets of an integrated account of one general physiological problem; that is, the mechanisms by which the blood serves as a remarkably efficient vehicle for transporting oxygen to the cells according to their needs and under wide variations in the availability of the oxygen in the surrounding atmosphere. Barcroft began his discussion with five chapters on the oxygen dissociation curves of hemoglobin and on theories about the nature of the chemical combination these two substances form. This knowledge is the necessary foundation for the whole topic, he pointed out, for it is upon the unique properties of hemoglobin that the possibility of the blood transporting oxygen efficiently enough to support the metabolic needs of the tissues depends. Next he reviewed his various investigations of the increase in oxygen consumption by organs when they become more active, linking these studies with those on the blood with a statement which also made clear the general point of view about respiration that he had reached: The classical work of Pfluger on the combustion of living material settled for all time, it seems to me, the logical order in which the constituent processes of respiration should be treated. The issue before Pfluiger [was] .. . Is the quantity of oxygen taken up by the cell conditioned primarily by the needs of the cell, or by the supply of oxygen? The answer was clear, the cell takes what it needs and leaves the rest. Respiration therefore should be considered in the following sequence. Firstly, the call for oxygen, secondly the mechanism by which the call elicits a response, the immediate response consisting in the carriage of oxygen to the tissues by the blood and its transference from the blood to the cell. Thirdly in the background 38. J. Barcroft, et al., Proc. Physiol. Soc., Jan. 18, 1913, pp. xlv-xlvii, in J. Physiol., 45 (1913); J. Barcroft, et al., Proc. Physiol. Soc., February 15, 1913, pp. iv-v, and December 13, 1913, p. xxvii, ibid., 47 (1913).

106

Barcroft and the Internal Environment you have the further mechanism by which the blood acquires its oxygen.39 Following the summary of his studies of the metabolism of organs Barcroft tumed to a closely related question on which he had recently begun to focus. "Unless the tissue when at rest is to be flooded with large quantities of unnecessary blood, the variations in its blood supply must be of the same order as those of the oxygen required." 40 Several earlier physiologists had maintained that organs can themselves regulate their supplies of blood by releasing substances into the blood which cause the local blood vessels to dilate. From his own recent work on the submaxillary gland Barcroft added what he considered to be a conclusive instance of this control. He dealt next with the question of how oxygen is transferred from the blood to the tissue cells, subdividing the problem into two parts: the chemical breakdown of oxyhemoglobin and the diffusion of the gas through the plasma to the cells. For the first time he brought together his previous findings concering the factors which influence the oxygen dissociation curve favorably for unloading it in the tissues. He pointed out, however, that these studies had been based on equilibrium conditions and did not show how rapidly the oxygen can be taken up in the lungs or discharged in the tissues. Drawing on recent experiments he and his colleagues had been doing to measure the rates of these reactions, he speculated on a mechanism which might explain the adaptability of these rates to the varying requirements of the organism. Barcroft's discussion of the "third mechanism" of respiration, that by which the blood acquires oxygen in the lungs, was dominated by a special issue over which physiologists were currently divided: whether physical diffusion is a sufficient explanation for the transfer of oxygen across the epithelium of the lungs or whether there is an active secretion. In the last section of the book he discussed his experiments on the effects of altitude upon the dissociation curve and the adaptive significance of the results. He had found, for example, on further mountain expeditions that a combination of high altitude and exercise caused lactic acid to enter the blood and increase its acidity, even while the breathing rate increased and lowered the concentration of carbonic acid. The lactic acid shifted the dissociation curve so that the saturation of hemoglobin was lower at a given oxygen tension. The 39. Respiratory Function, p. 73. 40. Ibid., p. 137.

107

FREDERIC

L. HOLMES

change probably somewhat decreased the degree of saturation the blood acquired in the lungs, but increased circulation and other factors could in part make up for that. In return the altered curve meant that the blood would discharge its oxygen more rapidly in the tissues, as needed in view of more rapid movement of the blood.41 The fact that he could incorporate most of his previous work into the synthesis his book provided may seem to indicate that Barcroft must have been guided after all by a similar overall view in the original choice of his research problems. Some support for this interpretation comes from an abstract of a lecture he gave on respiration in 1902 to the Belfast Natural History and Philosophical Society. "The problem [of respiration]," he had said then, "was to investigate the processes by which the oxygen of the inspired air is carried to the hidden recesses of the body, and those by which the carbonic acid is carried from the tissues to be cast out into the air of the lungs." 42 The coincidence between this delineation of a problem and the range of the subject covered by his research as portrayed in his treatise of eleven years later gives the impression that his investigations in the intervening years may be seen as the carrying out of a program designed to fill in the details of a general biological conception. Obviously, in a very broad sense this is true; all of his research in this time was devoted to problems of respiration. Yet the preceding tracing of the stages in his development shows that he probably did not have in mind all aspects of the problem at the beginning, but that his view progressively widened through the responses he made to more limited problems. It would be difficult to establish when Barcroft did begin to envision the results of his research fitting into the pattern which his book embodies. Perhaps the organization of ideas was clear to him considerably earlier, but his research papers did not afford him an appropriate means for expressing it. On the other hand, the pattern may have emerged more nearly as in the case of a later book, of which he said in its preface that he had started out to describe functional principles of the body which he thought of as "unconnected features of its architecture." Only as he wrote did it become clear to him that each of the principles was dependent upon the others.48 41. Ibid. 42. J. Barcroft, "Respiration," Report and Proceedings of the Belfast Natural History and Philosophical Society, Jan. 6, 1902, p. 27. 43. J. Barcroft, Features in the Architecture of Physiological Function (New York: Macmillan, 1934), p. ix.

108

Barcroft and the Internal Environment With little modification the synthesis Barcroft produced in his book on the Respiratory Function of the Blood could serve as a detailed exposition of the mechanisms that Claude Bernard had hinted at, which maintain the constancy of one of the crucial conditions of the milieu interieur. Yet, besides making no mention of Bernard's generalization, Barcroft did not really see the problem in the same terms. He viewed the blood as a vehicle for delivering to the cells the oxygen they consumed, not as a means for maintaining a steady concentration of oxygen in the environment of the cells. The distinction may seem trivial if one is concerned more with hard data than with conceptual shadings. If we consider the case of oxygen by itself the difference is largely verbal, and in fact Bernard's own discussions of oxygen fit as easily into the view of the blood as a supplier of oxygen as it did of the blood as a medium including oxygen; but the concept of the internal environment embraced also the maintenance of physical and chemical conditions which cannot be conceived merely in terms of something used up in the cell. The milieu interieur therefore is in a sense a broader biological conception than that which regards the blood as a delivery system. So long as Barcroft worked only on the problem of respiration these differences, however, might not have seemed appreciable to him, if indeed he had yet given much thought to Bernard's view. In the meantime other physiologists were beginning to pay more serious attention to the concept of the internal environment than had been customary when Barcroft entered the field. The studies of the internal environments of aquatic animals grew into a substantial research problem during the first decade of the century. In 1909 Ernest Starling wrote a treatise on body fluids incorporating the conception into his discussion of water balance in the body44 Then in 1917 one of Barcroft's most prominent colleagues in respiratory physiology, John Scott Haldane, published a series of lectures in which he viewed the regulation of respiration as a prime instance of the maintenance of the internal environment. Haldane summarized his past investigations of the regulation of breathing and the role of the blood in respiration, incorporating the results of Bohr, Barcroft, and others in addition to his own. He used the overall picture of respiration that emerged as a focal point for raising very fundamental questions about the nature of organisms, their relation to 44. Ernest H. Starling, The Fluids of the Body (Chicago: W. T. Keener, 1909).

109

FREDERIC L. HOLMES

their environments, and the explanatory principles appropriate to biology. One of these was that "it is characteristic of an organism to react towards disturbing influences in such a way as to maintain approximate constancy in its structure, internal environment, and even external environment." 45 Haldane's interest in Bernard's concept, then, was obviously due in part to its harmony with his own philosophical position about vital phenomena. The concept of the milieu interieur also fit well, however, with a more specific view he took toward the results of investigation of the properties of the blood. One of the general conclusions he had reached from his research on breathing was that the respiration is very delicately regulated so as to maintain a constant hydrogen-ion concentration in the blood in spite of the innumerable influences tending to alter it.43 There is also, Haldane said, a regulation of the circulation of the blood which serves to "keep the removal of COQfrom the body tissues and their supply of oxygen steady." Here Haldane's view of the role of the blood nearly corresponded with that of Barcroft, but Haldane also looked at the relation between the tissues and the respiratory gases of the blood from another perspective: The blood circulates at such a rate as is sufficient to keep its composition approximately constant at any part of the body, and the rate of flow seems to be greater or less at any one part in proportion as the causes tending to disturb the composition of the blood are greater or less at the same part. Among the chief of these causes [of disturbance] is consumption of oxygen and liberation of CO,. Hence the circulation rate is to a large extent determined by the activity of the latter processes, and varies, just as the breathing varies, in such a way as to keep the gas pressures in each part of the body approximately constant.47 Haldane did not, therefore, think of the variations in the oxygen needs of the tissues only as calls for supplies to be met by the blood, but also as disturbances to the constancy of composition of the blood to be adjusted for. He applied the same view to other constituents of the blood. He continued, "This is not an isolated fact in physiology. Claude Bernard pointed out in 1878 in his Legons sur les phenomenes de la vie 45. John Scott Haldane, Organism and Environment as Illustrated by the Physiology of Breathing (New Haven, Conn.: Yale University Press, 1917), p. 94. 46. Ibid., p. 37. 47. Ibid., p. 76.

110

Barcroft and the Intemal Environment that the blood is a fluid of remarkably constant composition, and practically provides a constant internal environment for the living cells." Haldane discussed a few other examples of this constancy, including his own recent findings concerning the regulation of the proportions of water and of salts in the blood.48 In 1922 Haldane again discussed the maintenance of the internal environment in the last chapter of his long monograph on respiration. Here he explicitly drew a distinction between "salts, water, and various other substances" of the internal environment which "are to only a very small extent used up or given off from the tissues," and oxygen which must be supplied in large quantities to the tissues. Even for oxygen, however, he disagreed with the usual way of regarding it as a supply passing in one direction into the tissues to be used up in an irreversible reaction. Oxygen, like other constituents of the environment of the cells, passes in both directions. Respiration and circulation are therefore regulated to maintain a steady diffusion pressure of oxygen in the internal environment, so that the balance between the inward and outward movement of that gas with respect to the cells is not disturbed.49 In this discussion Haldane remarked that his long series of investigations of respiration "may be regarded as an attempt to follow out in regard to blood reaction and oxygen supply" Bernard's idea that all vital mechanisms have the object of preserving intact the internal environment.50 Garland Allen has shown in a recent study of Haldane, however, that his research actually developed from more specialized, practical problems, and that only afterward did the milieu interieur become for him a way to relate his particular work to broader physiological issues.5' Barcroft was undoubtedly aware of the significance Haldane attributed to the milieu interieur with respect to the problems of respiration,52 but he still made no mention of the concept in his own writings in the years following the publication of Haldane's treatises. Barcroft did, however, begin to 48. Ibid., pp. 77-81. 49. J. S. Haldane, RespiTation (Yale University Press, 1922), pp. 382386. 50. Ibid., p. 383. 51. Garland E. Allen, "J. S. Haldane: The Development of the Idea of Control Mechanisms in Respiration," J. Hist. Med., 22 (1967), 392-412. 52. Besides the closeness of their previous association and interlocking of their research interests, which would make it very likely that Barcroft would read anything Haldane wrote, Barcroft reviewed Haldane's Respiration in Nature, 110 (1922), 803-804.

ill

FREDERIC

L. HOLMES

include in his research concems the problem of the hydrogen-ion concentration of the blood. Sometimes he described the relation between tissue activity and the maintenance of this property in the blood in a manner resembling Haldane's view. At the same time that Haldane was demonstrating the importance of the hydrogen-ion concentration in the regluation of breathing, Lawrence J. Henderson was showing the extent and complexity of the buffer mechanisms within the blood which stabilize its reaction. In 1922 Barcroft too began to investigate the relation between the concentration of hydrogen and other properties of the blood, including the oxygen dissociation curve.53 Instead of considering hemoglobin exclusively as a carrier of oxygen as he once had, he now became interested in its role as a buffer. In a report of the same year on his recent research carried out in the Andes he mentioned that the buffering properties of hemoglobin may explain the significance of the increase in its concentration in the blood at high altitudes, and he implied that he was following a trend in expanding his view of the function of that substance: The reason for the concentration of corpuscles at high altitudes has in recent years been somewhat of a mystery. Formerly it was supposed that the object to be obtained was merely the transport of a certain quantity of oxygen in the blood. More recent workers have recognized the insufficiency of this idea. It would now seem probable that the extra degree of buffering which the blood acquires is an important factor.54 In a review of "The Significance of Hemoglobin," Barcroft expanded on this theme in 1924. Although oxygen transport was still obviously the primary function of hemoglobin, he also regarded its action as a buffer as very important-important because "by buffering the blood, haemoglobin incidentally buffers the tissues." This power, he concluded, "makes it possible for the tissues to produce carbonic acid on a large scale and yet not alter the hydrogen-ion concentration to a serious extent." 65 53. J. Barcroft, et al., "On the Hydrogen-ion Concentration and Some Related Properties of Normal Human Blood," J. Physiol., 56 (1922), 157178. 54. J. Barcroft, et al., "Observations upon the Effect of High Altitude on the Physiological Processes of the Human Body, Carried Out in the Peruv379. ian Andes," Phil. Trans. Roy. Soc., [BJ, 211 (1921-23), 55. J. Barcroft, "The Significance of Hemoglobin," Physiol. Rev., 4 (1924), 341-346, 350.

112

Barcroft and the Internal Environment By 1924, therefore, Barcroft's view of the function of hemoglobin was broadening in a direction which approached the physiological point of view generalized in Claude Bernard's concept of the milieu inte'rieur. Although familiar with the connection between them through Haldane's writing, he was not discussing such a connection himself. Further, his developing view can be accounted for as a continued progression in response to current experimental problems and to the interactions between his work and the research of his colleagues. A few years later Barcroft was describing the role of hemoglobin and numerous other phenomena as among the physiological processes governed by the principle of the fixity of the internal environment.56 I am unfortunately not able from Barcroft's published writings to chart in detail his course between these two stages. One of the chief obstacles is that there is apparently no published account of the first presentation he gave of his ideas on the subject. In October 1929 he delivered a talk entitled "The Constancy of the Internal Environment" as one of a series of lectures at the Harvard Medical School which "excited considerable interest." 57 In a footnote to the article on the milieu interieur he submitted for publication in August 1931, Barcroft remarked that the paper contained "the essential points" given in that lecture and later expanded into three lectures in London University.58 Much of the crucial material for the published article, however, is drawn from research by Barcroft, his collaborators, and others published in 1930 and 1931, so that it is difficult to tell what aspects of his general ideas and of the evidence on which he based them he had in mind when he first expressed his views about the internal environment. Some of the experiments Barcroft and his associates took up in 1930 and 1931, such as the research he and J. J. Izquierdo did on the effect of temperature on the heart beat of the frog and the cat,59 seem to have been new lines of investigation for Barcroft. It is plausible that he might have taken up this work to explore implications of his ideas about the internal environment, but without knowing what he said in his original lecture it is not possible to judge the matter. Barcroft's published research papers between 1924 and 1929 56. Barcroft, Biol. Rev., 7 (1932), 24, 73-78. 57. Kenneth J. Franklin, Joseph Barcroft, 1872-1947 (Oxford: Blackwell, 1953), p. 179. 58. Barcroft, Biol. Rev., 7 (1932), 24. 59. J. Barcroft and J. J. Izquierdo, "The Relation of Temperature to the Pulse Rate of the Frog," J. Physiol., 71 (1931), 145-155; "The Effect of Temperature on the Frequency of Heart and Respiration in the Guinea-pig and Cat," ibid., 364-372.

113

FREDEBIC L. HOLMES

give little clue about the source of his ideas about the milieu His principal experimental interest in those years was the volume changes of the spleen and the role of that organ as a reservoir of blood and hemoglobin. These results he incorporated into a later expansion of his discussion of the internal environment in 1934, but the work on the spleen did not enter into consideration in his article of 1931. Undoubtedly the important treatise that L. J. Henderson published in 1928 linking the internal environment with his studies of the properties of blood, and the paper in which Walter Cannon incorporated Bemard's concept into his discussion of regulatory mechanisms significantly influenced Barcroft.60 A month after Cannon's paper appeared in July 1929, Barcroft arrived in Boston for the Intemational Congress of Physiologists. Between then and October when Barcroft gave his lecture there, the three physiologists must have had opportunities to discuss their mutual interest in the intemal environment.11 Though the writing of Haldane, Henderson, and Cannon may have stimulated Barcroft, his own discussion of the internal environment developed the topic in a distinctive way. It was an extraordinary occurrence that fifty years after Bernard's death four of the most eminent physiologists of their generation should have found in his generalization ideas so pertinent to their own times and work. And it is a measure of the richness of Bernard's thought that each saw in it germinal points for different yet complementary generalizations of their own. Henderson showed how the physicochemical properties of the blood are highly adapted for preserving that constancy of environment essential to the preservation of the very delicate protoplasm of cells. Cannon followed most closely the pattern of Bernard's treatment of 1878, developing in more specific detail and with more complete definition ideas Bernard had stated in a general way. Cannon gave a modernized list of conditions maintained constant, subdividing them formally

interieur.

60. In the introduction to his own discussion of the internal environment Barcroft said: "Within the last twenty years works of first-rate importance by Haldane, Henderson and Cannon have all dealt with the subject." Biol. Rev., 7 (1932), 24. 61. Franklin, Barcroft, p. 175. The possibility that such discussions occurred and may have played an important role in the formulation of Barcroft's ideas about the internal environment is enhanced by Dr. Barron's recollection that Barcroft relied heavily on conversations with his scientific friends for obtaining information. He did not ordinarily read widely in the literature, but listened very carefully in discussions and leamed in this way much of the knowledge which he did not derive from his own experience.

114

Barcroft and the Internal Environment into "material supplies for cellular needs" and "environmental factors affecting cellular activity." He discussed the maintenance of these conditions in terms of three general patterns of regulation: storage, overflow, and altering the rates of continuous processes. These corresponded with the types of regulation Bernard had described, but Cannon explicitly classified them, discussed categories of each, and enumerated many examples derived from discoveries made since Bernard's time. Cannon also defined more systematically than his predecessors the general characteristics of biological regularity mechanisms, designating these processes by his well-known term homeostasis.62 Barcroft made two distinctive contributions to the conception of the internal environment. First, he considered in detail the situations in which lower animals with less efficient regulatory mechanisms find themselves by comparison with higher animals. For each of the conditions of the environment he considered-hydrogen-ion concentration, temperature, oxygen, and blood sugar-he found that lower animals have ways of "evading" within limits the effects of disturbances, whereas higher animals are increasingly able to correct for the modifying factors; the more refined mechanisms are controlled by the higher nervous centers and are consequently closely correlated with the more highly developed brains of these animals. In the case of hydrogen-ion concentration the lower mechanisms are the buffers of the blood, the higher ones regulation by the kidneys and by the rate of breathing. In the case of temperature, lower animals evade by somehow reducing the dependence of the rates of the chemical reactions underlying their physiological processes upon the temperature; higher animals correct by maintaining a constant temperature. In discussing oxygen, Barcroft viewed the dissociation curve of blood as a buffer system resulting in the maintenance of a nearly uniform oxygen pressure in the internal environment despite substantial changes in the quantity of that substance carried in the blood.63 Thus he incorporated into a new framework the results of previous investigations of the interactions of blood and oxygen. With the other conditions the relation between the concept and the research is less clear. Some of the information he used came from the experiments of others, some from current experiments of his own. The latter could have 62. Lawrence J. Henderson, Blood: A Study in General Physiology (Yale University Press, 1928), esp. pp. 20-31; Cannon, Physiol. Rev., 9 (1929), 399-431. 63. Barcroft, Biol. Rev., 7 (1932), 24-87.

115

FREDERIC L. HOLMES

been stimulated in part by his ideas about the internal environment, though there is no indication of that in the research papers themselves. Barcroft's second contribution to the concept of the milieu interieur was to look in a new way at the meaning of the "free life" that Bernard had said the constancy of the internal environment makes possible. "In general," Barcroft said, "the 'end' is more important than the 'means', therefore it seems at least desirable to make some effort towards arriving at a conception of the liberty of life to be attained by the fixity of the milieu interieur." 64 The conclusion he reached was that that fixity is the condition of mental activity. By making this connection Barcroft found an intellectual framework for what had long been a matter of concern to him, less as an experimental subject than as the outcome of a series of personal experiences which befell him and his colleagues during the pursuit of their research. During his first high-altitude expedition in 1910 he found he suffered an "impairment of mental activity" when he had to work at 11,000 feet. He reported that when he "began to lead a more strenuous existence [at that height] symptoms, chiefly cerebral, of extreme apathy, incapacity to think and so forth set in." 65 Recounting such effects in 1914 in The Respiratory Function of the Blood, Barcroft hinted that an understanding of them was one of the ultimate, if distant, goals to which he hoped physiological research would lead: What fields of research lie in front of the physiologist before he can explain how the subtleties of climatic conditions affect the human mind . How gross seem the avenues at present open to such investigations. You are one person in one place, another in another. At the Alta Vista I became as one incapable of arithmatic . . . At Col d' Olen I have heard two clever and distinguished physiologists pause to discuss whether or not four times eight made thirty-two . . . At the Margherita hut I have seen one of the pleasantest and most considerate of companions behave as though he were suffering from alcoholic excess in a mild degree. One day these psychological changes, which in my opinion, appear much earlier than cerebellar ones, such as defective coordination and giddiness, or medullary ones, such as vomiting, will be studied for their own sake. In the meantime we have got the pioneer work to do, we 64. Ibid., p. 24. 65. Barcroft, J. Physiol., 42 (1911),

116

58, 61.

Barcroft and the Internal Environment have got to make a road into this forest wherever we can, we have got to find out the changes which take place in the blood at such altitudes, and in truth this is enough to tax our powers.06 During the expedition to Peru in 1921 Barcroft tried to assess the effects of altitude by comparing the performances of members of the party on mental tests with their performances at sea level. The tests failed to reveal any significant differences, but Barcroft believed they were not demanding enough to expose the subtle changes that occurred. The scientists there agreed that their ability to concentrate for long periods of time without mental fatigue suffered more at altitude than did their ability to solve a single mental problem.87 In the second edition of Respiratory Function of the Blood, in 1925, Barcroft included a chapter on the effects of oxygen deficiency on the mind. To descriptions similar to those he had given before he added accounts of difficulties he and others had encountered during experiments in chambers within which the oxygen pressure had been reduced. Their behavior supported the view that the highest mental faculties disappear first. For example, he had become unable to perform on his own initiative the manipulations required to carry out a given experiment, but could still complete the operations perfectly if told when to make each step.08 Such subjective experiences09 came to have a conceptual focus in Barcroft's interpretation of Bernard's statement about the milieu inte'reur. Besides summarizing again these personal incidents he collected evidence from the literature cited in Walter Cannon's article on homeostasis showing that whenever there is an alteration of the temperature or the hydrogen-ion concentration or a change in the proportions of oxygen, glucose, water, sodium, or calcium in the internal environment, the first symptoms are always mental-such as inertia, unconsciousness, headache, or nervousness-rather than failures of the grosser bodily functions Dogs and other warm-blooded animals tolerated 66. Barcroft, Respiratory Function, p. 249. 67. Barcroft et al., Philos. Trans., [B], 211 (1921-23), 435-437. 68. J. Barcroft, The Respiratory Function of the Blood. Part 1: Lessons from High Altitudes (Cambridge [Eng.] University Press, 1925), pp. 158159. 69. Dr. Barron has pointed out that these experiences were more highly subjective than Barcroft realized. Barcroft adapted unusually poorly to high altitudes. His lungs had a low diffusion coefficient, so that he suffered more from decreased oxygen pressure than did most of his colleagues. He seems to have assumed that the mental difficulties he encountered disturbed the others in a similar way, but it is doubtful if they were as much affected as he imagined them to be or as he himself was.

117

FREDERIC L. HOLMES

wider limits of change in their internal temperatures before losing the ability to function effectively than humans can survive. This difference Barcroft associated with their simpler mental processes. The more nearly perfect control of the internal environment, he concluded, was the prerequisite for the intellectual ascendancy of man.70 We have seen that Barcroft, like several other physiologists, found in the concept of the milieu int&rieur an integrative point of view providing a new perspective on his own previous work and that of others. There is no definite evidence that up until he formulated his own ideas about the internal environment he was guided in his work by knowledge of Bernard's generalization. A critical question now is whether once he had thought out his own position with respect to the milieu inte'rieur it then began to influence his subsequent work. The topic best suited for examining this issue is Barcroft's research on foetal physiology. He began to concentrate on this work in 1932, not long after publishing his article on the fixity of the internal environment; five years later he summarized this research in a series of lectures entitled The Brain and Its Environment. The situation would therefore appear to be highly favorable for such an interaction to have existed. The origin of Barcroft's interest in foetal conditions was in the familiar pattern of response to an unusual result of an experiment conducted for another purpose. He himself said that the research "commenced in a somewhat accidental way." In 1928, while investigating the changes in volume of the "exteriorized" spleens of dogs, he found in one case that the spleen became small and pale about a month after the operation, unlike what ordinarily occurred. A postmortem on the animal showed that it had been in an advanced state of pregnancy.7' Since Barcroft had been showing in other situations that contractions of the spleen ordinarily concurred with demands for increased blood volume, this result raised the question for him of what quantity of blood might be needed in the uterus during pregnancy. In 1932 he made measurements of the amount contained in that organ and found it to be unusually large, even while the embryos were very small.72 He then asked whether this quantity served for storing blood or whether an unusually large volume passed through the 70. J. Barcroft, Biol. Rev., 7 (1932), 80-86. 71. J. Barcroft and J. G. Stevens, "The Effect of Pregnancy and Menstruation on the Size of the Spleen, J. Physiol., 66 (1928), 32. 72. J. Barcroft and Paul Rothschild, "The Volume of Blood in the Uterus During Pregnancy," J. Physiol., 76 (1932), 447-459.

118

Barcroft and the Internal Environment uterus. Therefore, reverting to the approach of his early studies on the metabolism of individual organs, he measured the blood flow and gaseous exchanges. During this study he found that the percentage saturation of the venous blood in the uterus with oxygen gradually declined during pregnancy, until it reached such a low level that it was hard for him to see how any further oxygen could be transferred to the embryonic circulation. At a certain critical point, he thought, "If the two blood streams which leave the uterus are in approximate equilibrium it is clear that intra-uterine conditions have practically reached an impasse. Further growth of the embryos would demand more oxygen, but that oxygen could not be supplied from the available supply except at the expense of a grave decline in the already small pressure at which the gas is supplied to the embryos." 73 The gravity of the problem launched Barcroft on a full investigation of foetal respiration, for which he applied the basic methods he had earlier used in his studies of adult respiration. Part of the solution to this paradox he found through the discovery by his co-workers that foetal blood seemed to contain a special type of hemoglobin, whose dissociation curve differed from that of the maternal blood in such a manner that it could acquire oxygen at a lower pressure than ordinary hemoglobin can. He also found that the arrangement of the foetal circulation allows the blood which circulates to the lower body to be less saturated with oxygen than that which supplies the head.74 In 1935 and 1936 Barcroft and Donald H. Barron investigated the development of foetal movements from the first local reflexes to generalized respiratory-like movements and then to more integrated motions, representing the gradual increase in the level of organization of nervous control. They also found that asphyxia caused a regression to earlier stages, as though the higher brain centers were first affected by lack of oxygen.75 Thus the foetal research unfolded from certain special prob73. J. Barcroft, W. Herkel, and S. Hill, "The Rate of Blood Flow and Gaseous Metabolism of the Uterus During Pregnancy," J. Physiol., 77 (1933), 204. 74. J. Barcroft, "The Croonian Lecture. Foetal Respiration," Proc. Roy. Soc. [Bl, 118 (1935), 248-257. 75. J. Barcroft, D. H. Barron, and W. F. Windle, "Some Observations on Genesis of Somatic Movements in Sheep Embryos," J. Physiol., 87 (1936), 73-78; J. Barcroft and D. H. Barron, "The Genesis of Respiratory Movements in the Foetus of the Sheep," ibid., 88 (1936), 56-61; J. Barcroft and D. H. Barron, "The Development of the 'Righting' Movements in the Foetal Sheep," ibid., 89 (1936), 19P-20P.

119

FREDERIC L. HOLMES

lems to a broad description of respiratory, circulatory, and nervous development. Barcroft's old conception of the blood as a vehicle delivering oxygen according to the needs of the organism and of individual organs seemed adequate to handle the questions with which he was dealing, and he did not refer to the concept of the internal environment in his research papers or in his Croonian lecture of 1935 on foetal respiration.76 Nevertheless, Barcroft had not lost interest in his ideas about the internal environment. He reprinted his original article with some additions in his book Features in the Architecture of Physiological Functions in 1934, and related the concept to other general principles of animal organization. He seemed to imply some special relationship between the milieu inte'rieur and his embryological studies when he gave a pair of lectures in Dublin in 1935. One was on changes in the foetal circulation at birth, the other a condensed version of his discussion of the relation of mental activity to the fixity of internal environment, now entitled, "Chemical Conditions of Mental Development;" 77 but he did not directly connect these two areas of his interest. In 1937 Barcroft summarized his studies of foetal physiology in the Terry Lectures at Yale University. In contrast to his preceding discussions, he here made the milieu interieur a recurrent theme. "The environment of the organism," he said, "was divided by Claude Bernard into the external and the internal milieus. In the case of the foetus, the internal milieu is its own blood; the extemal milieu is in part the blood of its mother and in part the amniotic fluid. So far as chemical factors are concerned, it is the former; so far as physical factors are concerned, it is largely the latter." 78 Several times in his discussion of the quantities of oxygen in the foetal blood he referred to the "oxygen content of the internal milieu," and instead of discussing the oxygen supply to the brain he sometimes discussed the environment of the brain.79 Clearly he had in mind his view that the condition of the internal environment is most critical for the higher faculties of the brain. For example, after he described the finding that the more highly oxygenated the blood of a sheep is at birth, the more quickly 76. J. Barcroft, Proc. Roy. Soc. [B] 118 (1935), 242-263. 77. J. Barcroft, "The Mammal Before and After Birth," Irish J. Med. Sci., July 1935, pp. 290-301; "Chemical Conditions of Mental Development," ibid., pp. 302-313. 78. J. Barcroft, The Brain and Its Environment (Yale University Press, 1938), p. 1. 79. Ibid., pp. 10, 11, 14, 29, 33, 35, 41.

120

Barcroft and the Internal Environment it begins the rhythmical pneumotaxic respiratory movements controlled by higher nervous centers in the brain, he added, "If we regard this type of respiration as essentially pneumotaxic, then we must regard oxygenation of the blood as providing a favorable milieu interieur for the pneumotaxis process." 80 Though Barcroft wove the language of the internal environment into this discussion, it is not immediately evident that the concept itself was essential to his presentation. Removing the phrases involving the term would have left almost all of the information intact, except that the foetal circulation would be viewed, as he had described it in the original research papers, from the older standpoint of oxygen supply and consumption. This does not mean that the substitution was insignificant for Barcroft. Its importance was probably less in the particular description of foetal phenomena than in the relation it established between these phenomena and other phenomena which he viewed in the light of the milieu interieur. Thus, as in previous situations, the concept seemed to play the role of a synthesizing, organizing idea rather than one which aids in the analysis of the specific problem. Considered in its relation to the progress of the research itself, the milieu interieur appears once again to have provided only retrospective illumination on experiments undertaken for other reasons. Yet the situation could not have been that simple, for Barcroft was deeply interested in the concept at the beginning and throughout the period of his research on foetal physiology. It is hard to see his application of the concept to his lecture discussion only as an afterthought. The milieu interieur must have been on his mind at least some of the time while he was doing the research.81 Whether the course of his research would have been any different if he had not made any association between it and the concept of the internal environment is very difficult to judge. The conclusion I have reached may seem entirely negative. I have not been able to establish whether or not the idea of the milieu interieur exerted a tangible guiding influence upon the research even of one who wrote about and contributed to the development of the concept itself. I feel nevertheless that the outcome is of some interest for an understanding of 80. Ibid., p. 42. 81. 1 was especially interested to find out from Dr. Barron whether Barcroft discussed informally in the laboratory while they were carrying out the researches in foetal physiology the relations of those investigations to the concept of the internal environment. Dr. Barron said that he did not recall that Barcroft had ever mentioned the milieu int6rieur or Claude Bernard to him during the years they worked together.

121

FREDERIC L. HOLMES

the relation between general ideas and specialized experimental research. The milieu int6rieur has become a fashionable concept, and it is easy to find statements praising it as one of the most importantgeneralizationsin physiology.The present study suggests that we should be cautious about what we mean by an important biological generalization,and careful to define in what respects it is important. On the other hand, merely because the nature of the influence of the concept in this instance seems to evade precise definition does not mean that we can deny its importance. The dilemma is the general ambiguity inherent in analyzing any case of the impact of intellectual conceptionson the actions of men. This investigation was supported in part by Public Health Service ResearchGrantGM 11808-01 of the National Institutes of Health.

122

Commentary-Part I

ERNST MAYR HarvaTd University SCIENTIFIC EXPLANATION AND CONCEPTUAL FRAMEWORK

Garland Allen's paper on de Vries showed beautifully how important it is for an investigator in biology to have the proper conceptual framework. If he does not have it, his explanations will be vulnerable and most likely completely wrong. In the case of de Vries, I want to reemphasize some of the reasons why his mutation theory was such a complete failure. First of all, de Vries was an essentialist: he believed in discontinuous types and, obviously, if one believes in that, the origin of new types has to be by saltation. This was completely opposed to Darwin's populational thinking, and indeed de Vries himself has always emphasized how opposed he was to Darwin and Darwinism. We now know that de Vries' essentialism was wrong, and Darwin's populational thinking has been restored, largely through the efforts of the systematists. A second weakness was that de Vries, having been brought up in an experimental laboratory, had been completely brainwashed by his teachers to accept that there is only one method in science, the experimental, and this is why he never hesitated to belittle the comparative approach. Now we know that in evolutionary biology the comparative method is as important and as legitimate a method as is the experimental. The third reason for de Vries' failure is that with his background in chemical-physical thinking he believed in the essential identity of basic units. Hence, when he worked on Oenothera he automatically took it for granted that it would be exactly like any other species and that whatever he would find out about Oenothera could then be generalized and applied to all species. It was most unfortunate for de Vries that he picked such an odd creature as Oenothera, with its balanced heterozygous chromosome sets. Incidentally, it was equally unfortunate for the early history of genetics and of evolutionary biology that Johannsen took for his selection experi-

123

COMMENTARY-I

ments an equally odd creature, the bean (Vicia faba), which is an essentially homozygous, self-fertilizing organism, and naturally would be one of the very few among millions of species that would be largely refractory to natural selection. The point I am trying to make is that each of these species has its own highly specialized characteristics, and one has to be sure to pick the right one for one's researches. Oenothera and Vicia demonstrate how dangerous it is to generalize from one system to another, particularly in evolutionary biology. There is one other aspect of biology which is equally valid but which has always embarrassed and annoyed biologists. When it came to the explanation of evolution it turned out that one could arrive at the right interpretation, even if one was wrong in a considerable portion of one's assumptions. Darwin was completely wrong in believing that inheritance was blending; he was all wrong in thinking that use and disuse affected the subsequent fate of the genetic basis of the structures involved; and yet he came up with a theory of evolution which is completely correct in almost every detail. In contrast, the early Mendelians, even though they were the first to understand the basis of inheritance and the nature of totally wrong in their intergenetic change-mutation-were pretation of evolution because their conceptual framework was wrong. Facts, consequently, turned out to be less important than the concepts on which the interpretation is to be based. It is particularly important to stress this point at the present time, because enthusiastic but poorly informed physical scientists have lately tried very hard to squeeze all of biology into the strait jacket of a reductionist physical-chemical explanation. I will discuss later how much of functional biological processes indeed can be explained that way, and Simpson and all evolutionists have emphasized this all along. But Lewontin's explanations revealed how sterile a chemical-physical approach is in evolutionary biology. Here we deal with an entirely separate explanatory framework, and I was delighted that Dr. Shapere admitted this quite freely. I was perhaps even I've become extremely sensitive on more delighted-because this subject-that no one at the first session tried to accuse those biologists of being vitalists who are not satisfied with a purely chemical-physical reductionist framework. One additional point was implicit in much of the discussion yesterday, but nobody really articulated it concretely, and the way I am going to put it now is deliberately provocative. It is quite clear now that all through biology there is a complete

124

Commentary-I dualism. Just so you are not too horrified, let me explain that I am not going back to the obsolete dualism of an Aristotle or a Descartes; what I am talking about here is that there is a dualism which is responsible for the striking difference between organisms and inanimate objects. In my 1961 paper on cause and effect' I pointed out that we really have two biologies-functional and evolutionary-, and as Lewontin pointed out yesterday these two biologies ask entirely different questions about the same phenomena, because they have entirely different interests. These two biologies, and the dualism they engender, exist because every organism is the result of two underlying causes: (1) the blueprint of the genetic program and its genotype; and (2) the decoding of this program interacting with the environment in the making and subsequent life of the phenotype, of the individual. These are two separate sets of phenomena and they lead to two entirely different kinds of biology. Evolution is the differential success of different genotypes. And yet natural selection deals with individuals; that is, with the end-product of the decoding process, with the phenotypes. One can study the biology of the blueprints and one can study the biology of decoding. In methodology and Fragestellung they have remarkably little to do with each other. This separateness of the two biologies has caused great difficulties for at least the last hundred years. I remember discussing with Garland Allen the trouble this made for T. H. Morgan. Morgan was still an embryologist in 1900 when the Mendelian laws were rediscovered; at once he became one of the most vigorous opponents and critics of Mendelism, and he did not change his mind until 1910. Now what was the reason for his early opposition? I think the answer is the following. As an embryologist, Morgan knew that development is epigenetic, and to him Mendelism was an attempt to revive the obsolete and, "as everyone knows," completely wrong preformism. It was not until ten years later that he suddenly realized that the new dualism of biology permitted one to have preformism when it comes to the blueprint, but epigenetics as far as the making of the phenotype is concerned. Thus one can have the best of both worlds. As a result of this, as Lewontin pointed out, we can always give two sets of answers to any biological question: the blueprint answer and the phenotype answer. The same dualism has been implied by those who have said that there are two sets of causes in biology: the ultimate 1. Ernst Mayr, "Cause and Effect in Biology," Science, 1501-1506.

134 (1961),

125

COMMENTARY-I

cause (the evolutionary one) and the proximate cause (the one that deals with the making of the particular phenotypic phenomena). For instance, in my 1961 paper I asked the question, why does a certain individual bird migrate? One can give the blueprint answer and say he migrates because his genotype has been selected to respond to certain constellations of environmental conditions by migration. Or one can give the phenotype answer by saying he migrates because his physiology responds to an increase in some hormonal level, or to a particular set of changes in the environment, consisting of day length, temperature, wind, air pressure, etc. Schaffner asked, for example, "what determines a specific DNA sequence?", and of course he gave the typical answer of one who is interested only in the functional problems of the phenotype. Now the evolutionist would have answered this question, "what determines the specific DNA sequence?" quite differently. He would have said: It is selection among his ancestors together with a certain amount of stochastic noise which determines answer that the specific DNA sequence of an individual-an is just as legitimate as the functional answer which Schaffner gave. Let us look at some of the consequences of this biological dualism. I already mentioned causation. Even though they really know better, physical scientists and philosophers still have a strong urge always to look for the cause of every phenomenon. Nothing is more characteristic of biology than the fact that just about every phenomenon has multiple causation, and this goes way beyond the mentioned dualism. For instance, among the causes responsible for the making of a given genetic program, there are always simultaneously four causations at work: mutation, recombination, selection, and chance. Now let me take another topic-the target of selection. Darwin saw this very clearly and therefore always emphasized that it is the individual which is the target of selection. This was consistently forgotten or ignored in the early period of population genetics. The fact that there is a change in gene frequencies in the gene pool is not the direct effect of selection; it is an indirect by-product of the superior reproductive success of certain individuals. Much of the approach of population genetics during the past forty years, thinking strictly in terms of the fitness of individual genes, and adding up deficiencies in individual fitnesses into a calculated genetic load, was quite meaningless. It ignored the fact that it is not the individual gene that is the target of selection but the in-

126

Commentary-I dividual as a whole. It is only very indirectly, through the reproductive success or failure of the individual, that a given gene in his genotype acquires a fitness value. Uniqueness goes hand in hand with the populational nature of biological phenomena. The enormous potential of the genetic program for variation is indicated, for instance, by the fact that a mammal like man has enough DNA for something like 5 million genes, each consisting of perhaps 1000 base pairs. The amount of potential variability which this permits is so enormous that, except for identical twins, there are no two individuals alike. And even an individual does not remain phenotypically the same in all stages of the life cycle. No two populations are the same, no two species are the same, and no two natural communities are the same. Somebody recently said that there was a new way one could draw the line between physical-chemical phenomena and phenomena involving life. As long as one deals with identical entities and processes, one is in the realm of physics and chemistry, while life and biology start where one deals with uniqueness. In a recent lecture at Harvard Norman Ramsey stressed that the most important consideration in the study of elementary particles is that any meson, or whatever particle one deals with, is always identical with any other, no matter what element it came from, no matter where in the universe you would find it. This is the typical situation for the physical sciences, while in biology we find an extraordinary prevalence of uniqueness. Even when one deals with uniqueness, there are ways to arrive at generalizations, and this is by establishing statistical laws. This, we now know, reduces the sharp dichotomy between the physical sciences and biology because ever since the rise of quantum mechanics, it was found that many of the phenomena of physics are likewise statistical phenomena. As far as biology is concerned, this means that we must think in terms of variable populations rather than in terms of fixed types as de Vries did. The triumph of population thinking over essentialist thinking was perhaps the greatest intellectual revolution that has occurred in evolutionary biology. It is a fundamental change from one way of thinking to another and pervades everything we say and do. I hope that anybody who listened carefully to what Professor Lewontin said and to what I have just discussed has perceived how unimportant a concept reductionism is in biology. Reduction is totaUly irrelevant in nearly all evolutionary problems. Furthennore, a belief in reduction is not needed for

127

COMMENTARY-I

the unification of science because, as Simpson has demonstrated beautifully,2 the unification of biology and the physical sciences can be achieved in a different way. Finally, it has never been demonstrated that reductionism works, so to speak, upward. To be sure, most of the phenomena of functional biology can be dissected into physical-chemical components, but I am not aware of a single biological discovery that was due to the procedure of putting components at the lower level of integration together to achieve novel insight at a higher level of integration. No molecular biologist has ever found it particularly helpful to work with elementary particles. In other words, it is futile to argue whether reductionism is wrong or right. But this one can say, that it is heuristically a very poor approach. Contrary to the claims of its devotees, it rarely leads to new insights at higher levels of integration and is just about the worst conceivable approach to an understanding of complex systems. It is a vacuous method of explanation. ERNEST NAGEL Columbia University

Professor Holmes's paper on Barcroft's adoption of the notion of the internal environment raises the question of what scientists in general and biologists in particular understand by "theory." Certainly the concept of intemal environment is not a theory in the sense in which theories are frequently talked about (particularly in the physical sciences), according to which a theory is a set of well-formed statements from which one can make a large number and variety of deductions. On the contrary, it seems that the notion of internal environment is more like a determiniing factor or "variable" which can be made specific in a number of different ways; and by calling attention to this factor as one which deserves to be investigated, one provides some guidance and a stimulus to further exploration of the area under study. I understand Professor Holmes's paper to be saying that while Barcroft may not perhaps have used the notion of internal environment explicitly in much of his research, nevertheless that conception of a general determining factor of organic behavior was an implicit guiding idea in his work. Accordingly, I take his paper as a reminder to philosophers of science who have been concerned with much more formal views of the nature of scientific theory 2. George Gaylord Simpson, "Biology and the Nature of Science," Science, 139 (1963), 81-88.

128

Commentary-I that they are perhaps neglecting the role that such theoretical variables play in directing the course of actual scientific research. The other more explicitly philosophical papers deal in one way or another with questions about the relation of biology to the other sciences. Thus Professor Shapere's most interesting discussion of biology and the unity of science maintains, if I understood him, that despite much that is common to biology and the physical sciences, biology is in some sense autonomous -although I confess to being not entirely clear as to the respects in which he thinks biology preserves its autonomy. In developing his claim, he introduces a threefold distinction in philosophical questions about biology: those concerned with substantive (or subject-matter) issues; those with epistemological problems that may arise in trying to establish statements in biology; and those with methodological considerations of biological inquiry. And he appears to rest his claim concerning the autonomy of biology largely on the distinctive character of biological subject matter rather than on epistemological and methodological considerations which would differentiate in some way the cognitive status and the nature of biological explanations from those of explanations in other branches of scientific inquiry. However, though there is surely a sound basis for these distinctions, the questions subsumed under them may merge, as will, I think, be evident from considering briefly a point Shapere himself raises. Thus he notes a conspicuous difference between the internal structure of biological theories and the structure of theories in the mathematical physical sciences -a difference he attributes to the fact that the subject matters of biology and physics differ in at least two important respects: the greater difficulty in biology, first, of isolating the factors with which one deals, and, second, of developing limiting or "ideal" concepts. Shapere thinks these difficulties arise primarily not from logical considerations, but from the character of biological subject matter. It is at least debatable, however, whether the possibility of developing theories in biology more like those in physics is not contingent in considerable measure on methodological considerations and commitments. In the first place, it is perfectly clear that there are simplifying or idealizing assumptions in biological inquiry. Shapere himself cites the Hardy-Weinberg Law, which is valid only for biological populations behaving in an ideal way. Accordingly, it is far from being the case that while isolation of factors and idealization of behavior are common in physics, they do not occur in biology.

129

COMMENTARY-I

But, in the second place, the differences Shapere notes between physical and biological theories may be the expressions of different strategies in the formulation of theories in those domains, and may thus require to be explained by reference to so-called sociological factors. It is well known that although physical theories are generally formulated at a high level of abstraction, when the theory is applied to some concrete situation, much that was deliberately left out in the formulation must be reintroduced. For example, laws of motion are typically formulated for the case of bodies moving in a non-resistant medium. This is a limiting case; and when the laws are used in connection with actual bodies, supplementary assumptions must be introduced-assumptions which cannot be deduced from the laws dealing with the ideal case and which are indispensable if the latter are to be adequate to concrete situations. One strategy in constructing theories thus consists in stating some very general postulates that presumably hold in highly idealized cases, with supplementary but specialized assumptions to be added when the general postulates come to be applied to concrete situations. This is the strategy that has been employed in the physical sciences, certainly since the Renaissance. But there is an alternate strategy: the construction of a theory with a minimum of assumptions formulated for limiting cases, so that various special postulates which, on the first strategy, appear as supplementary assumptions, are taken to be characteristic and integral parts of the general theory. It is therefore an interesting question, though one to which I do not know the answer, whether the prima facie difference between biological and physical theories in respect to the degree of idealization employed in them is the outcome, not primarily of differences in the subject matters of biology and physics, as Shapere seems to believe, but of differences in the strategy of theory construction followed in these two areas of inquiry. One must surely not take for granted that when all the factors are listed that have to be taken into account in making predictions in physics, the complexity in the formulation and manipulation of physical theories is significantly smaller than in biological theories. In any case, the strategy of theory construction in physics has been an eminently successful one. Although no one really knows whether or not that strategy would have comparable success in biology, my guess is that if it were adopted by the biological sciences to a greater extent than it has been, it would have salutary consequences.

130

Commentary-I I come now to a question that has been of considerable concern to many of the previous commentators and is the central theme of Professor Schaffner's most instructive paper: the reducibility of biology to chemistry. Despite the general excellence of his contribution, however, he does not make sufficiently explicit just what it is that is to be understood by the successful reduction of one discipline to another; nor does he indicate what is the value of such a reduction. Since Professor Mayr in his comments has expressed some skepticism about current claims that biology is simply a branch of chemistry, a few words are in order concerning the potential value of such a reduction if it were to be effected. I will take my point of departure from what is perhaps the stock illustration of a successful reduction: the reduction of "pure" (or "phenomenological") thermodynamics to the kinetic theory of gases. This reduction is commonly counted as an important achievement in the history of science for the following reasons: The fairly well-established laws of pure thermodynamics were explained in terms of comprehensive assumptions of the kinetic theory of matter. Moreover, some of the previously accepted laws of thermodynamics were shown to need various kinds of corrections in the light of the theory to which they were being reduced. Furthermore, the kinetic theory made it possible to explore and analyze thermal phenomena that were not accessible to investigation in terms of the ideas of pure thermodynamics. I suggest that the reduction of biology to chemistry would be regarded as a valuable scientific achievement and not merely a logical exercise for similar reasons. But it is clearly an empirical question whether the reduction of biology would have such value, though it is a question to which no one has the answer since biology has not yet been reduced to chemistry. For despite the impressive material Schaffner cites, it does not show that biological laws have been reduced to chemical ones-i.e., that the former can be derived from the latter. For example, the important recent discoveries concerning the chemical composition and structure of genes still do not make it possible, as far as I know, to deduce from the assumptions about those structures any of established biological laws about the behavior of genes and chromosomes during cell division. In my view, at any rate, the complete reduction of biology to chemistry would require the completion of two tasks: the detailed chemical composition of biological materials would have to be given, and the laws or regularities of biological

131

COMMENTARY-I

phenomena would have to be derived from the assumptions of chemistry. It is therefore premature to claim that biology has been reduced to chemistry, for at best the reduction has thus far been only partial. It is also premature, it seems to me, to maintain that these partial reductions provide no genuine insights into biological phenomena, for the partial reductions are steps in realizing a comprehensive program of explanation-a program whose outcome cannot be guarantied but which promises to extend fruitfully our knowledge and our understanding. I would like to comment briefly on Bentley Glass's very stimulating paper, whose main conclusion is that the reduction of all biological laws to physical-chemical ones is impossible. In his view, that reduction is impossible because there are statistical laws at various levels of organization in biology which cannot be deduced from the laws that hold on lower levels. But his argument does not seem to me conclusive. For were it sound, it could be used to show not only that statistical laws in biology, but even non-statistical ones, are irreducible to others at lower levels of organization. However, Professor Glass apparently does not deny that nonstatistical biological laws may be deduced from such laws on lower levels of biological organization. And if it is possible to effect a reduction in this case, it is not at all clear why it should in principle be impossible to effect a reduction in the case of statistical laws. It should also be noted in connection with Professor Glass's argument that to reduce some given law to other laws, initial conditions for the application of the latter may have to be supplied. Moreover, the initial conditions may take the form of a random distribution of the values of some parameter. But the fact that initial conditions must be supplied, or that the initial conditions involve the use of statistical notions, does not count against the possibility of deducing a statistical law from other such laws referring to what happens on lower levels of biological organization. For the derivability of a law from a given set of assumptions does not depend on the derivability of the initial conditions needed for deriving that law-the initial conditions are simply part of the data which must be accepted as given in a stated problem, though conceivably the statement of those initial conditions may be deducible from other assumptions (involving a different set of initial conditions) which are the data in another problem. In short, it is indispensable for clarity to distinguish between the laws which are candidates for reduction to other laws,

132

COMMENTARY-I

and the initial conditions for the application of the latter that must be specified if deductions from them are to be made. And it is a non sequitur to conclude that the reduction of a given law to others is impossible because the initial conditions which are part of the premises in the proposed deduction of the given law are not themselves deduced. Let me conclude with an observation on a point that was made by John Platt concerning the open-endedness of physics. Much of the traditional discussion of the reducibility of one discipline to another is based on the tacit assumption that each branch of science has fixed boundaries and a constant content. But the history of science surely shows that this assumption is mistaken so that it makes no sense to ask whether, for example, biology is reducible to physics. What one can ask intelligibly is whether a given theory in a certain discipline (for example, quantum theory) is sufficiently comprehensive to permit the deduction from it of various laws (for example, chemical laws or laws about certain biological phenomena). It seems to me that questions of reducibility ought to be discussed relative to given theories or laws; and the possibility must be recognized that although a given body of laws may not be reducible to a particular theory, it may indeed be reducible to another theory. To say that physics (or some other brand of science) is open-ended is to say in effect, and among other things, that the currently accepted and dominant theory in that discipline may not be, or is not likely to be, the accepted theory indefinitely; and in the light of this assumption, the question whether biology is or is not reducible to physics is most unclear and really quite unmanageable. Although the question continues to be debated, none of the arguments I have encountered which seek to show the impossibility of reducing biology to chemistry is cogent. But it does not follow from this that biology is indeed so reducible. It seems to me, nevertheless, important to point out, as is done in Schaffner's paper, mistaken assumptions and fallacies in reasoning in arguments against the possibility of such reduction. For the claim that biology can be reduced to physics-chemistry is a program of research, requiring both experimental and theoretical investigation, a program whose success cannot be settled a priori and should not be rejected on the strength of inadequate arguments against it. Whether or not the program can be fully successful no one can say in advance. But its pursuit is likely to contribute to the unification of theoretical knowledge, to the satisfaction of having

133

COMMENTARY-I

a more rounded and integrated picture of the world, and to the expansion of our understanding into fresh domains of biological phenomena. REPLY TO COMMENTS BY ERNEST NAGEL DUDLEY SHAPERE

Professor Nagel's comment on my paper is illuminating precisely because it misses something essential and very important about scientific reasoning. For behind his introduction of the notion of "strategies" is an emphasis on a sharp distinction between the substantive content and the methodology of science-a distinction which has been characteristic of the philosophy of science for many years, and which I think is a very dangerous one. The danger is clearly evident in his statement that choice between alternative "strategies" for theory construction may be based on "sociological" grounds, which suggests, at least, that logical or factual considerations do not come into play. Any rationality that there might be in theory construction thus tends to be deemphasized or even ignored. The truth of the matter, it seems to me, is that in science we do what we can do-that is, we adopt whatever strategy seems to be the best or most appropriate to the overall situation. And we adopt that "strategy," and a resulting theory, not (or at least not in the best scientific work) because of sociological influences, but for good reasons, having to do with-as I suggested in my paper-the character of the subject matter, the techniques available, and the demands of the problem at hand. Furthermore, when alternative strategies or theories are available, that theory is taken as correct, or more nearly correct, which gives the more comprehensive and realistic (i.e., less "idealized") treatment of the subject matter. Even if an idealizational strategy or "theory" is used in particular cases for the sake of computational convenience, it is often realized, on scientific grounds, that that "theory" is an idealization, and the "theory" that is taken as fundamentally acceptable (if such a theory is available) is the more realistic one. The tendency of the history of science, despite positivistic universally, rewritings of that history, has generally-almost over the long span-been in the direction of aiming at more and more realistic and comprehensive accounts of nature. Choice of "strategies" (techniques), and of the theories which result from them, is, in the best cases of scientific reasoning on good -the most characteristically scientiftc cases-based reasons, among them the sorts of considerations I outlined

134

Commentary-I in my paper concerning the character of the subject matter involved. EVERETT MENDELSOHN Harvard University

I have thought of the problem under discussion in slightly different terms than the papers have presented it and suggest that a fruitful manner of approach may be through the sociology of scientific knowledge rather than from the perspective of either history or philosophy. The papers and the previous discussants have shown an unconscious concern with the sociology of the knowledge involved. Professor Mayr in his commentary indicated this when he said there are two biologies. I immediately wondered whether this meant that there were two totally different ways of thinking about biology, which in tum really reflected different underlying sciences; or were the two biologies reflected modes of operation, traditions which may or not be separated and may or may not be dealing with the same underlying science? Shapere made this point very nicely in his paper when he talked about the mode of "idealization" so often used by the physicist. The physicist slipped into this approach very easily and made good use of it; I am not sure, however, that he gets trapped by it. But he certainly uses it, stripping a situation of many of the slightly relevant or non-relevant variables, analyzing the components and then attempting to provide an explanation or to construct his explanatory model on this basis. Shapere immediately went on to point out that this was not the manner in which the biologist operated, and I could not really tell if he was asserting that it is something that the biologist could not do, or that it was something the biologist did not do because the tradition, the training, the social modes of which he was a part, did not lead hiln to do that. This difference reflects on a series of other arguments which we had at this session. For example, we can look back to an historical episode almost all of us know well: the difference between Descartes and Harvey as they attempted to describe the motion of the heart and the motion of the blood. Descartes said that for an explanation to be a valid one in biology it had to be deducible from general mechanical principles. He adopted this way of working and proceeded to develop an explanation for the motion of the heart quite at variance with the one proposed by Harvey. Harvey claimed to be operating in the mode of the anatomist; that is to say

135

COMMENTARY-I

he was a particularist. He looked at every single phenomenon and treated it; he did not assume that it was like any other phenomenon. The Descartes-Harvey difference has characterized biology ever since. Therefore I have wondered whether what we observe is not a difference of tradition of operation which grows up within a science and becomes part of the way in which the new generations of scientists are trained. Then, as I look back at the history of biology, I might be led to ask where have major conceptual changes in the biological sciences come from, not in every field but in several important ones? They come very often from what I would call role hybrids-people who crossed from one field of science into another. What they seem to bring with them is the search for different types of clues and indeed a tradition of using different explanatory models and different modes of thought. Their expectation of what might be considered an acceptable explanatory model is different from the expectations of the men whom they joined in their new area of research. Think, for example, of a Giovanni Borelli in the seventeenth century (whether his biology is successful or unsuccessful we can for the moment ignore). WVhathe did was bring to his analysis of the living organism the outlook of the mechanist, not in the broad sense but in the outlook of the Gallilean mechanist, the man who wants to see a simple machine in operation. This established within biology one of a very long series of trends which were used from that point forward. For a second example, look at Lavoisier, who in the eighteenth century brought to his analysis of respiration the insight and all of the tools of chemistry. It was this approach which allowed him to jump over the complexities of the organism and to claim that, after all, respiration is nothing but a slow form of combustion. He did what very few biologists themselves would have been capable of doing at that point; he drew an identity between the physical and the biological process. These processes were still very different, and we know that there has been 150 years of sorting out the differences, but nonetheless Lavoisier generated a kind of insight by carrying the explanatory models of one field and bringing them into another. This notion of the hybrid-the person with training in one field who moves into another-certainly has been with us since; think of a Delbruck and his impact on genetics in the middle of the twentieth century. Another point, which I think is a fundamentally sociological one and yet has enormous impact on what we have been talking about as the philosophy of science, or the problem of

136

Commentary-I explanation in science, can be seen in the very important contribution made by T. S. Kuhn. His work is so important not because he is right in the whole model which he built of how science works, but because he said that in order to understand how scientific ideas are generated and how scientific ideas become part of the working pattern of other men, one must look for the role of new ideas in their emergence and the way in which a scientific community responds to these new ideas and uses them. As I listened to Shapere talking about what it was that scientists did, alluding to their activities in the filling in of the complete picture, he seemed in some ways to suggest that this might well be the biologist's way of acting. This, of course, is not unlike the "normal" science that Kuhn refers to-the attempts to draw all the implications of a paradigm rather than either challenge it or attempt to develop an alternative. Certainly the minute you ask about the sociological implications of this, what it means for the next generation of scientists, it is quite clear that almost all the factors in the social institution that we know of as science favors the construction-the filling in-of the complete picture. The risks are much less, the rewards more obvious, the number of papers published is almost always much greater. Indeed, if you look about you at the scientific community today, most young men are involved in filling in the niches in one or another explanatory scheme. Turning to the other half of the question, I think the role of training came out in Holmes's paper when he very briefly noted that different people were responding in quite different ways to Claude Bernard's concept of the milieu interieur. Why they responded in one way rather than another, and why this strong organizing principle did not become the major operating mode of physiology in the late nineteenth century is one of the fascinating problems of the history of nineteenth-century biology. Holmes pointed to the fact that the tradition which loomed much larger in biology was what I would call the German biochemical or biophysical tradition. The difference between the German biophysical or biochemical tradition of the late nineteenth century and what might be described as the French vivisectionist, or organismic, tradition, is what marks the contrast in physiology during this period. It is not due to either the greater scientific success of one or the other mode of explanation, regardless of how we might attempt to measure scientific success. Rather, it seems to be due much more to the influences of traditions as established within a social milieu, particularly within the educational and training

137

COMMENTARY-I

framework. This factor has been one of the major formative influences of explanation in physiology at almost every period of time. The extent to which available analytical techniques shape the problems that people will work on, and to some measure then determine the mode of explanation which will be adopted, is another point worth examining. Techniques in many fields of science, and certainly in many fields of biology, have been extremely limited. When you cannot measure something, you turn to another problem or you come up with an explanation which would avoid a measurement factor. Certainly in physiology you can see this over and over again. It has to do with the way in which men operate, not the way in which the science itself might on some outside scale be constructed. Let me move on to a second point which Holmes raised in of Barcroft dealing with the problems of his paper-that integration, regulation, and control as a series of biological explanations. Integration, regulation, and control have most often been advanced as explanations by what I would roughly call antimechanists in the biological sciences. These three concepts have most often been adopted by people who were intuitively disposed against finding a physical basis for biological explanation, and ones who were searching for some way to say that the biological organism needed special explanatory models as compared to more generalizable explanatory models. Barcroft is a good example, although among his contemporaries he was probably one of the less philosophically minded men. When you turn to Claude Bermard himself, I think, we have a good example of a man who wanted to avoid being a reductionist; not that he was an anti-physicalist, but he certainly was antireductionist. He clearly wanted to say that there were special levels of organization at which the organism existed, and that consequently there would be special biological laws, special biological rules and explanations-a point which he makes several times over in the course of his own work. When you look at work done during the eighteenth century, you find that almost every time a physiologist tumed to a discussion of the nervous system it was to deny that the organism could be understood on the basis of either physical or chemical activity. The nervous system in a sense became the recourse of the men who wanted to say that there was something special and something biological about living systems. They wanted to express something about a level of organization or of design in which the organism existed which was different from anything else. They may have been right

138

Commentary-I in any specific example, but as a mode of explanation, integrative theory and theories of regulation and control were called upon as attempts to differentiate biology from the other sciences rather than a means of integrating knowledge of other sciences into biology. Let me turn now to what I would call a psycho-sociological approach to reductionism. As I have examined reductionism and the role it has played in science-in biology particularly -I have found that it has served primarily as a program or a policy; it has not served really, for the biologist, as a rule or law. The biologist might like to think of it as such, but more often than not he has used reductionism as a means of advocating a research strategy, of delineating what to him seems to be an acceptable type of explanation for the biological sciences. Claude Bernard very bluntly referred to reductionism as a conceit, and he urged reductionists to give up this conceit of thinking that everything could be understood in terms of simple physics and chemistry. And sometimes they really did think that everything could be thoroughly reduced to particles in motion and central forces. More often than not, though, I think that reductionism served for them as the very type of idealization of the situation that Shapere talked about as being the mode in which the physical scientists were able to work. When you look at the reductionists in biology, what they are really doing is constructing a program of propaganda, a policy which should guide research. This to them is where they think success will be and therefore it should have a strong influence on research strategies adopted. What I am doing, therefore, is separating two tendencies which exist in biology which are not easily separable. One is the tendency which insists that biology is nothing but physics and chemistry; the other is a tendency that says we must adopt a research strategy that is reductionist in approach. I think one can see the difference through periods of historythey often overlap. I would certainly say there is a spectrum along which almost any biologist falls. This spectrum runs from physics and chemistry as being important informants for the construction of any biological models, to physics and chemistry being unimportant. The problem on which biologists are working will often make a big difference as to where they fall on this line. You will find people accepting reductionism as a research strategy and yet believing that there are special vital forces extant in the organism. I think I would interpret Justus von Leibig in this framework. Leibig, as I understand him, really attempted to take almost every organic

139

COMMENTARY-I

phenomenon and process he could and reduce it to its chemical parts. Yet he insisted, as his policy statement, that the organism is not just physics and chemistry, that there are vital organizing principles. Certainly, historically, the impact of the man has been that of a reductionist. Let me close with a few remarks on what I would call problems of the real biology and of the ideal biology. In proposing three factors which provide an explanation of evolution, Lewontin very neatly indicated that all of evolution could be understood in terms of these three factors, none of which was really a causal mechanism. As I understand him, I think that what he was imposing, post hoc, was an order on statements and explanations which had been arrived at in a different form. It may be a good idealization of what evolution is, but it certainly is not a reflection of the process by which explanatory models in evolution were developed. It is important to recognize what it can be reduced to in this simple use of the word "reducing," but I do not think it tells us much about the means by which evolutionary models were developed. If one thing came through Garland Allen's case study, it is that most people interested in evolution in the nineteenth century were searching for much more explicit causal mechanisms of causal explanations. They were searching for entities, for sorting factors, for specifics of causal mechanisms.

JOHN PLATT Mental Health Research Institute University of Michigan

There are a number of aspects of our study and analysis of the complex external world that limit the application of "reductionism," even in principle. These limitations grow out of the following considerations. 1. Different approximations are needed and are generated in different domains of science, with unimportant correction terms in one domain-like gravity in the field of nuclear physics becoming dominant features in another. 2. Each science is open-ended-still developing-at any given tine, so any attempt to make from it a closed derivation of another science is always incomplete, with the chance that this opens up of the derivation being wrong. 3. A science exhaustive in its own domain may still miss completely some other aspect of the problem-as the exhaustive physics and chemistry of a neon sign may still miss the fact that it spells "Joe's Bar and Grill" in English.

140

Commentary-I 4. In systems of hierarchical complexity, the higher levels of organization must be consistent with the lower ones but are not necessarily predictable from them, any more than a "systems phenomenon" like a traffic jam-or the absence of one-is predictable from a complete knowledge of the physics and chemistry of an individual automobile and its driver. 5. There is a difference between explaining a developed subject in terms of a simpler one-for example, my detailed path on an auto trip to Chicago, in terms of the physics of the wheels and the bumps in the road and the guidance system-and being able to predict the former in terms of the latter. 6. Both reductionism and antireductionism are non-disprovable because of the open-ended character of science, which may turn up new confirmations or exceptions at any future time. This means that reductionism, or its opposite, cannot be a scientific theory or proof, but is more like a heuristic program, or an attitude, or a statement of faith. In fact, both attitudes are needed in a healthy science, the one as a pressure to make our views of the world mutually consistent, the other as a stimulus to new insights or new paradigms that can turn upside down the received science of a given time, and permit growth. An examination of these limitations from the point of view of the hierarchical complexities of science throws more light on several of them. Hierarchical organization and systems properties. The notion of hierarchical systems has not been discussed in this conference on a scale of size. For example, let us start at the level of atoms and molecules and macromolecules and go up to cells, organisms, and societies. The molecular level ranges from systems with the dimensions of 1 atom, up to molecules and macromolecules with 103 and 100 atoms. In the case of cells we get up to possibly 1015 atoms, in the case of organisms to something like 1029, and in the case of societies to 1038, giving a quasi-logarithmic scale very roughly but typically like the following (the ranges of sizes in each category are not shown): 1

108

108

1015

atom

molecule

macromolecule

cell

1029

organism

1038

society

The first thing to emphasize about this scale is that the difference between the big systems and the small ones is not just a "similarity transformation." Going from small sizes to big sizes in the case of organized systems requires an addition of information. The situation is accordingly quite

141

COMMENTARY-I

different, say, from the enlargement of a crystal, which represents no more than a trivial increase of information, if any, because of the physical similarity between the large crystal and the small one. In biological systems, if we go from a short segment of a hereditary chain of DNA to a large, long segment of DNA, it appears that there is a roughly linear increase of information (except for some unknown redundancies of segments). And this is recoverable, operational information, because the effects of each individual base pair at each point of the chains can be amplified up to macroscopic and observable effects, such as the presence of a blood disease in an animal or human being. The differences in the levels of information as we go from the largest protein molecules up to single-celled animals, or from these animals up to large organisms, or from organisms up to cultural societies, are factors of thousands or millions or more at each step. This means that we are dealing with hierarchical organizations which develop systems properties that go far beyond the properties of the subsystems. I think it is very doubtful whether they are even in principle predictable from the properties of the subsystems. Take the example of a traffic jam, which is a phenomenon occurring at the level of social interaction. The possibility of a traffic jam is a systems-property of all the cars in a certain region of the country. It is not predictable from the physics and chemistry of the explosion of gasoline in the cylinders or the design of the steering systems of the automobiles. The jam is generated, or is resolved, at the systems level, not at the automobile level. If you have a low density of cars in the county, you do not get the traffic jam, while if you have a high density of cars, you do -regardless of driver behavior or specific accidents of individual cars. You may choose to regard the traffic-jam systems behavior as "reducible" to the subsystems behavior, or you may not, but the jam is certainly describable more easily in terms of the higher-level systems-properties rather than in terms of the multiple-detailed individual complexities of the subsystem interactions, most of which are irrelevant to the final general result. Certainly the systems-properties at the higher level of the hierarchy are not easily predictable from the properties of the sub-systems at the lower level. It is true that a sufficiently detailed and insightful application of computers over a long period of time might permit the extraction of general properties, like traffic jams, invariant to the detailed subsystem interactions. But such a procedure begs the question of

142

Commentary-I how we decide which of an infinity of derivable general properties the computer is to search for. And, in addition, there are many, many hierarchical systems where it is simply not practical in a finite lifetime, with a finite computer capacity, to jump across several levels-S or 10 or 15 orders of magnitude-of a hierarchical organization in a predictive way. In such cases, reductionism might be abstractly tenable, but the daily operations of science in real time have to proceed by a different route, by studying each level of the hierarchy in terms of its own regularities of organization and not in terms of those at some other level. The different aspects of a system. The difficulty in making predictions from one level of organization to another is a special case of a more general point-that a complete and exhaustive description of one aspect of a system may in fact be carried out without any mention of some other aspect of the system. Thus, in the case of the neon sign, as Donald Mackay has emphasized, one might know all about the physics and engineering, the electrical behavior of the individual atoms and ions in the gas discharge tube, the construction and properties of the electrodes and the glass tube, and the electrical connections to the transformers and the city generators-and still not know that the sign says "Joe's Bar and Grill" in English. The message on the sign is a linguistic and social and historical aspect which has no place in the time-invariant physics of the problem. What has "meaning" at the social level of interaction is generally quite "meaningless" at the different level of examination of the detailed motion of the atoms and molecules. Different approximations in different domains. The hierarchical scale also brings us up against the general fact that the laws of science valid in any domain are only approximations with a certain range of validity, and that they change in going to a different range of size or domain of inquiry. For example, in dealing with the hydrogen atom, the most accurate equation at present is the Dirac equation. Even this elegant equation is not correct; for 20 years we have known that it must be patched up with an important little Band-Aid called the polarization of the vacuum (and who knows what smaller Band-Aids must eventually be added on top of this one ?) Even so, no one can solve the Dirac equation for larger atoms than the hydrogen atom, or at least no one has succeeded in doing so. The result is that we approximate larger systems by using the simpler Schrodinger equation. But no one can solve the Schrodinger equation with its relativistic components for more than a three-particle system, and so we go on to the still more approximate Hartree-Fock

143

COMMENTARY-I

equation. But the Hartree-Fock equation in its turn becomes so complex that one can hardly extend it to even as heavy an atom as copper. To make an accurate a priori calculation of the differences between the properties of copper and nickel-which only differ by one electron, but whose differences are obvious to any man with coins in his handl-would be almost unthinkable as a practical computer-program today if we were to base it on the fundamental Dirac or Schrodinger equations including relativistic corrections. This example, entirely within the supposedly fundamental domain of atomic physics itself, is only a special case of the general rule that approximations that are valid at one level are no longer valid at another. In the little puddle in the street, the ripples are dominated by surface tension, while in the Pacific Ocean the waves are dominated by gravity. As you go from the one scale to the other, the surface tension becomes relatively insignificant. In the hydrogen atom the relativity correction to the energy is about 10-5 of the total energy, but the correction increases as the fourth power of the atomic number, so that this meaningless, trivial, negligible correction in computing chemical reaction or binding energy for the little hydrogen atom becomes a correction of hundreds of electron-volts in the iron atom. At the present time, no physicist or chemist has actually been able to demonstrate that these hundreds of volts do not have modifications that affect binding, in iron or the heavier atoms. How then do you expect to predict, by quantum mechanics, the behavior of the iron-porphyrin molecules, or the iron cytochromes, or hemoglobin? If we still, after forty years of quantum mechanics, cannot cross reliably this simple gap of a little more than one order of magnitude in electron number, when would the reductionist expect this gap, or any larger gap, to be closed? As long as chemical binding cannot be computed from quantum mechanics with certainty, it evidently has to be measured first by experimental chemists, who can then tell the quantum theorist, with nonreductionist direct evidence, whether his calculations have put in enough corrections accurately or not. Perhaps the most dramatic example of this kind that we know occurs in the case of gravity. The physicists who look at atoms and molecules would probably never have predicted the existence of gravity if they had always been limited in their observations to the atomic world. The reason is that gravity is only about 10-39 times the size of the strong electrical interactions in atoms. At the present time, even with lasers, we are only beginning to reach accuracies of 10-12 to 10-15 in our most exact measurements. It would be hopeless to measure or to check any theory of

144

Commentary-I correction terms at the 10-39 level, unless fantastic cancellation occurred in some phenomenon, or unless these terms became important in some other range where they could be studied independently rather than derivatively. How then can gravity be so important to us large bodies when it is so trivial at the atomic level? It is because there is a fantastic cancellation: the interactions that were absolutely dominant down at the atomic level all cancel out when we have 1030 or 1040 atoms, leaving behind that absolutely negligible residue of gravity which did not cancel out and which thus becomes the most important phenomenon at the planetary level. The possibility of predicting or extrapolating these trivial corrections correctly from one domain to another is something which has yet to be demonstrated. Open-endedness of each science. This limited domain of approximations means that every science is subject to additional unforeseen corrections as it extends its domain, and is therefore inherently open-ended. At any given instant in physics and chemistry-if these are ongoing subjects in which our understanding of the trivial corrections is not yet complete-any attempt to reduce some higher-order subject to these supposedly more basic subjects is likely to be upset by some amplification of the most negligible of corrections not yet observed or computed. We see that any thoroughgoing reductionism therefore amounts to an assertion that the more basic subject is now known to be completely closed down to its last corrections, a thing that no scientists and few philosophers would ever want to admit for any field. A science which is continually developing, in which there are still some things left for our children to discover, must be perpetually incomplete, or at least of some degree of uncertainty as to its completeness. This is the most fundamental obstacle, as I see it, to any real-time assertion of reductionism. A thoroughgoing reductionism which, by the nature of science itself, can never be explicitly demonstrated at any given epoch, thus becomes a problem in infinite regress, a will-o'-the-wisp that should be as unsatisfactory to operationally minded philosophers as it is useless to science itself. It seems to me, therefore, that reductionism is not something that can be proved, but that it is rather, as Mendelsohn has emphasized, a program-or perhaps something more like an attitude, or an article of scientific faith. It is undoubtedly a magnificent program, always pressing toward the unity of science; and it is a program full of continually fruitful suggestions as it looks for regularities across the borders, and one to which all scientists must in large degree subscribe. But

145

COMMENTARY-I

it is a program that will never be completed, at least not until we have believed for perhaps 40 or 100 or 2000 years that physics and chemistry are complete, with nothing more to be said about them even at the 10-39 level, and until we believe that our computers are able to compute all the extrapolations and systems-properties to other domains and other levels of the hierarchy with sufficient accuracy to make experiments in other fields unnecessary. Until that time, every field must continue to do its own experimentation and discover its own approximations, and continue to create and manipulate its ovn symbols and regularities, and then simply raise the question whether these are consistent with the lower-level approximations that are known at the time. This leads to even longer thoughts. Who can say, for example, that there is not some further correction at the 10-78 level which only becomes significant and measurable at galactic distances? We see that ultimately there may be no "lower level" of more basic science. Each domain must inform all the others, the complex the simple, and the large the small. This is not reductionism, but interactionism. It will give us more perspective if we remind ourselves that in every era some great discoveries have been essentially unanticipated from the physics and chemistry of the time. Remember that it was the biologists who discovered current electricity. The physicists might not have discovered it for one hundred years, for they were looking at stars. It was biologists who discovered that bats avoid obstacles in the dark; the more sober and fundamental scientists said this was wishful thinking or vitalism, or some such dirty word, and then later found the echo-ranging mechanism. It was biologists and geologists who said that geological-times must have extended backwards for millions of years, and Kelvin said, "That's absurd; why don't you people be scientific? We can prove that the sun cannot have lasted that long." And it couldn't, according to his mechanism of solar energy generation. But it wasn't the sun's lifetime that was limited, it was Kelvin's imagination. I think we must insist on the right and necessity of each of our fields of science to organize its own observations, and then and only then to see how far they can be interpreted in terms of lower levels of organization. A field must not reject its own real regularities if they do not fit those from other supposedly more fundamental fields. Such situations should be carefully scrutinized for error, of course; but if the new regularities are indeed based on well-made observations and sound statistics,

146

Commentary-I they may be simply the first signs of a new scientific revolution, as happened so often in the past. Non-disprovability of reductionism or anti-reductionism. But if reductionism is not provable-or disprovable-in real time, then by the same token antireductionism is not provable or disprovable. Are not these different absolutist claims the result of different personal attitudes that often come simply from differences of temperament? It seems to me that both kinds of attitude or temperament are needed in a healthy science. At a given moment of history it is important for some men to say, "Oh, we can surely explain these regularities in terms of simpler subsystems." But it is also important for other men to say, "Your simple subsystem approach is sterile; its ideas are incomplete and its predictions dubious when so extended; we cannot know, without checking, how far they will extend into other domains; we must keep on the lookout for new principles coming to the fore." How can we continue to have new approaches in complex fields unless we have some men who believe that the phenomena are not adequately covered by supposedly more basic explanations? This dialogue must continue as long as science continues. If men should ever accept a conclusive victory by one side or the other, the dialogue would stop, and science would be dead. It seems to me, therefore, that the reductionists and the antireductionists, like the deductive and the inductive types of mind, both have a permanent role to play. Science is a balance of opposites in the search for new regularities. The believers in regularity must not discourage the searchers for the new, while the searchers for the new in their turn must not disparage the on-going search for further reduction, if science is to continue to be a living venture.

147

Functionand Teleology MORTON BECKNER Department of Philosophy, Pomona College Claremont, California

I. INTRODUCTION In trying to decide whether teleology in the sciences is good, bad, or indifferent, philosophershave tended to focus on three sorts of cases. They are exemplifiedin these paradigms: 1. "Thefunction of the heart is to pump blood." (Call this a functional ascription.)

2. "Thegoal of the rat is to reach food at the end of the maze." (Goal-ascription.)

3. "Jonesintends to retire early by workinghard."(Intentionascription.)

These sorts of cases are sometimes confounded,but they should be clearly distinguished. In a rough-and-readyfashion, some important differences and connections can be described in the following way. Function

Functional ascriptions describe the role played by a part or process in the activities of a larger or more inclusive system. Standard examples are the ascriptions of roles to the organs, tissues, cellular parts, biochemical processes, and so on, in the growth,regulation,maintenance, and reproductionof organisms. Functions are also ascribed to the parts of artifacts, especially such objects as machines, pieces of furniture, and so forth. In these cases we antecedently identify a system S and activity q such that the whole of S can be said to do 0; and functions are then assigned to the parts P of S or to the activities cl' of P, only if 4' or P do contributeto the ping of S. In general, function is always function in a whole system. Goals and Intentions

We ascribe goals to persons whenever we ascribe intentions; indeed, anythingdescribableas an intention is also describableas a goal. But not all goals are intentions-or so people commonly argue. For example, we might acknowledge (with Sartre) that 151

MORTON BECKNER

a person's goal is to become God, but he has no intention of becoming God. Another sort of case: we may say that the rat's goal is food, but that it has no intentions. Finally, we might say that a self-regulated system, such as a target-tracking missile, has a goal but no intentions. These examples may seem dubious. I think that the concepts of "goal" and "intention," as we have them, always leave room for doubt; there are no completely convincing arguments that establish the possibility of goals without intentions. One can maintain that rats and missiles have no intentions, but that they also have no goals; or that rats and persons indeed have goals, but that they also have the corresponding intentions. Many philosophers have bypassed this point by introducing a technical concept of "goals" which insures their existence in such cases as the rat and missile. The application of the technical concept is defined so as to leave open the question of intention. Its introduction is guided, however, by examination of selected paradigms, namely, cases which (1) involve intention, and also (2) exhibit certain "behavioral marks" of directedness. The technical concepts "goal," "goal-directed," "goal-seeking," "directive correlation" (Sommerhoff), and "directive organization" (Nagel) are yielded, in a fairly obvious fashion, by considering the behavioral marks and/or some very general features of the organization of a system that would exhibit them. An instance of such a paradigm would be a man trying to reach a destination in the face of a series of obstacles. Writers differ in the details of their accounts of the behavioral marks, but they have in common some reference to persistence and to the range of variation in the disposition of obstacles under which the goal still tends to be reached. Again, there are differences in detail in the accounts of the organization of goal-directed systems, but the acceptable ones have in common (1) some reference to the power of the system to compensate for environmental changes that might impede the system's progress toward the goal, and (2) some reference to the independence of the variables that define the system and its environment. Sommerhoff, I think, first saw the necessity of the latter reference for ruling out various unwanted cases. If conditions (1) and (2) are satisfied, the system is selfregulated by means of feedback. One aim of introducing such a concept is to provide for cases of goal-seeking that are not goal-intended. The analysis does apply to the rat in a maze, the self-guided missile, and to a large set of other biological and technological cases. Goals and Functions In typical biological cases, achievement of a goal (in the above sense) does have a function in the system. Suppose, for example, 152

Function and Teleology that the movement of the water flea, Daphnea, toward the surface is goal-directed; this movement also serves the function of respiration in Daphnea. In a general way we may say that any biological machinery capable of goal-directed activity is (or at one time was) also capable of performing some function or other. But it is easy to imagine cases of functionless goaldirection. We could, for example, build an ingenious mechanism, regulated by feedback, that pumped sea water out of and back into the sea at a constant rate. The activity of the machine would be goal-directed; but achievement of the goal serves no function. (Of course, the parts of the machine would serve the function of pumping sea water.) I think that many cases of human goal-seeking fall in this category. Moreover, there are processes that do serve functions without being goal-directed. The blink reflex is an example. Thus, there is a clear distinction between activities that serve functions, and activities that are, in the technical sense, goal-directed. To challenge this point would, I think, be quixotic. To insist that we ought to say the function of any activity that has a function is also its goal would amount to no more than a rejection of the technical terminology. There is a sense, hard to get at precisely, in which functions must be fulfilled, whereas goals need not be reached. Suppose that people are constantly trying to reach the moon by climbing ladders. It is no objection to saying that reaching the moon is their goal to point out that they will never make it that way. But it is a conclusive objection to the statement "The function of the brain is to cool the blood" to point out that the brain does not cool the blood. This point is complicated, however, by the following two considerations. 1. In the case of ostensibly nonintentional biological activity, we would not identify something as the goal, for example, of an organism, unless organisms of that sort sometimes achieved it. 2. We sometimes say that something has a function, but is not performing it (or is performing it poorly). For example, if my heart stops pumping blood it does not thereby lose its function. So it is not strictly true that if qbis a function of P, then P contributes to q. The following, however, is true: if the members of the class of P's never 0, then 0 is not a function of P. Thus we readily apply the vocabulary of success and failure to both goal-directed activities and to functions. Functions, Goals, and Intentions All three concepts share the following feature: we may say of intentional, goal-directed, and functional activities that these all take place "for the sake of" something and "in order to" do something. 153

MORTON BECKNER

Consider the following statements: 1) The heart beats in order to pump blood (for the sake of pumping blood). 2) The rat snuffles along the maze in order to get food. 3) Jones works hard in orderto retire early. I think (not everyonewould agree) that all these remarks are entailed by the correspondingfunctional, goal-, and intention-ascriptions,and are therefore on occasion true. Not only might 1) and 2) be true; so far from presupposinga theology or discreditedmetaphysic, they do have important scientific uses. Clarificationof these uses is one aim of this paper. II. FUNCTIONALASCRIPTIONS The task of this section is to give an account of the meaning of functional ascriptions; I shall concentrate on sentences of the form "A function of P (or of q/) is o," although functions may

obviouslybe ascribedin a variety of alternativevocabularies. I divide functional ascriptionsinto three classes; my working examples, one from each class, are the following: 1. (Call it FA1): "Afunction of the heart is to pump blood." 2. (FA2): "Afunction of the heart is to produceheart sounds." 3. (FA8): "A function of the earth is to intercept passing meteorites." FA1 is clearly true. The heart has many functions that we know of (others are distributingoxygen, removing wastes, etc.), and, no doubt, some we know nothing about; but pumping blood is certainly one of them. FA2is in all probabilityfalse. The heart does produce heartsounds; the production of heart sounds is even a necessary condition of the heart'spumping blood. But this is not one of its functions. In the case of FA3,something has gone wrong. The earth does interceptpassing meteorites, but very few people, if any, would be willing to say that this is one of its functions. Is, then, FA3 false? Some would say so, on the grounds that meteorite intercep-

tion is not a function of the earth. Others would say, not that FA3 is false, but that it is somehow inappropriate-that it is pointless, or nonsensical, or involves a category error. I do not think much turns on this difference.If FA3is regardedas false, we do have to distinguish two ways in which functional ascriptions may be false, that is, between FA2and FA8.I shall call the third sort of case "inappropriate"; the readermay please himself as to whether inappropriatefunctional ascriptionsare also false. In the analysis of statementsof the form "Afunction of P is S," we may distinguish two questions: 1) If we suppose that a functional ascription is appropriate,what is the relation between P 154

Function and Teleology and q? 2) What distinguishes those which are appropriate from those which are not? First consider question 1). One answer is that P, or an activity +' of P, is a necessary (or perhaps both necessary and sufficient) condition of S. For example, it might be held that "a function of the heart-beat is to pump blood" states that "the heart-beat is a necessary condition of blood-pumping." This answer can be construed as no more than a first approximation, in view of two sorts of difficulties. First, if the heart-beat is necessary for blood pumping, then so are heart-sounds, since the heart-beat is a sufficient condition (neglecting some inessential points) of heart-sounds. Thus, no heart-sounds, no bloodpumping. But heart-sounds do not have the function of pumping blood; whereas we would be committed to this on the hypothesis that the ascription is otherwise appropriate. Moreover, since the heart-beat is sufficient for heart-sounds, we are committed to the conclusion that a function of the heart-beat is to produce heartsounds. This is false. Clearly the difficulty here lies in the fact that heart-sounds are an accidental by-product of the heart-beat, and are no part of the cause of blood-pumping. We should not say that P and q/ have q as their function unless they were, in some sense, part of the cause of 0. The difficulty here is closely connected with the famous difficulties associated with attempts to analyze the causal relation in terms of the relations of necessary and sufficient condition. The second sort of problem is this: a function of the heart is blood-pumping, but the heart is not really necessary for blood pumping. At least, not if by "heart" one means the muscular chambered organ usually meant by the term. For example, the blood can be pumped by a machine. This might seem a rather trivial objection. It might be suggested (as Nagel does) that since a function of P is always a function in a system S, the relevant class of S's in which a P has the function be restricted so as to rule out such cases as blood-pumping machines. Perhaps this could be done, but I do not believe a general method of ruling them out has been described successfully. And if it were to work for the heart example, there would still be a problem. Organisms commonly have alternative means of performing the same function. My right kidney excretes urea, but if it is damaged the left one does the job; so my right kidney is not necessary for urea excretion, although that is one of its functions. Sweating aids in temperature regulation; but if I lose the ability to sweat I can make do by panting and suitable choice of behavior. And there are the many cases in which we say that P or p' has the function + when P or 0' does no more than increase the probability of + under certain, perhaps rare, circumstances.

155

MORTON BECKNER

Nevertheless, there is a grain of truth in the "necessary condition" view. The grain is extracted in the following formulation, which also escapes the second type of difficulties. (See below for the way out of the first type of difficulties.) I state it only for activities of parts of S; with obvious minor changes, it holds also for the parts of S. Granted that the functional ascription is appropriate, then "A function of q/ in S is 0" is true if and only if there are regularly occurring states of S and its environment in which (' occurrs and in which the occurrence of .' causes an increase in the probability of the occurrence of qb.Under these circumstances we may say that q' "contributes to" the performance of +. We shall now turn to the question of what makes FA3 inappropriate, whereas FA2 is appropriate but merely false. Philosophers who have addressed this question have all, I think, thought that the answer must lie in some difference between, for example, vertebrate bodies and solar systems. All functional ascriptions, in this view, presuppose (or implicitly assert) that the system in question has certain properties that merely physical, chemical, and other systems lack. Nagel, for example, suggests that functional ascription "presupposes" that "the system under consideration ... is directively organized." (Nagel defines "directive organization" carefully; I have indicated in Part I no more than we need for my subsequent arguments.) This suggestion works for FA2 and FA3: organisms with hearts are certainly directively organized, whereas the solar system is not. Hence the necessary presupposition is missing in the case of FA3. But there are difficulties. First, we do commonly ascribe functions to the parts of machines and other artifacts. These systems are not, or need not be, directively organized. Second, the functional relationship, as either I or Nagel define it, can be present in a directively organized system, but still have nothing to do with the system qua directively organized. FA2 provides an example. Third, some systems are directively organized, but we do not apply functional analysis to them. An example is the ecosystem of a mountain lake. This system is directively organized with respect to the biomass ratio of predator and prey fishes. But we would not say that a function of the trout is to eat the bluegills, although this does play a role in the regulation of the ratio. It might seem possible to avoid these difficulties by holding not merely that functional ascriptions presuppose directive organization, but that they presuppose that the ascribed function contributes to directive activity, and that, moreover, the directive activity has adaptive significance. Another possibility would be

156

Function and Teleology to delete reference to directive activity altogether, and define "appropriateness" in terms of adaptive significance. In this view, functional analysis of nonliving systems would be treated as a somehow parasitic or derivative procedure. Nagel's position, and the variations on it just suggested, do provide clarification. Nevertheless, I think they narrowly miss the mark. These positions seek the distinction between FA2 and FA3 in some feature of the systems, some difference between e.g., animals and solar systems. My view is that the distinction lies in logical differences between the conceptual schemes we are prepared to apply to animals and to solar systems. Functional ascriptions presuppose conceptual schemes of a certain logical character. The ascription is inappropriate if such a scheme is missing. In order to describe these conceptual schemes I shall first define two notions: 1) the "net-like organization" of a system, and 2) a "contributory system." 1) If we consider a function such as respiration, we find that the function is performed by a complicated set of parts that causally influence each other in complicated ways. The parts are all parts of the organism; except for the indivisible parts (if there are any) each part has parts; each part (except the whole organism) is part of another part; and finally, given any part P, there are other parts of which P1 is not a part. (These are rather trivial remarks about the concept of a "part.") Every part can in principle, and in practice does, have causal influences on every other part. Of course, only some parts, and only some of the causal relations between them, need be considered in an analysis of respiration. The part-whole relations of a system can be represented by a "tree"diagram, such as that shown in Fig. 1. In the diagram, a and b represent parts of c; and a, b, c, and d are parts of e. Such diagrams would represent perfect part-whole hierarchies if we were to define "part-levels" in such a way that every part at a level was the same kind of part, and if every part at a level were exhausively analyzable into parts of the same kind. The systems we analyze functionally, however, are not perfect hierarchies, if we define the parts in the way we ordinarily define them. Suppose now we identify some activity 4 performed by the system marked "e"in Fig. 1; and a set of causal relations between some of the parts of e which contribute to ,. These can be represented by the arrows in Fig. 1. What I have in mind by "net-like organization" could be represented by "tree"diagrams, except for the fact that a part P1 can contribute to the activities of part P2 when neither P1 or P2 is

157

MORTON BECKNER

Fig. 1

part of the other (for example, the diaphragm influences the lungs). I representthese causal influences by wavy lines. Now suppose that a theory T leads us to identify parts of S, and causal relations between the parts of S, in such a way that, as a matter of fact, the part-wholeand causal relations in S can be representedin a net-like diagram such as Fig. 2. Then I shall say that S possesses "net-likeorganizationwith respect to T."

a/ Fig. 2

Any relatively complex material system obviouslywill possess net-like organization with respect to some existing theory or other. So net-like organizationis possessed by systems, such as the solar system, which we refuse to subject to functional analysis.

(2) If the activities 4l in a system S1 "contribute"(in the sense defined above) to the activities 02 of another system S2, I shall call it a "contributorysystem,"and we may speak of ql as a S1 need not be a part of ST The arrows i "contribution"to Figs. 1 and 2, on both straight and wavy lines, represent the "contribution"relation. Any straight line with an arrow point on it is a contributorysystem, and so is any part with a wavy line leading from it. Again, any complex system will contain 158

Function and Teleology contributory subsystems, so contributory systems are present in, for example, the solar system. Both parts of a system with net-like organization and the contributions they make are identified with the help of a conceptual scheme. When we refer, for example, to the lungs, we do so by means of a term ("lungs") that has a certain definition, or at least a certain standard use. The definition or use places the term within our conceptual scheme. A part (or activity), such as a lung (or respiration), is a part or activity relative to the concepts we employ in identifying and describing them. Now suppose we have identified a system S; that S exhibits activity 0; and that we are interested in showing how S does q by ferreting out the contributions that the parts P1, P2. . Pn (and their corresponding activities 4' 02. . .pn) contribute to +. I want to call attention to two alternative ways of identifying the parts of S. Let "P" refer, by virtue of its definition, to P. Granting that "P." makes a contribution to p, "P" could be defined (in part) by reference to p);or "P1"could be defined quite independently of any reference to p. For example, the heart pumps blood (P, contributes to p); and the term "heart," by definition, cannot be applied to any system unless it pumps blood, or is the sort of system that contributes to the pumping of blood (the term "P." is defined (in part) by reference to k). On the other hand, the earth does intercept passing meteorites; but the term "the earth" is not defined (even in part) by reference to the activity of meteorite interception. In sum: it is a logical truth that hearts pump blood; it is not a logical truth, but only an empirical one, that the earth intercepts meteorites. Let us refer to the parts of systems which, like the heart, by definition contribute to an activity p of a larger system as "definitionally contributory" parts. Moreover, the activites O', of parts of a system, or of the whole system, may simply, as a matter of fact, contribute to an activity 0 of a system; or, alternatively, the term "4/i" can be so defined that o', necessarily contributes to p. The revolution of the earth about the sun contributes to the interception of meteorites, but this fact is not exploited in the definition of "revolution." On the other hand, a) the courtship rituals of a bird do contribute to the selection of mates, and this contribution is part of the concept "courtship ritual"; and b) breathing by definition contributes to respiration. Call such activities "definitional contributions." They are, I think, relatively much rarer in biological theory than definitionally contributory systems. We are now in a position to state rather briefly the major theses of Part IH.First, functional ascriptions explicitly state that a part P, or activity p',contributes to an activity pin or of system 159

MORTON BECKNER

S. Second, functional ascription presupposes that S possesses a net-like organization such that (a) the strands of the net are identified by the same general conceptual scheme which is employed in the ascription itself; (b) a significant number of critical strands in the net definitionally contribute to one or more activities p of the whole system S; and (c) the contribution ascribed is a contribution to activity p of the whole system S, where ( is an activity to which a significant number of strands in the net definitionally contribute. Let me now summarize by applying the foregoing remarks to FA1, FA2, and FA3-omitting as far as possible my own jargon. FA1: It is true that a function of the heart is to pump blood. The heart does pump blood; the body is a complex system of parts that by definition aid in certain activities of the whole body, such as locomotion, self-maintenance, copulation; the concepts "heart" and "blood" are recognizably components of the scheme we employ in describing this complex system; and blood-pumping does contribute to activities of the whole organism to which many of its organs, tissues and other parts definitionally contribute. FA2: The heart does produce heart-sounds; but heart-sounds do not contribute to any of the activities of the whole body which our conceptual scheme singles out as privileged benefactors of the activities of our bodily parts. FA3: Although the earth does intercept meteorites, we do not, I think, identify the parts of the solar system, or any of its activities, in terms of the contribution they make to activities of the whole solar system. The solar system certainly does things as a whole (for example, it revolves about the galactic center), but we have not found it useful to construct a theory which involves the definition of celestial bodies in terms of their contribution to the motions of the solar system. Note that this analysis applies to machines and other artifacts; there is no need to regard the ascription of functions to them as in any way derivative or parasitic. III. ELIMINABILITY It has been suggested that teleological statements such as "The heart beats in order to pump blood," "The function of the heart is to pump blood," "The heart beats for the sake of pumping blood," and so on can be eliminated in favor of nonteleological statements. This is a very complicated question. In order to present a supportable answer, we have to pay some attention to the following questions: 1) How are we to distinguish in general between teleological and nonteleological statements?; 2) What 160

Function and Teleology are we to understand by the "elimination" of statement A in favor of statement B? Supporters of the elimination thesis usually speak of the "translation" of teleological statement A into nonteleological statement B; but there are other sorts of relation not as stringent as translatability that might reasonably fill the bill. I will deal with these two questions together, and in doing so, explain my own view, which is: (a) teleological statements are not translatable into nonteleological ones, but (b) every activity of a system, even those of systems which we ordinarily describe with the help of functional ascriptions, can be described in nonteleological language. Clause (a) of my opinion is difficult to prove, in view of the lack of clarity both in our criteria for identifying teleological language and in our criteria for deciding whether B translates A. There would, no doubt, be general agreement that sentences containing the expression "for the sake of" and "in order that" are teleological; and similarly for those sentences I have called functional, goal-, and intention-ascriptions. But consider the following sentences, which do not fit one of these molds: (1) "Vultures break open eggs with stones." (2) "Myrtle warblers migrate in the spring into regions of abundant food." (3) "The missile swerved toward its target." (4) "He acted out of avarice." (5) "The arms of the Dean's chair are upholstered." (6) "The sight of a barracuda releases an escape reaction in anchovies." In all of these cases there are strong lines of argument tending to show that they are teleological, if explicit functional, goal-, and intention-ascriptions are teleological. These arguments have a common pattern; they indicate that any differences between the phrasing of cases 1-6 and the phrasing of admittedly teleological sentences are matters of prose style only. For example, (1) is very much like "Vultures use stones to break open eggs," which is very much like "Vultures use stones in order to break open eggs." I think these three sentences do indeed differ only stylistically, and that if one is teleological, all three are. Similarly, in case (2), the movement of birds is not a case of "migration" unless it is movement of a regularly recurrent sort that does indeed serve a mating or food-getting function. In case (3) an object is not the "target" of a self-guided missile unless the missile is directed toward it as a goal. In case (4), the reference to "avarice" indicates (among other things) that the action is done for the sake of some goal. In case (5), the "arms" of a chair are by definition parts that serve a certain function. Finally, 161

MORTON BECKNER

in case (6), an "escape reaction" is an activity whose function is the avoidance of danger. I would suggest that there is no touchstone for identifying the teleological, and that teleological language is far more prevalent and more firmly embedded in the language of science than a casual inspection might suggest. My conviction that teleological language cannot be translated into nonteleological language is based on the following hunch: if A is a teleological sentence, and if B translates A, B would also be teleological. I would tend to regard preservation of teleological character under translation as a condition of adequacy for any account of the translatability relation. The teleological character of a sentence is not a matter of vocabulary alone; it is a matter of the logical structure of the conceptual scheme employed in the sentence. I have indicated in Part II what features of a conceptual scheme are presupposed by functional ascriptions. I think this account can be easily generalized to cover the features of the conceptual schemes presupposed by goal-ascriptions. We have to adopt a technical concept of goal-direction, similar in principle, if not in every detail, to those proposed by men such as Sommerhoff and Nagel. If we do this, we may regard any goaldirected activity as defining an activity S6of a system S which is privileged in two ways: 1) any narrowed line in the net-like organization of S which contributes to <, is a function in S (thus admission of goal direction extends the range of admissibility of function ascriptions); and 2) the vocabulary of success and failure is applicable to S, to b, and to the parts (and their activities) of S. All language which essentially exploits these conceptual schemes I regard as teleological language. We may now turn to clause (b) of my thesis, that even though teleological language cannot be translated away, there is a sense in which teleological language is eliminable. The sense is this: given any single case of an activity which is describable in teleological language, that case is also describable in nonteleological language. By this I mean that every observable aspect of the activity which in fact serves as the basis for our application of teleological concepts can also be described by means of a conceptual apparatus that is not teleological in character. This point is best clarified by means of an example. Suppose we are watching a tank in which there is a single anchovy, and that we introduce a barracuda into the tank. At first nothing happens; then, when it would be reasonable for us to suppose

162

Function and Teleology that the anchovy spots the barracuda, the anchovy behaves as follows: he turns sharply, swims quickly to the surface, leaps out of the water, reenters, and repeats the sequence. This whole description (call it B) is, let us at least suppose, in nonteleological language. And it is a description of exactly the same behavior that would serve as the basis for the teleological description (call it A) "On sighting the barracuda, the anchovy engaged in an escape reaction." Of course B does not translate A. A in some ways says less, and in other ways says (and presupposes) much more. A says less than B since B offers details not mentioned in A. It says more, since calling the anchovy's behavior an "escape reaction" implies that it serves the function of escape, whereas this is not implied by B. The utility of the teleological sentence A in relation to nonteleological B lies in the fact that there are contexts in which the 'less" said by A is hardly worth saying, and in which the "more" said by A is very much worth saying. These are contexts in which we are trying to fit the single case into a certain general picture. To see this, consider how the concept of an "escape reaction" can be related to behavior-segments such as that described in B. 1. The details described in B may vary considerably, and still constitute an escape reaction. For example, the same anchovy, or other anchovies of the same species, might omit the direction change, or not break the surface. In other words, the concepts employed in the description of these various cases might not provide the concept of a "kind" with respect to which all of these cases are the "same kind" of cases. 2. The B-type description of the behavior of anchovies might be a description of an escape reaction, while the same ("B-type same") behavior shown by another fish might be behavior of another kind ("A-type kind," such as a mating dance.) 3. Anchovies may have more than one sort of escape reaction in their repertory. The reactions of members of a school might be entirely different. In this case even the term "anchovy escape reaction" would not apply solely to cases like B. 4. The term "escape reaction" applies to very different sorts of behavior throughout a very wide range of taxonomic groups. All these escape reactions might have virtually nothing in common other than the common features noted in teleological language. It should not be supposed that the term "escape reaction" could be defined in nonteleological terms as a complex disjunction of B-type conditions. Such a concept could be constructed, but, even if it were in fact to apply to all and only escape reac-

163

MORTON

BECKNER

tions, it would nevertheless not be the concept of an escape reaction; for it would still be possible for a pattern of behavior not mentioned in the definition to serve the function of escape. CONCLUSION The view of teleology sketched in the above remarks seems to me to offer a piece of candy to both the critics and guardians of teleology. The critics want to defend against a number of things: the importation of unverifiable theological or metaphysical doctrines into the sciences; the idea that goals somehow act in favor of their own realization; and the view that biological systems require for their study concepts and patterns of explanation unlike anything employed in the physical sciences. An important part of their defense has been the contention that teleological language can be eliminated without loss from the sciences. I have argued that this is true: any phenomenon that can be described in teleological language can be described otherwise. On the other hand, eliminability does not mean translatability. I have suggested-I do not see how to prove it-that the teleological character of a sentence is so fundamental that it would be preserved under translation. Teleological character is conferred on a sentence by the manner in which it fits into a conceptual scheme designed for the description of certain classes of systems possessing net-like organization. The elimination of teleological language thus involves a conceptual shift, and involves a different method of classifying the elements of a system. The guardians of teleology-myself among them-have insisted that teleological language is perfectly legitimate. This conclusion is plain if my account of "appropriateness"is correct. Indeed, the development of conceptual schemes which render functional ascriptions appropriate is seen to be just a special case of a general scientific procedure: the designing of languages aimed at bringing to light those causal relations which are most interesting to us. BIBLIOGRAPHY Beckner, Morton. The Biological Way of Thought. New York, 1957. Hempel, Carl G. "The Logic of Functional Analysis," in Symposium on Sociological Theory. Evanston, Ill., 1959. Lehman, Hugh. "Functional Explanations in Biology," Philosophy of Science, 32, 1965. Nagel, Ernest. The Structure of Science. New York, 1961. Scheffler, Israel. "Thoughts on Teleology," BTitish Journal for the Philosophy of Science, 9, 1959. Sommerhoff, George. Analytical Biology. London, 1950.

164

From Machine-Theoryto Entelechy: Two Studies in Developmental Teleology FREDERICK B. CHURCHILL Department of History and Philosophy of Science Indiana University, Bloomington

The 1880's and 1890's were exciting decades for a young embryologist. At the outset, Wilhelm Roux announced his program of Developmental Mechanics (Entwickelungsmechanik ) ,' designed to seduce investigation away from endless comparative descriptions of embryos into an experimental analysis of developmental events. Eduard Pfluger, Gustav Born, and Oscar Hertwig, to name a few, championed the same experimental approach in embryology at nearly the same time. By the 1890's the ranks of experimenters had swelled to include Hans Driesch, Curt Herbst, T. H. Morgan, and many others. It was a time when certain techniques developed in plant physiology and cytology converged onto embryology to broach questions about the activities within the cleaving egg -questions, more specifically, about the role of the nucleus in heredity and development, about the formative influences exerted by some cells upon other cells or by external stimuli, and about the regenerative capacities of experimentally altered embryos. Amphibians and echinoderms were the chief martyrs in these quests, but roundworms, gastropods, even protozoa, served embryology too. There were almost as many explanations of development and heredity as experimental animals, and often the choice of the latter determined the tenets of the former. Yves Delage, at the turn of the century, described over thirty general theories of heredity, most of them contemporary and many of them arising directly from experimental embryology.2 1. Roux first used the term Entwickelungsmechanik in the introduction to a series of papers in experimental embryology. "'Einleitung' zu den 'Beitragen zur Entwickelungsmechanik des Embryo'," Z. Biol., 21 (1885), 411-428. 2. Delage, Yves, L'H&redite et les grands problemes de la biologie

165

FREDERICK B. CHURCHILL

One of the most highly speculative yet provocative theories of the time came from the pen of August Weismann. Since I will refer to it on a number of occasions it is worth describing it briefly. Known as the germ-plasm theory of 1892,3 it was an all-embracing theory which attempted to explain evolution, development, and heredity completely within mechanistic terms. Weismann postulated the existence of a germ-plasm which was located within the nucleus and was subdivided into a hierarchy of self-replicating units, each level of which served to explain a different set of phenomena. The pertinent feature of the germ-plasm theory, as far as this paper is concerned, was Weismann's belief that embryonic development consisted largely of qualitative nuclear divisions which segregated the developmental or chromatin material into increasingly divergent and differentiated parcels. The system functioned like clockwork, it was beautifully logical and clearly stated, yet it was all too often the bete noire against which other embryologists designed their experiments and deployed their own explanations of development and inheritance. Against this background of experimental embryology and Weismann's germ-plasm theory, I wish to discuss Hans Driesch and teleology in biology. As a student, Driesch had for two semesters heard Weismann lecture in Freiburg i/Br. ;4 he had then studied under Ernest Haeckel and in 1889 had earned his doctorate on a subject which reflected the morphological interests of the master. For the next decade Driesch and his close friend, Curt Herbst, played the roles of footloose travelers and free-lance biologists. Interspersed between long trips were stops at the marine stations at Trieste and Naples, where Herbst analyzed chemical stimulants on development and Driesch turned his hand toward experimental embryology. In his first and best-known experiments Driesch succeeded in 1891 in separating the first two blastomeres of sea urchin eggs and showed that each half developed into a whole pluteus.5 This work contradicted earlier experiments performed by Roux on frog blastomeres and directly challenged Weisg6n6rale, 2nd ed. (Paris: Schleicher Fr&res & Cie., 1903). Particularly useful is the synoptic chart on p. 437. 3. I have discussed the origins of the germ-plasm theory in "August Weismann and a Break from Tradition," J. Hist. Biol., 1 (1968), 91-112. 4. Hans Driesch, Lebenserinnerungen, Aufzeichnungen eines Forschers und Denkers in entscheidender Zeit (Basel: Ernst Reinhardt, 1951), 53-54. 5. Hans Driesch, "Entwicklungsmechanische Studien. I. Der Werth der Experibeiden ersten Furchungszellen in der Echinodermenentwicklung. mentelle Erzeugung von Theil- und Doppelbildungen." Z. wiss. Zool., 53 (1891-92), 160-178.

166

From Machine-Theory to Entelechy mann's germ-plasm theory when it appeared full-blown the following year. Later experiments performed on different stages of development, as well as regeneration experiments on Tubularia, confirmed his earlier results, and being no exception to the prevailing mood, Driesch too brought forth elaborate explanations of development based on limited evidence. Two of these explanations will be examined in detail: the first, published in 1894, was later referred to as Driesch's "machine theory of life;" the second, published in 1899, became his declaration of vitalism. In turning to Driesch's Analytische Theorie of 1894,6 one finds an elaborate and at points sophisticated explanation of embryogenesis. The text itself was nearly two hundred pages, and Driesch tells us it was written in Zurich during his summer retreat to the Alps.7 The introduction of the essay presented a critical summary of what he called the "Roux-Weismann theory," a theory which Driesch viewed as "a pyramid stood on its point: below, the hypothetical basic notion, above, finally as a heavy base on top-my simple experimental results." The hypothetical notion at the inverted apex was that of qualitative nuclear division, and as Driesch up-righted the pyramid he intended to replace this basic assumption with a broad base of demonstrable cellular reactions. The analytical theory presented "the typical processes of ontogeny as being triggered by the typical protoplasmic differences inside the germ." 8 Driesch devoted the first section of his monograph to the details of these processes, and it is worth following his argument closely to see the sort of explanation he fashioned. Drawing upon the familiar sea urchin as the paradigm, Driesch illustrated how his analysis proceeded. The formation of the mesenchyme cells at the vegetative pole of the blastula was the first organ-forming process in development. "Why did it occur?" 9 Because such a question was too broad and, in fact, would not be answered by analytic means, Driesch rephrased it. 'What was mesenchyme formation?" If this were viewed as a process of cell multiplication at the vegetative pole of the blastula, his causal analysis forced him to the next question of why this particular place. Some cause of mesenchyme formation must be present and spatially directed. 6. Hans Driesch, Analytische Theorie der Organischen (Leipzig: Wilhelm Engelmann, 1894). 7. Driesch, Lebenserinnerungen, p. 86. 8. Driesch, Analytische Theorie, pp. 11-12. 9. Ibid., 30-31.

Entwicklung

167

FREDERICK B. CHURCHILL

'For every variation which appears," Driesch explained, "we inquire a priori about a cause." 10 What then were the causes, or better yet, the elementary cellular processes upon which Driesch could call? Cytological studies of the preceding two decades had suggested a rich variety of processes which Driesch singled out as exerting a formative influence. These included cell division and growth, the phenomenon of cell migration, and special cell secretions. Moreover, he recognized that internal physical events had a formative role. Such were the pressures of cell growth and the pulls and pushes of germ-layers which Wilhelm His, August Rauber, and Roux had recently described; to these Driesch added capillary action and osmosis." Most basic, however, of all the elementary cellular events were the little understood chemical processes which ranged from changes in salt concentrations to specific reactions in the protoplasm. Driesch concluded that all ontogenetic processes were either chemical in nature or chemically induced, and this assumption made it possible for the investigator to understand development: If the elementary morphogenetic processes make up in their totality the whole of the events of ontogeny then we can say the result on ontogeny comes about through chemical stages. This sentence thus means, that the effect of every elementary ontogenetic mechanism [Auslosung] is a chemical one which has physical and therefore morphological results (growth, etc.). Nothing stands in our way of presenting the majority of such chemical effects as fully graspable and comprehensible; that is we can become clearly acquainted with that newly made substance which initiates the new morphological event as such although we may never say why this alone now arises.12 In intimating that there was a limit to human knowledge about the organism, Driesch also indicated that this was at some point beyond the basic chemical processes. Finally, he argued that the nucleus played the dominant role in directing such chemical changes, although he took a firm stand against Roux's and Weismann's contentions that differentiation was a matter of qualitative nuclear division.13 Driesch described none of these cellular events in detail, but it is well to remember that in his 10. Ibid., 32. 11. Ibid., 35-39, 40-42.

168

12. Ibid., 43-44. 13. Ibid., 45-48.

From Machine-Theory to Entelechy Analytische Theorie he offered a guide to a solution and not the solution itself. Besides the above cellular processes Driesch utilized a specific mode of physiological activity which became an essential part of his understanding of development. He collectively described these reactions under the general category of Auslosungen or "releases," yet since "biological releases" were commonly invoked during these years it is hard to say how Driesch became inspired to think in such terms. His constant companion, Curt Herbst, in an endeavor to show how they could be useful in understanding ontogeny, was involved at that very moment in cataloguing all the known stimuli and responses.'4 Herbst's information was largely derived from the works of plant physiologists, such as Julius Sachs and Wilhelm Pfeffer, and his intent was to apply the methods and some of the conclusions formulated about tropisms to cell differentiation. Toward the end of his long monograph Herbst distinguished between Auslosungreize (releasing stimuli), and Strukturelle Reize (structural stimuli). It is clear that Herbst took as his model for the Auslosungreize the mechanical opening of a valve in a machine-hence the notion of a surge of new activity.15 Driesch, too, used the same mechanical model and referred in a general way to Pfeffer and Eduard Strasburger,'6 although he undoubtedly also discussed the matter many times with Herbst. The important feature of applying the methods and terminology of tropism studies to differentiation rested in their extreme mechanistic implication, and it is not uninteresting that Sach's prize pupil, Jacques Loeb, at that very time was also employing the same tools to analyze animal form.'7 The releases were viewed by Driesch as chemical events which in turn released secondary chemical changes of far greater magnitude than the release or trigger itself. The released reactions could in themselves play the role of trigger for a third generation of chemical reactions, and so ontogeny, becoming an ever-expanding constellation of stimuli and re14. Curt Herbst, "Ueber die Bedeutung die Reizphysiologie fur die kausale Auffassung von Vorgangen in der tierischen Ontogenese," Biel. Centralblatt, 14 (1894), 657-666, 689-697, 727-744, 753-771, 800-810; 15 (1895), 721-745, 753-772, 792-805, 817-831, 849-855. 15. Ibid., 15 (1895), 818-822. 16. Driesch, Analytische Theorie, pp. 178-179. 17. See particularly "On Some Facts and Principles of Physiological Morphology [1893]," reprinted in Jacques Loeb, The Mechanistic Conception of Life, (Cambridge, Mass.: Harvard University Press, 1964), pp. 78-104, as well as the Introduction to the same volume by Donald Fleming.

169

FREDERICK B. CHURCHILL

sponses, continued to progress. Dnresch envisioned that such triggering reactions occurred within cells, between neighboring cells, between germ-layers, or even between organs. He furthermore designated all such triggering interactions as instances of induction, and although his discussion was agam limited to a theoretical level, he cited Roux, Herbst, and Loeb as having discussed actual examples.18 Perhaps the most intriguing of the fundamental processes which Driesch described arose out of his attempts to explain the temporal sequence of stimuli and responses. Cell division itself offered a temporal scale for the ordering of events; so the inductions which followed one another in regular order were associated directly with the passing generations of embryonic cells and organs. In pursuing the problem further, Driesch made one of his most innovative suggestions.'9 Development itself, he proposed, implied a process of differential stimulation. This meant portion x of a hitherto uniform layer or organ acquired histological characteristics (viz. a chemical state) which distinguished it from the remainder of that layer, let us call it portion y. What interested Driesch was the fact that when portion x developed its new characteristics, its capacity to follow the pattern of development which portion y then took became suppressed. His evidence was drawn from halving experiments which he had performed on late blastulae of sea urchins. Where the vegetative half of the severed blastula closed over its wound and carried on the process of mesenchyme formation and gastrulation, the animal pole seemed incapable of developing any further than sealing the cut surface. It remained a blastula, and as far as Driesch could discover its ectoderm was incapable of invaginating and becoming endoderm; it had become determined as far as these two primitive gern-layers were concerned. Driesch introduced two terms to describe what had happened. The prospective potency referred to the multifarious capacity of a particular part to develop along different lines. In the case of Driesch's sea urchin blastulae, the ectoderm at the animal pole had lost its capacity to develop into endoderm; its prospective potency had become restricted. Driesch argued that as ontogeny progressed the prospective potency of all parts increasingly narrowed; so that at the ultimate stage of development the final organs possessed a prospective potency of zero. The termnprospective value Driesch used to denote the ultimate fate of a given part. It became one of 18. Driesch, Analytische 19. Ibid., 75-79.

170

Theorie, pp. 50-62.

From Machine-Theory to Entelechy the mooted questions in experimental embryology of the day as to the point in ontogeny when the prospective value of a given part was actually set. Driesch, like many of his contemporaries, believed that the nucleus played the dominant role in development and heredity,20 but it was also his contention (and this distinguished him from Roux and Weismann) that all nuclei preserved a totality of Anlagen. The stand put him in a curious position, for he had to explain how the nucleus could be selective about incoming stimuli, determine in its own turn the histological character of the cell, and yet preserve the totality of its original constituents-presumably in an unchanged state.2' His solution to an apparent dilemma came in the form of a distinction between the capacity of a cell to receive a stimulus and the capacity of the nucleus to respond. The former depended upon the protoplasm surrounding the nucleus, or, as Driesch expressed it, "The protoplasm is thus the mediator (the 'perception zone') between stimulation and the nucleus (the 'action zone')." This meant that the nucleus could always contain a totality of Anlagen and that the progressive restriction of the prospective potency showed by the cells in development became a cytoplasmic problem: With this we now solve the contradiction just disclosed: in so far as it possesses a nucleus every cell of ontogeny is the bearer in fact of all Anlagen, however, in so far as it possesses a specific plasma body it is only capable of receiving certain causes; since the response to a cause presupposes its reception, the cell therefore can respond each time as a whole to only certain causes.22 For Driesch in 1894 the process of histological change entailed a twofold process on the part of the cell and a stimulus extemal to that cell. The nuclei acted both as stimulators and receptors, and the cytoplasm, acting as selective filters, permitted only certain stimuli to reach the nuclei at given places and at given times. By this elaborate set up of a cytoplasmic "mediator" Driesch believed he could explain the coordinated rhythm of differential development and the har20. Driesch listed the evidence for such a position as: (1) the complicated process of Karyokinesis (mitosis), (2) the dominant role of the nucleus in the events of fertilization, (3) the evidence of inheritance from the father particularly in hybridization, and (4 ) the results of regeneration experiments with single-celled organisms, (ibid., 46). 21. Ibid., 48. 22. Ibid., 79-81. The quotation appears on p. 81.

171

FREDErICK B. CHURCHILL

monious restriction of perspective potencies obvious in every developing embryo. In its process of responding Driesch envisioned the nucleus as discharging chemical "ferments" into the surrounding cytoplasm. There they initiated the histological changes which put the cytoplasm into a different state of receptivity or induced a reaction in a neighboring part. It seems clear that he envisioned a specific ferment for each identifiable histological type, embryological and adult,23 and he argued that this supposition, in contrast to Weismann's germ-plasm theory, kept the complexity of the nucleus to a minimum. After all, there existed far fewer histological types than unit characters. The skeleton-forming mesenchyme cells were all histologically the same regardless of where in the pluteus larva they were secreting the calcareous skeleton. Driesch, however, was not explicit about the nature of these nuclear ferments; in fact, he considered them the most hypothetical feature of his whole system.24 It is interesting to note that Driesch discovered in Hugo de Vries' work the appropriate model for this sort of nuclear discharge; it is particularly interesting since Driesch must have been referring to de Vries' Intracellular Pangenesis of 1889. Here de Vries had described a particulate theory of inheritance which by chance had a strong influence on Weismann's germplasm theory as well.25 That Driesch, however, found de Vries an inspiration not for a preformed nuclear arrangement but for a nuclear melange of assorted chemicals attested to a peculiar divergence in possible renderings of de Vries' work. "The Nucleus," Driesch explained, "appears to us a mixture of stuff, we speak not of a structure of the nucleus; thus our theory in respect to form as such is in its fictitious dress also epigenetic: the starting form, the structure of the egg, is far simpler than the end form." 26 Driesch thus presented the nucleus as a mixture of ferments with no set arrangement, and as though to distinguish his ideas even further from the germ-plasm theory, he made no effort to relate cellular differentiation and mitosis. In his Analytical Theory Driesch brought a number of fundamental assumptions to bear on the general question of the spatial ordering and temporal sequence of embryological events: (1) the great array of elementary chemical and physi23. 24. 25. 26.

Ibid., 179-180. Ibid., 91. Churchill, "August Weismann," p. 106. Driesch, Analytische Theorie, p. 86.

172

From Machine-Theory to Entelechy cal phenomena, any one of which could act as a release of further events; (2) the distinction between the prospective potency and prospective value of the cell; and (3) the threecornered balance between the nuclear reception of a stimulus, the release of specific ferments, and the continued and restrictive differentiation of cytoplasm. The process of development implied that each stage was not so much the cause of the next stage as its "receptive station." 27 Since each nucleus remained the bearer of all "Anlagen," ontogeny was not to be seen as a determined unfolding of set steps. The prospective potency simply narrowed as the development of each part proceeded toward an ultimate value. The fate of any cell was in truth a function of its position with respect to the whole, and if its position changed, as was bound to happen with sectioning experiments, the fate changed within the limits of the prospective potency at the time. When one turns to the second major section of Driesch's text, it is immediately clear that he was dissatisfied with the highly imaginative and completely mechanistic explanation of ontogeny which he had just completed. Dissatisfied not so much because of the explicit details, but because of what he felt was a serious limitation to the entire analytical approach. This second section was entitled "The Ontogeny as Development in Light of Teleological Considerations." Driesch stated his position at the outset: One has indeed expressed the opinion that every stage of ontogeny is the necessary consequence of the preceding stage and the cause of the following one; this sentence, however, is not to be accepted without further ado, for ontogeny is not a united process but is a combination of many processes which are not only in part totally separate from one another but in this independence are actually "given" processes. In other words, we understand . . . these processes only in part by understanding their causes. We must therefore explain . . . ontogeny as a causal regularity striding along in a pronounced mystery; causal analysis of ontogeny leaves us only fragmentary pieces.28 His analysis had left Driesch a description of individual processes and a recognition of the balance between stimuli and responses; yet he felt denied an understanding of the given harmony of ontogeny which invariably fashioned a mature animal from distinct and often unrelated events. Driesch 27. Ibid., 82-83.

28. Ibid., 128.

173

FREDERICK B. CHURCHILL

claimed that he remained as unenlightened about what was going on as if he had wandered onto a pier during the initial stages of the construction of a ship.29 Parts and pieces lay in confusion about him, and he could understand their relation only when he knew their ultimate purpose. Ship and pluteus larva were the undeniable goals of given processes, and only by introducing a teleological view could he understand them. Yet Driesch made it clear that such a teleological view was not to be construed as simply an intellectual reversal of the causal chain. Such a heuristic query about the end point of a process could be helpful, he remarked, in understanding the upheaval of a mountain range or creation of a river delta; the geologist simply worked backward in time so as to gain a better understanding of the phenomenon at hand. This heuristic use of teleology was a common practice of contemporary embryologists; C. 0. Whitman, E. B. Wilson, and Frank Little used such language throughout their cell lineage studies.30 Driesch, however, had something very different in mind. He turned to the authority of Kant in invoking a teleological interpretation of the organic world. Setting aside the question of whether he properly understood the Critique of Teleological Judgment, it is clear that he embraced a teleological approach as necessary for the understanding of the harmony in ontogeny which he had so belabored in the first section. The completed adult or mature larva was an explanation in itself for the structure of an egg. "Said in another way," Driesch argued: Because the viable whole is given as a clear recognizable end of the totality of all the processes of ontogeny, we judge on the ground of an objective necessity therefore, these processes to be as though they were fixed according to quality and order by an intelligence. With these words we give the really adequate expression to the critical teleological standpoint." 31 For a full understanding of the mature functioning system, 29. Ibid., 129. 30. Ibid., 129. See also Charles Otis Whitman, "The Embryology of Clepsine," Quart. J. MiCroscop. Sci., 18 (1878), 215-315; Edmund B. Wilson, "The Cell-Lineage of Nereis. A Contribution to the Cytogeny of the Annelid Body," J. Morphol., 6 (1892), 361-480; Frank Lillie, "The Embryology of the Unionidae, A Study in Cell-Lineage," ibid., 10 (1895), 1-100. J. H. Woodger has made a similar observation about Wilson in Biological Principles: A Critical Study (London: Routledge & Kegan Paul, 1967), pp. 431-432. 31. Driesch, Analytische Theorie, p. 131.

174

From Machine-Theory to Entelechy Driesch applied the same standards, and his teleological interpretation of both ontogeny and physiology seemed to imply that the architecture given to the egg at the outset was beyond analytical investigation. At least one critic, Wilhelm Haacke, was quick to point out that one of the great achievements of the theory of organic evolution was its capacity to explain the origin of animal forms and hence particular egg structures.32 In this connection it is not surprising that Driesch extended the same teleological argument of a goal-directed process to organic evolution. Like many of his contemporaries Driesch saw in Darwin's theory of natural selection an argument for blind chance creating complex forms. He added the further reproach that it was impossible for him to conceive how a viable and perfected species could change into a second viable species by passing through generations of imperfect forms. The ridiculousness in this stance approached the comical from the embryologist's point of view (". . . so verwandelt sich die Absurditat der 'Zufallsansicht' beinahe in Komik"), for imperfect eggs couldn't even reach maturity. The logic of the Darwinian argument seemed to Driesch as absurd as his understanding of Darwin may seem to us. But his rejection of natural selection forced him (or perhaps allowed him) to explain organic evolution in terms of a "given" functional harmony in egg and adult alike which purposefully responded to a changing environment.33 In concluding his teleological discussion, Driesch introduced a final term which he declared was nothing more than a substitute for the intelligence which seemed to order biological events. "Instead of saying the ontogenetic elementary-events are so arranged with respect to time, place and quality as if they were arranged by an intelligence concerned with the goal of development, we now say that the Bildungstrieb has arranged these events." 34 The term Bildungstrieb served as a useful expletive. It was the Bildungstrieb which created the structure of the egg, it was the Bildungstrieb which constructed the self-regulating physiological system, it was the Bildungstrieb which induced the new phylogenetic adaptations. 35 Driesch did not appear to rely on the term once he introduced it and explained its teleological import; he was, however, careless at the outset when he likened its use to the physicist's use of Kraft. 32. Wilhelm Haacke, "Die Formenphilosophie von Hans Driesch und das Wesen des Organismus," Biol. Centralblatt, 14 (1894), 712-718. 33. Dnesch, Analytische Theorie, pp. 133, 135-139. 34. Ibid., 139. 35. Ibid., 139-142.

175

FREDERICK B. CHURCHILL

In the last section of his Analytical Theory Driesch reaffirmed his intention to stay within the strictures of what he considered Kantian teleology. Both causality and teleology, he argued, were a priori concepts, but were perfectly valid and even necessary components of research. The causal analysis which he had performed in working out the processes of ontogeny had revealed the egg to be a complex chemical and physical machine. The structure of the egg, however, could only be understood in terms of the ultimate goal of development. Causal analysis revealed the harmony in the developmental processes, teleology explained the givenness of the starting structure.36 With his Analytical Theory and particularly with his introduction of the Bildungstrieb, Driesch appeared to his contemporaries to be steering all too close to the forbidden shoals of vitalism. Both Roux and Emil du Bois-Reymond were critical of the tum in Driesch's thought,87 and Driesch two years later was forced to disclaim any such vitalistic intent. "Vhat I represented therefore was absolutely not 'Vitalism', but, at least as far as living phenomena come into the question, was directly the current view of physico-chemical dogmatism." 38 He went on to explain that he had simply been willing to see that one consequence of this dogmatism was its inability to account for the purposefulness of form. In its restatement the argument remained essentially the same as that of 1894. Driesch had the opportunity to emphasize the difference between the causal understanding of the individual physiological and embryological events and the teleological understanding of a given structure. Together, the "physico-chemical" and "tectonic-teleological" view of ontogeny became what Driesch called his machine-theory of life. The mechanisms were researchable, the architecture merely describable-and describable most satisfactorily in teleological terms. Driesch completed the paper announcing his support of 36. Ibid., 162-166. 37. Wilhelm Roux, Gesammelte Abhandlungen uber Entwickelungsmechanik der Organismen, 2 vols. (Leipzig: Wilhelm Engelmann, 1895) II, 1020; Emil du Bois-Reymond, "Uber Neo-Vitalismus [1894]" in Reden von Emil du Bois-Reymond, 2 vols., ed. Estelle du Bois-Reymond (Leipzig: von Veit & Comp., 1912) II, 507-508. Du Bois-Reymond is here actually criticizing an earlier paper of Driesch's which referred to "ein unfassbarer Regulator" instead of a Bildungstrieb; the implications are the same however. 38. Hans Driesch, "Die Maschinentheorie des Lebens. Ein Wort zur Aufklarung," Biol. Centralblatt, 16 (1896), 353-368. Quotation appears on p. 365.

176

From Machine-Theory to Entelechy vitalism in 1898, and it was published the following year.39 It may seem curious at first glance that he should have submitted this declaration to Roux's Archiv fur Entwickelungsmechanik, but Driesch always considered himself part of Roux's experimental tradition, and in his own autobiography Roux was glad to list Driesch as one of his disciples.40 That Driesch turned away from his earlier machine-theory of life and eventually from science altogether should not conceal the fact that he had been a highly ingenious experimenter and that many of his theoretical discussions of embryological events carried on in a style very similar to Roux's own quasi-philosophical forays. The problem of localizing embryonic events was one of the major concerns of experimental embryology in the 1890's and indeed is still one today. Both Weismann and Driesch pitched the question of "Why an event occurred at a given place?" in a manner and at a structural level which seemed far beyond the means of an anatomical or chemical solution; both of their discussions undoubtedly suffered for it. For Driesch the question about the position as well as the specificity of elementary embryological events was important in the framing of his machine-theory of 1894; it was essential to his paper of 1899 which he entitled "The Localization of Morphogenetic Processes, a Proof of Vitalistic Phenomena." In 1894 the limits of a causal analysis for Driesch had been at the "givenness" of the entire structure of the egg; by 1899 he felt analysis was unable even in a hypothetical way to explain certain cases of localization. He was led to such a conclusion largely by pursuing the logical consequences of his machine-theory, but a number of experiments performed on the gastrula stage of the sea urchin highlighted the difficulties in a way which the blastula did not. The most interesting of these for his argument consisted of dividing experiments which he performed in Naples during the winter of 1894-95 and which he published in Roux's Archiv the following summer, the year after his Analytical Theory.41 39. Hans Driesch, "Die Lokalisation morphogenetischer Vorgange. Ein Beweis vitalistischen Geschehens." Arch. Entwickelungsmech., 8 (1899), 35-111. 40. Hans Driesch, "Wilhelm Roux als Theoretiker," Natur-wissenschaften, 8 (1920), 446; Wilhelm Roux, "Wilhelm Roux in Halle a. S.," Die Medizin der Gegenwart in Selbstdarstellungen, 2 vols., ed., L. R. Grote (Leipzig: Felix Meiner, 1923), I, 173. 41. Hans Driesch, "Zur Analysis der Potenzen embryonaler Organzelen," Arch. Entwickelungsmech., 2 (1895-96), 169-203. In 1899 Driesch referred to this work as the stimulus which forced him to rethink the whole matter of localization. Driesch, "Die Lokalisation," p. 40.

177

FREDERICK B. CHURCHILL

When he sliced the completed gastrula near the equator so that each half contained ectoderm and endoderm, both halves sealed over the wound of incision and developed into healthy pluteus larvae of reduced size (compare Fig. 6 with Fig. 5). What fascinated Driesch was not only that both halves possessed the potential potency to continue normal development, but that there was a rapid establishment of the normal proportions of the original whole. The clearest case in point was

8.

j

XjcircicYL

9.

Figs. 1-14. Sphaerechinus granularis from Hans Driesch, "Zur Analysis 2 der Potenzen embryonaler Organzellen" in Arch. Entwickelungsmech. (1895), Plate XV. 1-4. Diagrams which show defects of the ectoderm and the endoderm caused by slicing operation on the gastrula: 1. animal portion with both germ-layers removed; 2. vegetative portion with germ-layers removed; 3. a lot of animal ectoderm removed; 4. nearly a half of the ectoderm removed. 5. Normal pluteus in outline. 6. Pluteus in outline, arising from a gastrula with a marked defect especialiy on the ectoderm. This figure is drawn on the same scale as Fig. 5 and stems from the same culture. 7-8. Plutei arising from gastrulae with large defects; notice the gut. Scale and culture same as Figs. 5 and 9. 9. Threefold segmented gut of a normal pluteus, same scale as Figs. 7 and 8. 10. Normal pluteus from the side in reduced scale. 11. Pluteus with a one-sided skeleton arising from a large defect. 12. Animal portion of a sliced mesenchyme-larva in outline, in order to show the curled form. 13. Gastrula with a small lateral gut arising from only a portion of the animal pole of a sliced mesenchyme-larva, which was sliced more as shown in Fig. 14.

178

From Machine-Theory to Entelechy the development of the primitive gut, which in normally developing gastrulae contained two constrictions and consequently a fore-, mid- and hind-gut (Fig. 10). These three sections maintained strict proportions to one another regardless of the size of the gut (compare Figs. 7 and 8 with Fig. 9),42 proportions which in undisturbed embryos could be adequately explained by Driesch's 1894 conception of inductive stimuli and responses. Driesch argued that one could envision two constriction-determining stimuli, which, having been released from either pole, traveled along the longitudinal axis and elicited corresponding constrictions in the gut at the two places where the cytoplasm acted as receptive filters. Or, he suggested, if one preferred, one could suppose a single stimulus which similarly localized the mid-gut and incidentally determined the appropriate size of fore- and hind-gut.43 To apply the same explanation to an arbitrarily halved gastrula, however, introduced the complicating problem of explaining reduction of strength of the polar stimuli and a relocation of the receptive cytoplasmic filters. "In the events here described," Driesch noted in retrospect, a sufficient basis does not reveal itself which would explain the correctly proportioned three-fold arrangement of the gut of the cut larvae; that correctly proportioned segmentation points far more to a type of phenomenon which is essentially not of a mechanical but of a specific vitalistic sort.44

Postponing for a moment what he meant by phenomena "of a specific vitalistic sort," let us trace closely Driesch's analysis of the events. The way out of the new difficulty was through the postulation of what he called a harmonious-equipotential system and through the enlargement of the role of teleology. In introducing his concept of the harmonious-equipotential system Driesch put the emphasis on the problem of localizing certain ontogenetic events. The sea urchin blastula presented the paradigm. His early experiments had shown that every cell of the blastula had an unlimited potency and could differentiate into all possible embryonic and adult organs; with the gastrula there was some restriction, but each primitive gut cell could develop into any of the possible products of the endoderm. In short, Driesch claimed, "Jedes Element kann Jedes" (every element can become every element). So 42. Ibid., 176-177. 43. Driesch, "Die Lokalisation," p. 76n. 44. Ibid., 39.

179

FREDERICK B. CHURCHILL

far the problem seemed little different from his 1894 description of prospective potency, but the experiments on the gastrula and his discussion of the proportions of the gut forced Driesch to go further. Not only could every element become every other, but every element remained within strict proportions to the whole: "in that every effect happens only once or a definite number of times and stands in a fast relationship to all other effects, so then I call such a living system . . . as the cleaved sea urchin's egg: a harmonious-equipotential system." 45

Driesch warned his readers that a harmonious-equipotential system did not mean an isotropic system. The blastula of the sea urchin, after all, must possess an orientation which acted as the initial cause for the localization of mesenchyme growth and invagination. This was an important point to remember when one considered the development of two normal pluteus larvae from a divided blastula. In this case, Driesch explained, the severed halves upon closing-over reinstated the original orientation and thereupon possessed within their new systems the appropriate localizing stimulus necessary for further normal development. By contrast a Ctenophore egg, even when divided at the two-celled stage, appeared to lack the regulative ability to reinstate the initial orientation and therefore was unable to develop along normal lines.46 Driesch seemed to be arguing that the questions of localization and prospective value in the harmonious-equipotential system were bound to the question of regulation. Referring again to his experiments on the gut, if he could assume that the original orientation of the severed gastrulae halves had been reinstated, that is, if the sealed halves were identical to the normal gastrula in every respect except size, then Driesch could sally forth with his physico-chemical machine of stimuli and responses. The size of the gut would be smaller, but the proportions of the three gut segments would remain normal. It was the harmonious-equipotential system which possessed the "regulatory capacity" to reinstate that all important initial orientation.47 To elucidate the distinctness of the harmonious-equipotential system Driesch reviewed the explanation of regeneration offered by his machine-theory.48 This earlier explanation had 45. Ibid., 74. 46. Hans Driesch and Thomas Hunt Morgan, "Zur Analysis der ersten I. Von der Entwickelung des Ctenophoreneies. Entwickelungsstadien einzelner Ctenophorenblastomeren," Arch. Entwickelungsmech., 2 (189596), 204-215. 47. Driesch, "Die Lokalisation," pp. 75-77. 48. Ibid., 70-72.

180

From Machine-Theory to Entelechy sufficed in limited cases only, such as with willow trees and marine whelks. Both organisms exhibited great regenerative capacities; a willow cutting could sprout new shoots and roots when properly planted, and the whelk could regenerate an excised foot. In both cases, however, Driesch maintained that regeneration was of a determined and limited sort. The cells of the willow cutting contained a twofold prospective potency, and depending upon the stimulus, either a shoot or a root budded forth in an unorganized fashion. Each cell of the whelk's foot contained the prospective potency of the cells distal to it, so that under the stimulus of having a portion of the foot removed the remaining cells simply regenerated the missing part. Localization during the regenerative act presented no problem, for the operation itself determined the place of the new part. In neither case was there a necessary restoration of the initial orientation; although the prospective potency of all cells remained greater than their prospective value, the outcome of regeneration depended upon nothing further than cell multiplication and differentiation. In brief, these were cases of determined-equipotential systems. As an harmonious-equipotential system, the gastrula of the sea urchin was called upon to perform the far more complex task of reestablishing a new orientation and confining further development to new dimensions. Driesch made his analysis seem more thoroughgoing, although he hardly clarified matters, by a brief excursus into the very meaning of cause and effect.49 The upshot appears to be his conviction that all physico-chemical causes reappeared in some manner in the consequent effect; with cases of mechanical events and energy conversion this meant that effect was directly related to its cause. The "regulatory" response of the harmonious-equipotential system, however, did not carry a similar relationship. Since the response of each cell was a function of its place in the whole rather than the initial cause which brought it forth, the total response was not related to the embryologist's scalpel but seemed solely coordinated to the goal of normal development. The effect was goal-directed in a much more immediate sense than the teleological organic-machine of 1894. Although Driesch did not say so explicitly, his excursus was an indirect way of saying that his postulated three-cornered balance between stimuli, cytoplasmic filters, and nuclear ferments could not explain the "regulatory" response. To argue from the position of his machine-theory of life, Driesch would have had to suppose 49. Ibid., 84-85.

181

FRDERCK

B. CHURCHML

that the arbitrary act of cutting a gastrula in half caused the appropriate chemical readjustinents in all the cytoplasmic filters and at the same time elicited the release of appropriate stimuli from newly located inductive centers. This was too much for the machine-theory to bear. Driesch argued then that this purposeful regulatory response (or that which he renamed an Antwortsgeschehen) in the harmonious-equipotential system was unique. To be sure, man-made machines could have regulatory mechanisms, but only in the sense that the machines substituted one "typical" course of events for another "typical" course; put into modem parlance this meant that if one program failed, the machine could switch over to another which had already been built into the system.50 In the harmonious-equipotential system, however, there was no substitute causal chain since there was an infinite number of arbitrary ways to cut the sea urchin in two. It was the "atypical" state which led to "typical"development. It is worth noting that these two terms, "typical" and "atypical," had a very definite meaning in late nineteenthcentury embryology. In drawing up his program for Entwickelungsmechanik in the 1880's, Roux designated "typical" as the correlative adjective for the genetic type;51 this meant that a "typical" response was not just a normal response but any response, normal or abnormal, which might still be part of the type or species itself. "Atypical" referred to those responses which were not of the type, and in Roux's mind this meant an externally induced and organized response. I think it highly likely that Driesch was using these terms in the same way; if so, it makes the failure of his machine-theory all the more understandable. The hereditary material, and for Driesch this implied both the nuclear ferments and the cytoplasm, could not direct an "atypical" response because the latter was not part of the system. The regulatory response was sui generis because of the dilemma which arose out of Driesch's peculiar understanding of the stimulus and response systems. The centers of induction were able to control the proportionality of the new organic whole by means of controlling the extent of its own sphere of influence; on the other hand, the reacting system 50. Ibid., 86, 89-90. 51. The clearest exposition of the use of these ein Wilhelm Roux, Die Entwickelungsmechanik, ogischen Wissenschaft (Vortrage und Aufsatze chanik der Organismen, vol. I; Leipzig: Wilhelm 182-186.

182

terms can be found in neuer Zweig der bioluber EntwickelungsmeEngelmann, 1905) pp.

From Machine-Theory to Entelechy had the capacity to rearrange all its elements to the new needs of the reconstituted whole. The necessary mutual adaptations of both sides of the system arose before a normal physico-chemical chain of events could bring about their mutual adaptations. Yet instead of trying to tinker with the mechanics of the entire system, perhaps directing more explicit questions at the nucleus and designing experiments to find its make-up, Driesch supplemented the original teleological element of his machine-theory. Throughout his entire discussion of the harmonious-equipotential system Driesch continually referred to the evident purposefulness in the "regulatory" response. Must the physical and chemical events themselves become the intimate bearers of purpose in a way which was not intended in 1894, when Driesch considered the egg a goal-directed machine arranged as if by an intelligence? The real reason for Driesch postulating the existence of the harmonious-equipotential system with its unique regulatory response appears to be that it was the one way for him to preserve the integrity of these physical and chemical events and at the same time allow for the complex contingencies of regeneration. "Only in this way," Driesch remarked: can we free the "causes," in the strong sense that we give to the word, from teleological impurities, and only in this way can we transfer, in the form of its "answerability," the teleological entity entirely into the "conditions of the system"; as such, it becomes the integrating ingredient as shown in the sense of an elementary or indivisible characteristic.52 Driesch thus charged the "answerability" or regulatory-capacity with the purposefulness of the system. The unfathomable element, which for the machine-theory in 1894 had legitimized teleology as a mode of scientific inquiry, by 1899 become attached to a regulative response. It was at this point that Driesch found it useful to distinguish between "static" and "dynamic" teleology. I do not know whether these terms are original with him, but as Driesch applied them they were convenient for contrasting the teleological status of his 1894 machine-theory with the teleological processes of his harmonious-equipotential system.53 It was 52. 53. when Hans don:

Driesch, "Die Lokalisation," p. 96. Ibid., 103. Driesch elaborated upon this distinction in 1907 and 1908 he delivered the Gifford Lectures at The University of Aberdeen. Driesch, The Science and Philosophy of the Organism, 2 vols. (LonAdam and Charles Black, 1907-8), II, 135-136.

183

FREDERICK B. CHURCHILL

the dynamic teleological behavior which was unique and permitted Driesch to invoke a vitalistic theory. In 1899 he did not specify the causal contents of the Antwortsgeschehen, or regulatory response, for after all, the main thrust of his declaration was to show the goal-directedness of the harmoniousequipotential system. He did observe, "Only this one thing, that namely the 'vitalism' exists not perchance in the introduction of a new form of energy . . . what it 'introduces' as 'agent' is something completely and essentially of a different sort."54 Driesch placed the Antwortsgeschehen beyond the constellation of physico-chemical events. The ontological status of such an agent was now experimentally secured; it appears to be only an incidental part of the story that Driesch waited till later to embellish it with the designation of Entelechy. Reflecting upon the change of mind which I have just related in detail, Driesch remarked in his own biography that as early as 1895 he had begun seriously to consider a vitalistic explanation of his experimental results. In a well-known passage he remarked that such a possible solution occurred to him suddenly while taking a stroll in the woods about Zurich. Further on he was less dramatic but more explicit. "But where in the mechanistic sense," he claimed: "was there then the cause for the fact that in a given case this element renders this result and that element that result from the contents of its own completely homogeneous potencies? Formative stimuli in the sense of Herbst just weren't there." 65 Driesch's remark was more telling than he perhaps realized. He had commented on the failure of Herbst's formative stimuli in his 1899 declaration of vitalism,56 but he failed to recognize the extent to which his machine-theory had snared him by the promise of the simplistic chemical solution as suggested in Herbst's catalogue of trophic and taxic responses. The three-cornered interaction between centers of stimuli, nucleus, and cytoplasm was an ingenious exploitation of Herbst's study of formative stimuli, but when it failed, as it surely had to, it was Driesch's original teleological commitment which led him out of the dilemma. 54. Driesch, "Die Lokalisation," p. 109. 55. Driesch, Lebenserinnerungen, pp. 108-111. Quotation appears on p. 109. 56. Driesch, "Die Lokalisation," p. 57. "Auch verdanken wir ihm [Herbst] eine logische Klassifikation *formativer Reize* nach ihrem und an sie wollen wir bei unserer Frage nach dem *Causalwerth* lokalisirenden Werth derselben zunachst ankniupfen, wobei aber von vorn herein betont sein muss, dass HERBST eben seine KMassifikation nicht zu diesem unserem Zweck angesteilt und die Lokalisation des morphogenen Geschehens uberhaupt nicht zu einem gesonderten Problem gemacht hat."

184

From Machine-Theory to Entelechy I suspect Driesch's ultimate turn was closely linked to his conviction that the nucleus was comprised merely of a uniform mixture of chemical ferments; that is, that the chromosomes had no formatively significant organization. Such a view was not uncommon, and with at least one other adherent bore all the marks of an immediate and strongly "reductionist" point of view.57 If the "reductionist" appellation is appropriate for the Driesch of 1894-and I think it has to be-it suggests that he denied himself the possibility of utilizing a broad range of internal cellular "formative" events. Weismann, for all his shortcomings, promoted this second approach as he discussed on a theoretical level the role of the chromosomal elements in development. It was the experimental cytologists, however, men such as Boveri,58 who fruitfully pursued that approach in vivo. 57. Ernst Haeckel had the same attitude. See Churchill, "August Weismann," pp. 98-99. 58. Baltzer has a nice comparison of Driesch's and Boveri's scientific approach. Fritz Baltzer, Theodor Boveri, The Life and Work of a Great Biologist, trans., Dorothea Rudnick, (Berkeley and Los Angeles: University of California Press, 1967) pp. 106-114.

185

Explanationin the BiologicalSciences MICHAEL SCRIVEN Department of Philosophy University of California, Berkeley

The changes in biology and in other sciences since the eighteenth century have made biology an increasingly typical science. If the historians of science can forgive a caricature, I would say that biology was once a mixture of magic and mechanism, in which the latter element gradually gained the upper hand; that it came to be dominated by a relatively novel mode of explanation which served as a paradigm for the burgeoning social sciences; and that finally it was invaded by the reductionist forces of molecular biology as were classical thermodynamics and behaviorist psychology among the other "pure" macro-subjects. Today it contains, and for obvious reasons will always retain, substantial commitments to both the ecological-evolutionary domain and the molecular-chemical domain; in the jargon-an idiographic-historical aspect, as well as a nomothetic-dynamic one. The peculiar problems of uniqueness and predictability arise most clearly in connection with the former aspect and hence it is to this that my remarks are most immediately applicable. But this does not imply that I think it is more important than the other; no such comparative evaluation seems useful. Nor do I think the methodological phenomena to which I attend are only to be found in evolutionary biology; they most certainly occur in molecular biology, as indeed in classical astrophysics. But they are most clearly seen in the historical side of biology or its short-term sidekick, ecology, precisely duplicating-it appears to me-the situation in geology, geography, government, and psychiatry, in all of which there is currently sharp professional competition between the historical-idiographic enthusiasts and the nomothetic-dynamic. I shall treat briefly several issues that bear on the methodology of these double-aspect subjects, beginning with a discussion of the nature of philosophy of science and proceeding to a general sketch of what one might call macro-philosophy of science.

187

MICHAEL SCRIVEN

THE NATURE OF PHILOSOPHYOF SCIENCE In my view, philosophy of science is best thought of as simply a meta-science, like the sociology of science, the economics of science, the psychology of science, and so on. Philosophy is not exactly a science, any more than history is, but it is best to think of it this way because such a conception puts its descriptive and nomothetic features into correct perspective. Just as a physicist will unhesitatingly evaluate explanations on the basis of his physical knowledge, so a philosopher of science is entitled to evaluate accounts of explanations, theories of explanation (descriptions of their general logical features), on the basis of his knowledge of the logical properties of good individual scientific explanations. Of course, it is some set of scientists' judgments that determines the philosopher's data, but that does not mean the individual scientist or most scientists at one time cannot be corrected by the philosopher, since he may have extracted from his research certain key features of explanation the necessity of which has been overlooked by scientists engaged in the thrills of scientific conflict. This situation rarely persists for long since many scientists are also part-time philosophers of science and usually appeal to logical considerations quite quickly, often prematurely, in their debates with other scientists. But there is nothing more surprising about this possibility of normative philosophy of science than there is about normative science, that is, about one scientist correcting another on the ground that this theory does not fit in well with well-established physical principles. Metascience is empirical in the same sense as anthropology, though it studies verbal behavior rather than other quasi-arbitrary social customs. It is certainly closely analogous to the lexicographer's activity, which is surely empirically based, and just as clearly it forms a basis for normative and corrective claims. Frederick Churchill has kindly defended philosophers of science on the grounds that they are not really telling scientists how to do science. But sometimes they are, and certainly they are involved in a subject which entitles them to do just this at times. It is really much better to think of philosophy of science as an activity defined by its relation to "straight" science, or "science proper"-in other words, science as it is usually conceived, rather than as what people called philosophers of science do. For what we do is not distinguishable from what some scientists do some of the time and very properly. We are simply specialists in the logical analysis of scientific concepts, just as we might be specialists in the economic effects of scientific research. And some scientists achieve their most striking results from

188

Explanation in the Biological Sciences consideration of what are simply logical points about science, whether raised by a colleague with the same union card or a philosopher, just as some philosophers of science have made significant contributions to straight science, from Aristotle to Kant's hypothesis about the creation of the solar system and Russell's hypotheses about the fundamentals of mathematics. Scientists who do not like philosophers often say that such contributions are not made by a philosopher but by a man who was partly a scientist, or even "really" a scientist. But the other side of that coin is that many scientists, including Einstein, must then be regarded as partly philosophers or even "really" philosophers. The payoff is scientific progress, and that requires some logical analyses, for example of the concepts of distant simultaneity, or of species, or of information-content, or of causation, whoever does them. The real question is whether any straight philosophers of science ever say anything that leads their listeners to improve their practice as scientists. And that challenge I regard as a legitimate, indeed as the principal, criterion for criticizing my own work. I think there is another defense for the philosophy of science, viz. as a "pure" subject. But I find that a less challenging and, it is obvious, socially a less valuable goal. In either case, the task of philosophy of science is the same: to produce good analyses of scientific concepts. And the task of the history of science is similar with respect to the time dimension. We have to tell it like it is, and that means both philosophers and historians provide data in terms of which it will be possible to criticize current practice destructively and decisively. To give two examples: it is inconceivable that any historian of psychology could fail to point out some howlers in recent psychology of perception that have simply been repeated because their perpetrators were ignorant of their earlier resolution. And I do not think a student of the Whorfian hypothesis or of Talcott Parsons' work could have failed to benefit from the trenchant criticism philosopher Max Black has directed to those topics. So we philosophers have swords in our sheathes, but so do you as scientists. Just as we can appraise one aspect of your work, so you can do the same to us. Our task is to benefit each other's main enterprise as well as we can. MACRO-AND MICRO-PHILOSOPHYOF SCIENCE The debate between the reductionists and the historical biologists is simply one example of a debate that has occurred in many subjects at many times in their history-one thinks

189

MICHAEL SCRIVEN

of Ostwald going to the grave (or nearly that far) denying the reality of the atom and of Skinner condemning the psychophysiological approach as irrevelant to psychology. On the basis of my analogy between philosophy of science and science, I wish to propose the existence of a radical difference of approach within the philosophy of science which corresponds to that difference in science. The terminology which the discipline of economics uses to mark the distinctions is the one I shall use, the difference between the macro and the micro approach, partly because it already has a more general currency and partly perhaps because in economics the two approaches have been reconciled, and I would hope for the same outcome in my own subject. Let me first mention one or two alternative terminologies that I considered, for this is as good a way as any to illuminate the meaning I wish to attach to the terms I do use. In cryogenics about twenty years ago an ingenious hypothesis called the two-fluid theory was proposed to explain the bizarre behavior of liquid helium. It was widely described as a phenomenological theory, meaning one which (allegedly) had no ontological commitment as to the real nature of liquid helium but only proposed an "as if' account: if one regarded this substance as if it were a mixture of two fluids of very different properties, one could account for the fountain effect, and so on. I wish to introduce what might, in this sense, be called a "phenomenological" philosophy of science. The reason I do not call it so is a problem of confusion with the branch of contemporary philosophy called phenomenology, which may well have its own philosophy of science (though I know of no substantial body of such) and which most certainly would be very different from what I am talking about. Incidentally, the sense in which the two-fluid theory could actually explain phenomena while denying reality-content is quite obscure, and that same conclusion applies to my own proposal; the situation reminds one of Galileo and his "calculational device" defense against the heresy charges. Real explanations are reality-based. One might also talk of a philosophy of science-in-practice as opposed to a philosophy of science-in-principle, and that distinction conveys much of my intent. Or one might talk of a phenotypical philosophy of science by contrast with a genotypical one, and convey a similar point; or of a philosophy of the approximational sciences, and so on. Professor Simpson has said that most philosophers of science are, as everyone knows, really philosophers of physics. Indeed, it could be said with some plausibility that most

190

Explanation in the Biological Sciences philosophers of biology are really philosophers of physics; the same, indeed, could be said of most philosophers of the macro-sciences in general. It is of great importance, therefore, that one distinguish macro-philosophy of science from philosophy of the macro-sciences. The latter might be simply, in one worker's view, the same as philosophy of physics. But macrophilosophy of science, like macro-economics, is by definition different; it may be ultimately reducible to the other but it is the phenomenological account of the logic of the macrosciences, which is obviously different with respect to many considerations which we shaUl shortly itemize. In saying that philosophers of macro-subjects are usually philosophers of physics, one does not just mean-and I believe that Professor Simpson did not just mean-that these philosophers do not know (or care, which may be more important) as much about biology as they do about physics-though that is often true -but rather than they are meta-reductionists who wish to reduce, not one science to another, but the philosophy of one science to the philosophy of another. I think someone should be trying to do that, just as reductionists should be working on biology and psychology and chemistry and engineering. But I doubt if it is the only useful approach in the philosophy of science, any more than in the substantive sciences. Let us set out some propositions in macro-philosophy of science and see whether it may not be a more useful instrument for the macro-scientists, regardless of whether one claims it is different for logical or merely factual reasons. In macro-philosophy of science: 1. Determinism, with respect to macro-state variables, is often an excellent belief, especially in limited areas. That is, determinism is true, or true in very nearly all cases, within certain domains; it covers much of the action, including voluntary human behavior, in other areas (for example, spread of phenotypes in large populations). Of course, in the ultimate micro-science of particle physics it is false and very significantly so; its substitute there is statistical determinism, not limited determinism (each is both stronger and weaker than the other). The function of macro-determinism is thus twofold: as a methodological inspiration to further macro- and micro-research, and as a descriptive formula in certain restricted areas (see 8 and 9 below). 2. Emergence is obviously a common phenomenon. (In the microsubject, the issue is lost in morasses of competing definitions and interpretations.) 3. Definition of key concepts like species and function is

191

MICHAEL

SCRIVEN

chiefly by giving examples, contrasts, and indicators rather than necessary and sufficient conditions or essences. (The micro-theorist may and often does continue to believe that we shall one day be able to identify the truly defining properties of each concept in some ideal reconstruction.) 4. Randomness is common in the macro-environment, since it means the absence of regularities which have any causal significance at the macro-level. (A roulette wheel is macrorandom.) Randomness usually occurs in the behavior of macrosystems as the result of random input from some environmental system. We might describe this by saying that it is usually antecedent conditions and not the laws which have an irreducibly stochastic element (macro-irreducible of course). 5. Reducibility of the macro-subjects (like biology) in the sense of their eliminability in favor of micro-subjects (like physics) is pragmatically absurd, since the macro-subjects are identified in terms of macro-problems which are to be solved using the available macro-data. Of course, some macro-phenomena can not be explained at the macro-level, for example, the law of effect or the connection between the curvature of space and the distribution of matter-and these impose some of the limits on macro-determinism and provide an important reason for continuing some reducibility research. But the macro-subjects exist in response to an environmental demand which frequently allows far too little time and/or opportunity for access to micro-data, so their macro-defined concepts are necessary, which means their autonomy as subjects is assured. Often this practical demand is not only pragmatically but definitionally connected with macro-variables, as in the request for the historical account of the development of the mammals. 6. Reducibility of macro-phenomena in the sense of their explicability in terms of micro-variables is simply the last resort of whatever truth there is in micro-determinism. It has in fact been achieved on some occasions and may be achieved on many more, but it is chiefly a project for the micro-subjects with the above exceptions. When achieved, however, two points should be insisted on: (i) an insight has been achieved, for we do then understand something in a respect we did not before; (ii) no insight has been achieved with respect to possibly important questions at the macro-level couched in terms that require a macro-answer-an example is any request for historical background, but there are other totally different kinds, such as requests to fit a macro-event into some meaningful macro-pattern. ("That movement is part of a released escape mechanism"-M. B.'s example.) (iii) Both (i) and (ii) are in-

192

Explanation in the Biological Sciences dependent of the question whether new and valuable macroresearch has been suggested by the micro-theory which gives insight. 7. Reducibility in the sense of conceptual eliminability or the "defining-away" or "translating" of macro-concepts is nearly always impossible (except in the almost irrelevant sense of material equivalence); but then it has, I think, never been achieved nor would it had been very valuable in any field of science. 8. The most important explanatory concept at the macrolevel is that of cause, and it is a completely autonomous macro-concept. That is, it has no logically necessary microcommitment at all (for example, to the existence of intervening links between spatially or temporally separated causes and effects). Of course, the success of micro-theories in general makes it plausible to look for micro-equivalents of macrocausal claims, in terms of which micro-equivalents it is sometimes possible to answer other and interesting questions (about the mechanism of causal influence for example). But this in no way alters the fact that the account in terms of macrocauses is (a) a complete and correct answer to important scientific questions, (b) discoverable without any micro-investigation. Before going into some detail about this and other points, let me complete my list of macro-claims by mentioning the most important consequences of this one. 9. a) The information which enables us to identify the cause of a macro-event beyond any doubt is often entirely inadequate for predicting that same macro-event prior to its occurrence. (Radiation sometimes produces a viable mutation, but you have to bet that it will not on any given occasion.) b) The information which allows a prediction of a macroevent is often entirely inadequate for its explanation at the macro-level. (A mere correlation is not explanatory, but is predictive. ) c) The information which allows a micro-explanation of a macro-phenomenon is often entirely inadequate for its macroexplanation, as in the case where the required macro-explanation is historical and refers to the unique features of the event. d) The same point as (c) applies to micro-prediction but for a relatively trivial reason and under restrictions. To sum up point 9, explanation and prediction are essentially different processes practically as well as logically; and the macro- and micro-levels are essentially independent levels with respect to explanation, and to a lesser extent, prediction. 10. The conceptual apparatus of methodology-the concepts

193

MICHAEL SCRIVEN

of concept, law, model, theory, hypothesis, and so forth-must be defined in macro-appropriate terms for macro-use, and that means abandoning any Nervous Nellieism about their approximative status. Almost no physical laws are literally true, especially those which are precisely expressed, and any useful generalization at the macro-level which needs a convenient name-tag can justifiably be called a law or principle unless too weak ("hypothesis") or too strong ("central dogma"). To support points 8 and 9 requires a radical and very extensive reassessment of the whole logic of knowledge, understanding, and inference, which cannot be done adequately in the present paper. I shall try instead one simple line of argument, hoping this will give some assurance that an alternative line of approach is possible. I propose to argue that (a) explanation and prediction are connected in the sense that both processes often draw on some of the same data, although they use them in different ways and often use and need other data; (b) a cause is simply a single-factor explanation. The notion of "the cause" is therefore appropriate in just those cases where a simple, single-factor explanation is appropriate, that is, roughly those cases where the inquirer is looking for and there exists one missing link in the general picture that he has of the phenomenon in question. The two main "axioms," of widely different kinds, on which I base the theory of explanation that leads to these conclusions, are as follows: 1) An axiom from pragmatic logic. Explanation is an essentially pragmatic rather than syntactical or semantic notion. In particular it is highly context-dependent, in the sense that a proper answer to a request for an explanation can only be identified by means of cues provided by the context, for example, the level of knowledge of the inquiry (or, in particular, that of an inquirer), and the type of interest that he has. The notion of "the scientific explanation of x" must be regarded as an abstraction from the basic notion of "explaining x to y" via the notion of "the explanation of x for y." 2) An axiom from the evolutionary theory of cognitive processes. The need for understanding and explanation transcends but springs from the need for knowledge. It is a further requirement for effective adaptation of an information-processing system in an environment providing it with more information than it can efficiently handle if it employs the simplest sequential-storage, total-search-and-retrieval system. Let us begin by identifying the typical questions that call

194

Explanation in the Biological Sciences for explanations and predictions. It has been wrongly suggested that explanations are always or nearly always responses to "why" questions. They may in fact be responses to questions of almost any form, for example: What is the nature of this complex process? How did this result come about? Where and in what way do the waste products get treated in this system? Indeed, it is probably just as satisfactory to think that "why" questions can be translated into these terms as the reverse. Requests for predictions, on the other hand, come from a completely different family; for example: What will be the future development of this life-form? Will it be 6 feet tall a year from now? Where will Mars be next April 9th? When will the next solar eclipse occur at San Francisco? How will the Fifth Republic end? To put the matter in its simplest terms, prediction require us to say something about the nature of the future, whereas explanations require us to identify a feature of the past that bears a special relationship to the present. Speaking atemporally, the simple fact is that any kind of empirically reliable correlation between events is an adequate basis for prediction, but explanations require a special kind of connection, such as causal. Whereas any kind of nonsimultaneous correlation will give you some kind of prediction, if you want to predict where Mars will be next April 9th, then you have to operate in a more directed manner. You work from the orbital equation for Mars and the antecedent condition of its position at a certain time to generate the required position. This is very like a theorem-proving program in the computer field, where we specify the task in a certain way, as, for instance, by asking the computer to prove a theorem about a certain relationship, to prove or disprove, for example, the collinearity of three specified points in a certain family of triangles. The program begins with certain axioms, perhaps with intermediate theorems, and tries to apply rules of inference to them according to certain heuristic instructions until it generates a result which has the required formal properties to within a single application of the negation operator. On the other hand, the kind of activity involved in the search for an explanation varies a great deal more, depending upon the particular level and type of explanation that is being sought. The computer must first infer or be given data about level and type. Then, in response to a request to explain the nature of a complex process (such as a tribal activity to an anthropologist unfamiliar with the tribe in question, or the

195

MICHAEL SCRIVEN

operation of a complicated distillation plant), the prime task of the explaining program is to identify in this process some already understood patterns, that is, patterns which it is reasonable to infer that the inquirer will understand, and attempt to tie together a set of these with understood links so that they completely describe the observed phenomenon in the required degree of detail. The simplest kind of request for an explanation-for example, that posed by the police to the coroner-calls for a single cause, e.g. of death, and here again a pattern recognition activity is involved. The explainer examines the circumstances and compares them with his repertoire of known causal patterns until he obtains the best match. An intermediate case is involved in responding to a question of the form, "How did this come about?", where one is expected to produce a rather long but simple chain, as in answering a request to explain the emergence of the black nations as a signfficant world force in the latter half of the twentieth century. Here we have a recognition and construction task, the over-all requirements of which are very similar to those of the complex prediction type, if described on a sufficiently general level. But the moment we get down to looking carefully at the two tasks, we notice that in the prediction task we are restricted to using patterns of a particular kind, those with "projective reliability," that is, those with highly stable later portions tied to a highly specific early portion. For we are trying to build up a bridge to a future event described in informative detail, with a degree of accuracy that mere guessing would not lead us to achieve. This is the sole dimension of merit in a prediction-the degree to which it is precise or informative with respect to the required prediction-request. But in building the connecting links between one or several causes and their effect, we are not at all concerned to show that the effect was the only one that could have come about-our task is rather to show how it did, in fact, come about, since we already know that it did. That is, we only need patterns with good fit to a completed process: no projective reliability is required. On the other hand, the acceptable patterns must meet another criterion; they must themselves be comprehensible. This does not just mean familiar, or true, though it is connected with these concepts; it means, roughly, reducible to certifiable models where "certifiable" means that their consequences are known (to at least a considerable depth) and true. So we see that there are affinities between the prediction task and the explanation task which are related to those be-

196

Explanation in the Biological Sciences tween theorem proving and pattern recognition tasks, but there are also crucial differences. Without going into great detail, I shall now describe one particular case that bears on claim 8. This is what I call the basic statistical case of causation. In this case, the following conditions obtain: a) The effect Y never occurs except in circumstances where the factor X is also to be found. b) In those cases where the factor X is introduced randomly, as well as where it is spontaneous, a certain small percentage are observed to also include the factor Y. No scientist would hesitate to say that on the occasions where both X and Y occur, X is the cause of Y. I say that a scientist would not hesitate to describe this as a case of causation with some confidence because it is a common case in quantum physics, and is there so described. But its importance has not been recognized by scientists in general, or philosophers of science in particular, who are still committed to the view that there cannot be any definite undeniable cases of causation in an individual situation if there is no underlying determinism. The case just described can be weakened to allow a spontaneous base rate for Y, to cover the case of several independent possible causes of Y, etc. In all of these situations, a substantial amount of causal vocabulary can be retained if certain other experimental results are known. Some philosophers faced with considerations such as the above have been led to suppose that there is an analysis of causation which would handle them. This would equate a cause with a probability-modifying operator. Unfortunately, this is both too restrictive and too inclusive a definition. It is too inclusive because the symptoms of a disease are probability-modifying operators indicating the presence of the disease but not causing it; it is too restrictive because of a family of cases which are also crucial for certain other analyses in the concept of cause, such as cases of over-determination. These are cases where there is present in a situation more than one factor or group of factors which are independently sufficient to bring about the effect. Since it is possible to predict the occurrence of the effect with probability from any such group, and since in such circumstances it is frequently true that only one of them is, in fact, the cause of the effect occurrence, it must be false that a cause is a proba-

197

MICHAEL SCRIVEN

bility modifier. These basic cases are, of course, crucial to one or more of the earlierclaims, particularlyclaim 8. To conclude, then, I have proposed that we should develop and employ a macro-philosophyof science just as long as we find it appropriateto employ macro-sciences. The day when chemistry, biology, and psychology are reduced to atomic physics it would be quite possible that we should reduce macro-philosophyof science to micro-philosophyof science. It might be appropriate at an earlier date, or a later one; but until it can be demonstrated that there are greater advantages to talking always in terms of what is "in principle possible," then I would be inclined to think that we should stick as closely as we can to a description of the methodology and logic of science which is as closely tailored to its reasoning and procedures as its own concepts are tailored to its phenomena.

198

OrganizationalLevelsand Explanation CLIFFORD GROBSTEIN

University of California, San Diego School of Medicine

La Jolla, California

The problem of levels is especially critical for certain areas of biological investigation, particularly that of development. The question is posed whether such biological phenomena require special forms of explanation because the phenomena themselves are hierarchically organized. By explanation I mean explication of the properties of a phenomenon in logically interrelated and substantively relevant statements. By "hierarchically organized" I refer to the fact that biological phenomena can be dissected conceptually and physically into subsets which are again dissectable into subsets (and this several times repeated), with each subset having a reasonably integral set of properties. The question of whether this requires explanatory principles that are fundamentally peculiar to biology is, I believe, to be answered negatively. However, the question of whether there are special difficulties to be taken into account and special conceptual problems to be solved, I believe must be answered positively. The special difficulty is the relationship between properties at successive subset levels. The special difficulty is heightened by the fact that biological subsets are not the same in isolation and in combination, and the difficulty is compounded when new sets and subsets are appearing during the process under investigation with emergence of new properties within a set. The problem was classically presented in the phenomenon of development, where the seeming miracle of emerging order has confounded biological thought since Aristotle. We are no longer confounded, though the miracle is not yet thoroughly illuminated. What we now see is a process involving conversion of properties to successively more inclusive sets. More substantively, we see information contained in the linear sequence of certain macromolecules transformed into spatial, temporal, macromolecular patterns, and these in turn into cell behaviors which give rise to tissue organization,

199

CLIFFORD

GROBSTEIN

organ formation, and ultimate organismic and supraorganismic order. Developmentally emerging order involves continuous translation of the properties at lower levels to higher, with subsets interacting with each other even as their properties are interlocked in some sense with properties of sets above and below. What are the rules which interlock properties of subsets, sets, and supersets? Is a new form of explanation required to comprehend this? Is there a miracle of indeterminacy at any step in the process? Is the notion of emergence of new properties a figment of vitalism, or is there a valid phenomenon to be subjected to causal analysis? This is the way I see the substantive problem which we are forced to deal with in development. It is essential to understand the translation rules between successive levels, because such translation is a mechanism of development. Let me illustrate the problems with a particular case. The development of the kidney is almost diagrammatic in illustrating the situation. Let me state certain facts about the phenomenon and then return to the more formal aspects. The adult kidney of the mouse is an intricate structure; it has a very complex form and function that are essential to the life of the whole organism of which the kidney is a part. We may say that the organ is a subset of the whole organism, which in turn can be dissected into subsets called glomeruli, secretory tubules, collecting tubules, and so on. These parts are normally in a characteristic array and each can be further dissected into subsets (the cells) which again are in characteristic array. The cells too can be dissected into organelles, and these into macromolecular arrays. The important first point to make is that this impressive complexity has been reproduced over successive and countless generations of mice in essentially identical form. The elaborate duplicative performance is a regular and predictable phenomenon, which appears to occur de novo in each generation. However, kidneys do not produce kidneys. Rather, the mouse produces an egg, the egg an embryo, and within the embryo the kidney appears from nonkidney. Our problem is the genesis of this intricate order. Confessing immediately that we cannot fully explain this genesis, let me state some facts that give hope that ultimately we will explain it, and without qualitatively new forms of explanation. I have noted that successive generations of mice produce essentially identical kidneys. There are, however, significant exceptions. There is a strain of mice which on suitable breeding yields offspring which are abnormal in a number of re-

200

Organizational Levels and Explanation spects. Included among the abnormalities are defects of kidneys. The breeding behavior indicates conclusively that there is a relatively simple genetic factor involved in the kidney genesis. This is no surprise, in view of our general knowledge of the role and mechanism of action of hereditary factors in development. We summarize this by asserting that there is a replicative program transferred from generation to generation, and that within this program there is information whose proper interpretation leads to kidney formation. Localized alterations or defects in the program lead to developmental defects of the kidney; hence there must be particular portions of the program which have special relevance to the kidney. Whatever the mechanism of registry of the informationand I think that very few biologists today will doubt that it resides initially in important part in base sequence in DNA -all of the information is contained in the kidney rudiment of an 11-day embryo. I mean that one can remove such a rudiment from the embryo and this isolated subset under suitable conditions will continue essentially its normal development: the rudiment contains all of the necessary information for completion of the complex, intricate genesis. The sorudiment itself is dissectable into two components-the called ureteric bud and the surrounding mesenchyme, and each of these is a cellular system. A whole rudiment placed in culture will continue reasonably normal development. Separated components, however, do not, although they do if recombined. Clearly some interaction of the components is required. At this stage the genetic information, initially present in the egg in the form of linear sequences in DNA, has been processed sufficiently so that neither component without the other has an effective program to continue development. Something exchanged between them (and it can be shown to be separable from cells and to be macromolecular) is essential to continued expression of the genetic program. The exchanged material is known not to be DNA itself, although we do not know exactly what it is. Since it contains essential information for continued development it must be a translation product of DNA, at least in the sense that there is some systematic correspondence between its properties and those of DNA. It might, for example, be a protein whose amino acid sequence is determined directly or indirectly by the base sequence of DNA. This would conform to what is now regarded as the fundamental translation step in genetic read-out. Let me add that I refer to it as protein for simplicity. It could equally well be RNA, and the same argument would apply.

201

CLIFFORD GROBSTEIN

To relate the material to the emerging order of kidney development we must describe some further biological facts. If we treat the kidney rudiment properly we can dissociate it without significant damage to a suspension of individual cells; that is, we can progress in our dissection from rudiment to subset components and then to subset cells. These interspersed cells will recohere into components and these in turn will interact to give continued kidney development. The separated cells contain information that enables them to reform the next higher level, which involves components having properties that can interact to continue the development. Analysis shows that the cells have surface properties which cause them to sort as to kind through differential adhesion; the surface materials involved are macromolecular and at least in some instances separable from the cells. The macromolecular materials promote aggregation and contain the recognition specificity. They appear to be similar in general chemical type to those referred to earlier as exchanging between the components. Moreover, if one examines the development of the kidney mesenchyme into tubule rudiments one notices that cellular aggregation is one of the first consequences of the action of one component on the other. Cells within the affected mesenchyme cohere more tightly to form packages, or aggregates, and these gradually transform into tubules. The first step in the formation of the new subset is cohesion to form a package, and this step is promoted by exposing dissociated mesenchyme to macromolecular extracts of tubule-inducing sources. Hence, assembly of supercellular systems appears to be dependent on macromolecular materials associated with cell surfaces. Tubules emerge in combinations of suitable cells and macromolecules. It is well known that complex macromolecular pattems can self-assemble from dissociated molecules; that is, what we have been talking about at the cellular level also occurs at the molecular one. Native collagen, for example, can be solubilized in dilute acid so as to completely lose its fibrous character. On return to neutrality, the still intact collagen monomers reassemble into fibers and these can be shown to have a characteristic ultrastructural banding pattern whose detail can be altered by the conditions of the reassembly. Depending upon relatively simple conditions the reassembly can duplicate or alter the original macromolecular assemblage. It happens that the detailed structure of the reassembled material is important in development, because, if it is normal, calcium will deposit on the collagen to forn normal bone. If

202

Organizational Levels and Explanation the substructure of collagen is not normal, deposition of calcium is not normal and bone formation will be defective. Thus large molecules specified in their linear sequence by the linear sequence of DNA can self-assemble in arrays and these in turn can influence cellular assembly. The emergence of tubules in the developing kidney is an epigenetic conversion, in the proper context, of properties at the molecular level. I am not anxious to press the point too far as to detail because there are still large gaps in our knowledge; rather, what I have been sketching is a somewhat impressionistic example of what most of us now see as the essential course of development. It is a series of translations of information stored in chemical terms, but read and registered by scanners of increasing complexity as the developmental course progresses. The process is frightening in its intricacy but it does not appear to contain elements requiring new intellectual feats to comprehend. The basic translations may be broadly indicated as: micromolecule-*.macromolecule1---polymer----r ultrastructural array-ocell-*tissue-, organ->organism. I have tried to make it clear that we see ways in which cellular behavior is altered by the specificity of molecules, in turn laid down by genetically transmitted chemical information. Cellular behavior, like specific aggregation and specific synthesis, leads to specific structure and activity in cellular systems, and interaction among these leads to specific behavior at the level of the organism. It is clear that we move upward in levels with conversion of information from one level to the next. The rules, however, we only partly understand. Moreover, I have emphasized only the outflow of genetic information. There are a whole series of feedback operations as well which I have not touched on. The issue posed, however, seems to me to be worthy of joint consideration by biologists and philosophers. Are there general rules by which properties at a given level are summed, integrated, and transformed, so as to emerge into properties at the level above? To illustrate what I am asking we can write the relationship as follows (S = set; [] = combining relations ): SI

S2*-

+Sn

[S-

+ S2 + *

+ Sn]=S

The issue is the relationship between the properties of S. and the properties of any one of its subsets (P..). There are a number of

203

CLIFFORD

GROBSTEIN

cases of increasing complexity which can serve as models in trying to make some sense of the rules of conversion. The simplest case may be referred to as "collections." Professor Scriven made reference to sand dunes, and I will also use these as an example. In this case the properties of subsets are identical whether they are within or outside of the set. In other words, it makes virtually no difference in the properties of a sand grain whether it is by itself or whether it is within the sand dune. One can further define a collection as the case in which association is fully reversible; you can pile up the sand or you can scatter it, and the grains are otherwise unaffected. What is the relation in this simplest case of properties at the level of the individual sand grain (subset) and the level of the sand dune? One can make certain obvious statements; for example, that the mass properties of the sand grains are summed in the properties of the dune. Further elementary interactions do go on between grains, and these lead to characteristic packing of the sand and hence to dune properties. There are more complex effects as well. If one examined the temperature gradients in the dunes in the early morning or late at night, I suppose that the gradients from the exterior to the interior would be different depending on the size and shape of the dune. This is a mass property of the whole dune which is not readily predictable from properties of grains. Beyond this, there is the effect of context and vantage point-the sand dune has different significant properties depending on where we are and where it is in relation to other objects. The dune does not change at all, but its properties are defined very differently from the inside, from immediately above, or from 2000 feet above. This is not an inconsequential matter because there is an often unappreciated vantage-point effect in looking at more complex systems like biological ones. Even this simplest case of collections shows a striking new dimension when heterogeneity becomes a factor. As soon as more than one kind of subunit is involved in the collection, a whole new range is opened in terms of higher-level properties. Differential distribution of kinds, as happens when wave patterns move over the beach and different colors of sand grains are sorted out, produce patterns-properties which are predominantly relational and relatively independent of unit properties. One can say a good deal in this relatively simplest case about the relationship of the properties at the upper and lower levels, and the kinds of rules which relate them. This becomes much more difficult when the case gets even a little more complex; for example, when there is some kind of co-

204

Organizational Levels and Explanation herence among the subsets. Assume that the only differences in the properties of a subset are those immediately resulting from coherence with a neighbor subset. Assume further that coherence is not fully reversible and that the coherence is limited to a particular point on the subset. Packing of subsets is immediately strongly influenced, and patterns of coherence appear (crystallization). Now assume a second kind of complication, that coherence involves affinity, for example, complementarity between antigen and antibody or between enzyme site and substrate. Relationships depend now on properties stemming from conformation; that is, pattern itself becomes a source of binding force. Real complexities obviously occur, however, when elements moving into a set are strongly altered in their properties by the relation itself. A whole new level of transformation is introduced. The kidney behavior I referred to is a case in point. The two components are altered very importantly in their properties when brought into interaction. Such cases may be reversible: the properties are altered within the set but return to initial state on disassociation of the set. The slime mold, for example, moves sequentially from a higher ordered aggregative state to a lower ordered disaggregative state. It is only in the higher ordered aggregative state that one finds significant biological differences of properties between the cells. Such a system may be called facultative in that interacting elements change their properties in interaction, but can very readily be returned to initial state. The case that presents most of the mystery is the obligatory system, the one in which interaction has reached the point where subsets cannot, without very great difficulty, be induced to show their characteristic properties because these are "bound" to the set. The requirements for the characteristic properties of the subset are intricate and subtle, and very difficult to reproduce outside of the set. Cells of multicellular organisms belong here; they have only recently and incompletely yielded to the blandishments of controlled environmental conditions in vitro. In such obligatory systems one nearly always finds an internal environment which is controlled and separated from the external environment so as, in a sense, to acquire a life of its own. The internal milieu begins to develop properties which are partly self-controlled, not independently of the cells but not fully under their control. This kind of phenomenon occurs not only in complex organisms with respect to cells, but in ecologic systems like kelp communities where a controlled

205

CLIFFORD GROBSTEIN

environment is set up with the kelp in which only certain kinds of organisms will be found. The combining conditions at this point acquire their own dynamic properties and their own stability. This means new properties in the total complex resulting from this controlled environment. The objective of this discussion has been to pose a question. Is there a useful formal approach to the problem of the conversion of properties from one level to those of another? The question is no longer idle for there is a wealth of new data in the last few decades stemming specifically from the success of the reductionist approach. The reductionist approach has multiplied problems for the constructionist approach: how do we put things back together again conceptually, what are the rules operating within the whole system that are abrogated in the process of reduction? Can we look at vertical transformations of properties in simple systems and generalize rules that will allow us to "see through" more complex systems? If there are general rules, do they constitute a new kind of explanation? I think not, but in part this may be a matter of definition. Perhaps, again, we deal with levels. Perhaps it is a new level of explanation which emerges from the interaction of explicatory process with the complexities of hierarchical order. It is not new in kind, but it has a new power. Perhaps it will prove to be the explanatory counterpart of the enormous biological power of hierarchial organization.

206

Commentary-Part II

EDWARD MANIER University of Notre Dame

Michael Scriven has raised a most fundamental question concerning the philosophy of science: Can methodological studies succeed in both of the following tasks? 1) the analysis of actual scientific practice; 2) the formulation of criteria for the evaluation of scientific progress. The tasks are strikingly similar to those confronting the evolutionary biologist who seeks to develop an empirically significant account of progress in organic evolution. In line with the first task, Scriven proposes a "macrophilosophy of science," which is "phenomenological," or "phenotypic," in its own approach to science. It is as if the philosophy of science were confronted with a given: the verbal behavior of the various sectors and levels of the scientific community. If I am correct, the first constraint which must be met by the philosophy of science as Scriven has described it is that of fidelity to the actual diversity of theoretical perspectives and methods within biology. This diversity is schematically represented by the polar relation of the idiographic-historical methods of ecology and evolution on the one hand, and the nomothetic aspects of molecular biology on the other. Within the boundaries thus determined, Scriven seeks to undercut such questions as that of the "in principle" reducibility of biology to physical theory, admitting that an insight is achieved by the explanation of, for example, the behavior of an anchovy in terms of micro-variables (molecules or nerve pathways) but denying that this provides any insight into "possibly important questions at the macro-level," such as those dealing with the evolutionary development or ecological context of the anchovy's behavioral repertoire. Presumably, a macro-philosophy of science would illuminate and not eliminate the diversity of such insights. Scriven offers at least two reasons for pursuing macro-philosophy of science (while not denying the intrinsic merit of its micro-counterpart:

207

COMMENTARY-II

devoted to the ideal of a unified science). First, macro-philosophy is the more challenging and valuable enterprise. "The real question is whether any straight philosophers of science ever say anything that leads their listeners to improve their practice as scientists." This pay-off is more reasonably sought among the results of a philosophy of science "defended by its relation to 'straight' science, or 'science proper . . . rather than as what people called philosophers of science do." Second, fundamental philosophical considerations (from pragmatic logic and the "evolutionary theory of cognitive processes") support the principal tenets of Scriven's macro-philosophy of science. As an essentially pragmatic notion, explanation is relativized, not only to specific scientific contexts, but to specific investigations or investigators. This is supported and expanded by a highly schematized "evolutionary" axiom expressed in terms of the effective adaption of the "simplest" information-processing system in an environment which provides it with more information than it can efficiently use. (Scriven does not indicate whether survival, surviving progeny, or some other criteria can be used to mark the effective adaptation of an information-processing system.) We might sustain this last metaphor by noting that macrophilosophy of science deals with natural populations of scientists rather than the logically simple but fictional population of ideal scientists. Nevertheless, Scriven is also committed to the development of what he calls a "normative" philosophy of science. The critical relation of this philosophy of science to science itself is not direct; it evaluates accounts, theories, and descriptions of the logical features of good individual scientific explanations. These considerations exhaust the province of the philosopher of science, but they do not exhaust the features which account for the "effective adaptation" of a "good" scientific explanation. This leaves the project for a macro-philosophy of science which is at once faithfully descriptive and critically and actively normative with two unfinished tasks. First, a philosophy of science which is based on the view that explanation is essentially context-dependent must give an account of the means by which it could hope to defend a position critical of the theories and methods prevalent at a given time in a given area of science. Since what counts as an explanation may vary with contextual cues and the level of interest of the investigator, the grounds for independent criticism must be seriously reduced. Scriven assumes that such grounds are available,

208

Commentary-II but his presentation includes no illustration of their employment on any specific item in the biological literature, let alone the preferable account of their general availability. The point is simply this: if explanation is context-dependent, the door is open for the evasion of all criticism via the claim that the putatively aberrant features of an explanation are requisite for the context in question. The second problem is one which confronts Scriven and every philosopher of science who admits that the features of a "good" scientific explanation cannot be settled by logical criteria alone. This is the problem of providing a method for the straightforward comparison of competing scientific perspectives, and arriving at some measure of their scientific merit. There is no a priori guarantee that in any specific case the palm will be awarded on logically or philosophically assessable grounds. This renders the subject matter of macrophilosophy of science extremely tenuous-so tenuous that such a philosopher would be well advised to double as a historian in order to avoid the recurrent frustration of analyzing a scientific episode only to find that the crucial factors were changes in experimental apparatus or in experimental subjects (for example, the shifts to Drosophila, then Neurospora, and then to E. coli as subjects for experiments in genetics). Such shifts are obviously of tremendous significance in the history of science; but it is difficult to imagine the logic or the epistemology subtle enough to make much capital of them. Scriven's difficulties in this regard may be seen more clearly by reviewing his own earlier analysis of a specific problem in evolutionary biology: the meaning and logical status of the familiar slogan "the survival of the fittest." 1 Scriven decides that "fitness" is too ambiguous to allow expressions containing it to have definite predictive significance; nevertheless, he claims that the slogan has limited empirical significance as an "explanation indicator." It emphasizes the reasonableness of attempting to explain the occurrence of a flourishing population in terms of some variant of the concept "biological efficiency," and not in terms of mere chance. Scriven accurately lists the sources of the ambiguity of "fitness": a) the tension between specialization (stability) and adaptability (variability); b) the probability that the descendants of a small or decreasing population will represent a skewed sample of the ancestral gene pool; c) the effect of random mutations upon the biological prospects of a popula1. Scriven, "Explanation and Science, 130 (1959), 477-482.

Prediction

in

Evolutionary

Theory,"

209

COMMENTARY-

Il

tion; and d) the unspecifiable effects of random pre-adaptation. Nevertheless, he fails to recognize that the specific theory he is commenting upon is aimed at reducing the ambiguity which results from the factors mentioned. J. M. Thoday has sought to illustrate, by both experimental evidence and conceptual models, the genetic mechanisms required for the resolution of the antagonism between genetic stability and genetic variability.2 Scriven reads this article as if it maintained that the antagonism is altogether irreducible: "maximum specialization is incompatible with maximum flexibility." Thoday in fact holds that while it is impossible to generalize concerning the fitness of any given phenotypic trait, it is reasonable to stipulate that a population of organisms will show an increase in fitness when the sum of its genetic stability and its genetic variability shows an increase. Such increases can be achieved if the organisms involved exhibit genomes possessing balanced heterozygosity, according to Thoday. This is a condition in which a gene on one chromosome is paired with a distinct allele on the matching chromosome capable of sustaining similar functions under different environmental conditions. Organisms with such a buffered system can exhibit "developmental flexibility," that is, individually adaptive response to various forms of developmental stress. Simultaneously, balanced heterozygosity maximizes the possibilities of genetic variability under the ordinary mechanisms of recombination. Thus one and the same type of genome equips a population to meet the ordinary demands imposed by the variability of its natural environment and to transform itself through successive generations in the face of the demands of sudden and drastic or long-protracted environmental change. This example casts some doubt on Scriven's thesis concerning the distinction of explanation and prediction in science. He intended the example to show that scientific theses which lack predictive value nevertheless may have explanatory significance. In fact, the example suggests that explanations of evolutionary history which refer to organic fitness, broadly understood, may succeed in fitting a particular episode into a general picture in a satisfactory way and yet prove unacceptably fictional when an effort is made to specify the meaning of "fitness" to be used. The required judgment can hardly 2. J. M. Thoday, "Components of Fitness," Symposia of the Society for Experimental Biology, 7, 96-113; "Natural Selection and Biological Progress," A Century of Darwin, ed. S. A. Barnett (Cambridge, Mass.: Harvard University Press, 1958), pp. 313-333.

210

Commentary-II be made unless "fitness" is specified in some fashion open to predictive, experimental testing. In spite of the fact that I have yet to see the enterprise successfully executed in the philosophy of biology, I am in agreement with Scriven that the philosophy of science must focus upon problems crucial for the historical development of the science under study. It is appropriate, I think, to examine Morton Beckner's account of function and teleology in this light. Beckner offers a comparative analysis of three statements: 1) A function of the heart is to pump blood. 2) A function of the heart is to produce heart-sounds. 3) A function of the earth is to intercept passing meteorites. Beckner indicates that the first is true, and the second and third either false or meaningless, or inappropriate in some other sense. He explains this distinction in terms reminiscent of C. H. Waddington's account of organization. The degree of organization is the extent to which the entity is internally coherent, in the sense that the properties of its parts are essentially dependent on the relations between the parts and the whole. The relation between the part and the whole is spoken of as the function of the part.3 Beckner speaks of the parts of a system being identified in terms of the same conceptual scheme which enables the identification of the system's principal activities and of the "definitional contribution" of the functional parts to those activities. It cannot be denied that such discussions are relevant for the work of Waddington and many other biologists. But it can be argued that their relevance is at the most general level when one is seeking those features of biological studies which distinguish them from investigations of inorganic systems. The relevance of this distinction to progress in biology is open to question. The suspicion lingers that this is a philosopher's issue: "How many kinds of things are there in the world?" It may, therefore, be useful to suggest an alternative approach to the study of functional analysis in biology. Generally, previous discussants of the logic of functional analysis in biology have paid little or no attention to the history of biology or to the diversity of theoretical perspectives within contemporary biology. In consequence, their efforts 3. C. H. Waddington. Unwin, 1957), p. 3.

The Strategy of the Genes (London:

Allen &

211

COMMENTARY-II

produce succinct, lucid definitions which assure us that biological science contains no unwanted metaphysical contaminant, but leave us otherwise uninformed about the logical cohesiveness and experimental suggestiveness of current theory. Questions and statements concerning organic functions occur at various stages in the process of theory construction in biology. In general, biologists ask at least three different types of questions about the functional components of organic systems. 1. Is a functional system of type a to be found in organisms of type x? (Do lower organisms resist infection by anti-body formation?) Or, at the same level, what is the function of the item i in organisms of type x? (What is the function of the thymus in mammals?) At this level, either the functioning part or the functional activity may be observable in some sense. The aim is the initial delineation of functions and functional parts, often by an assessment of analogies or homologies among organisms and their activities. Often this level of analysis contains no theoretical component other than the general assumption that the major activities of living things bear some analogy to each other. (If higher organisms, for example, exhibit resistance to infection, analogues for this activity may be sought in lower organisms as well.) This assumption need not carry evolutionary corollaries with it. If this level of investigation is taken as typical of all functional analysis, such analysis will be regarded as a relatively primitive, nontheoretical, and eliminable phase of biology. But there are other, more advanced variations of this approach. 2. What is the contribution of this functional system to the overall fitness of organisms of type x? How did such a system evolve? (How, for example, could a system of warning cries evolve in flocking birds if the sentinels which make the cries are subjected to heavier predation than silent birds?) Again, the functional activity is given, in some very broad sense, but the question that is raised assumes a particular theoretical perspective. This perspective may vary from the relatively simple question concerning the historical lineage of the functional system to more complex questions concerning the type of selective and genetic mechanisms involved. These questions touch such fundamental issues that it is difficult to see them as corrigible aspects of a preliminary phase of research.

212

Commentary-II Again, if this phase of investigation is taken as typical, credence is given to the view that an organic part or activity can be regarded as functional only if it makes a positive contribution to the fitness of the organism in question. In fact, much more is at stake as biologists seek to unify the evolutionary and the physiological perspectives. 3. What is the mechanism of a functional system of type a? What is its basic organization or design? Under what conditions does it perform at optimal efficiency? Under what conditions does it break down? (What, for example, is the design of the system which regulates the internal temperature of mammals? What causes this system to break down in instances where a fever is produced?) An even more theoretical approach seeks to identify or construct functional systems which could not be given to observation in terms of their status as components of fitness. It has been argued that this approach would lead to an expanded set of functional categories, that-in particular-it would correct a tendency to think of functional systems only at the physiological level where the recognition of the functional unit is facilitated by its mechanical compactness and the relatively short time-interval within which it operates. In this vein, Waddington argues for the effectiveness of functional regulators of the embryological process, and WynneEdwards and Lorenz have called attention to the functional organization of the behavior of whole populations. In each case a sketch of a theory of fitness has been used to guide the search for empirical data which would confirm the presence of functional units not directly observable as such. The concept of the organization of the gene pool of a population, advanced by Waddington, Mayr, Lewontin, and others, is an instance of such interaction between theoretical and descriptive uses of "function." This suggests that our list of questions may be supplemented by 4. What functional systems must be postulated to account for the evolutionary record of organisms of type x? (For example, how can the gene pool of a population be organized so that the population will be more likely to survive a capricious sequence of environmental changes?) It is my belief that the possibilities for biologically fruitful analyses are greatest for that philosophy of science which undertakes the comparative analysis of such questions, the

213

COMMENTARY-II

contexts in which they are raised, and the methods by which they are answered. At least the following is true: if philosophers are unable to develop techniques for dealing with biological questions as specific as these, it is not likely that their contribution to biology or the history of biology will be more than an increasingly perfect expression of the issues of the ancient debate between mechanism and vitalism. DAVID B. KITTS

University of Oklahoma

I should like to comment upon Michael Scriven's remarks concerning the explanation of unique events. If a geologist were asked to explain the deposition of a sand dune, to use Scriven's example, he would respond by offering a set of individual events which occurred before the deposition of the dune and which were supposed to be causally related to it. His reason for supposing that these antecedent events were causally linked to the deposition would lie in the acceptance of general hypotheses, some of which might measure up to the rigid standards that have been set for a scientific law. In any case, the explanation would conform closely to the explanatory model with which we have all become familiar. It has been claimed that all kinds of scientists, including physicists, might respond in much the same way to the request for an explanation of an individual event. What would it mean to explain an event in its uniqueness? It would mean, presumably, to explain each of the properties which taken together were sufficient to distinguish it from all other events. Anything short of this would constitute the explanation of a class of events rather than of an individual event. A physicist would infrequently go so far as to explain the complete set of properties necessary to establish the uniqueness of an event. But often geologists and paleontologists do not bother to do so either. They do have a special interest in inferring antecedent conditions; that is, after all, their way of getting at the past. It is interesting and significant that geologists and biologists are more inclined than physicists to cite a large set of initial conditions in explaining events. I do not see, however, that this fact points to any fundamental methodological difference between physics and the so-called historical sciences. Dr. Lewontin has called attention to what he considers to be the postulatory-deductive character of evolutionary theory.

214

Commentary-II That Darwin regarded the theory as having such a character is clear, I think, from the following passage. If under changing conditions of life organic beings present individual differences in almost every part of their structure, and this cannot be disputed; if there be, owing to their geometrical rate of increase, a severe struggle for life at some age, season, or year, and this certainly cannot be disputed; then considering the infinite complexity of the relations of all organic beings to each other and to their conditions of life, causing an infinite diversity in structure, constitution, and habits to be advantageous to them, it would be a most extraordinary fact if no variations had ever occurred useful to each being's own welfare, in the same manner as so many variations have occurred useful to man. But if variations useful to any organic being ever do occur, assuredly individuals thus characterized will have the best chance of being preserved in the struggle for life; and from the strong principle of inheritance, these will tend to produce offspring similarly characterized. This principle of preservation, or the survival of the fittest, I have called Natural Selection. It leads to the improvement of each creature in relation to its organic and inorganic conditions of life; and consequently, in most cases, to what must be regarded as an advance in organization (Origin of Species, 6th ed. rev., p. 98). Darwin's account contains four fundamental propositions, or postulates: (1) Under changing conditions of life organic beings present individual differences in almost every part of their structure. (2) There is a geometrical rate of increase in populations. (3) There is an infinite complexity of the relations of all organic beings to each other and to their conditions of life. (4) There is a strong principle of inheritance. Darwin seems to have made a claim of a kind with which we are familiar in scientific discourse. It is that if certain fundamental propositions are granted, then other propositions follow from these logically. Darwin's account has characteristics we have come to expect in a scientific theory. But it has been pointed out here that the consequence of the theory is simply that evolution will occur. It does not, it has been claimed, permit the explanation of individual evolutionary changes. As stated by Darwin and Lewontin, it certainly does not. But the theory of mechanics does not permit the treatment of individual cases either unless it is instantiated with particular initial

215

COMMENTARY-II

and boundary conditions. Darwin does not present a set of explicitly stated rules for the instantiation of the theory, but a major portion of the Origin of Species is, in effect, devoted to this very problem. In his 1959 paper on evolutionary theory Scriven claimed that Darwin and the Darwinians had attempted to formulate scientific laws of the traditional kind and had failed. I should like to ask Professor Scriven whether or not he would be willing to count any of Darwin's four fundamental principles as laws of the traditional kind? Could not any of them be used as the basis for predictions? And, turning from Darwin to a Darwinian, would he be able to accept Fisher's fundamental theorem of natural selection as a law, since it can clearly provide the basis for predictions? WILLIAM COLEMAN Johns Hopkins University

It will be obvious that I am speaking as a historian and that my biases are many. I think most historians tend to a kind of systematic despair before the word "reductionism" and it may be that we are not alone in this business. A good part of the problems we have been facing here revolve around the definition of this term. No one yet, it seems, has told us what we must do and what we must find; that is, how should we define the reducing science? What is the reducing science? We have all said physics and chemistry; I for one do not understand what we mean by this, simply because as a historian I find chemistry and physics both to be very shifting grounds. It might even be possible that they are both replaceable, that there might be other more fundamental sciences. The historian's view seems to me to have been conditioned by the kind of reductionism which came to the fore in the midnineteenth century. This is usually referred to as materialistic monism and is associated with a brilliant school of physiology. It postulates simple and largely unspecified matter and motion as the reducing science. This is usually a fairly clear statement on the part of the historian. We saw an example of this in the latter part of Frederick Churchill's paper where he points to the strongly reductionist views of a man like Driesch and to Driesch's disapproving recognition of a homogeneous material constitution of the nucleus and its inadequacy as an explanation. The historian in biology has therefore adopted materialistic monism as the science toward which his science can be re-

216

Commentary-II duced. In the abstract sense, this is perhaps not desirable. Given the problems which he himself is dealing with, it may be acceptable for the given period, the given problem, and above all for the context with which he as an historian is dealing. He should not overlook, however, the possibility that contexts vary and other alternatives are available. One, available in the late nineteenth century, was that based upon the Kelvin-Maxwell conception of the vortex atom, a conception by which one can virtually reduce matter itself to a particular dynamic system in the ether, the ether of course being stated to have very few properties which are relevant to traditional reductionism. You have then a different set of terms to which you can reduce phenomena, and there are, of course, biologists who can play this game to a certain extent. This particular approach disallowed its base in traditional materialism, and yet I think it is still reductionist. The point to be taken from all of this is that the historian, when he talks about reductionism, is making ontological statements. There is definite content to what he means by reduction; it is to express something in very particular and tangible form, if mental construction can be regarded as tangible. If we turn to the terrain of the philosophy of science, I think we find a quite different sense of reduction. It certainly is not the sense the historians use. If I understand this rightly, the philosopher of science is dealing with what we can call logical reduction, the reducing of phenomena or, better, the explanatory system of phenomena, to a separate science through explicit logical procedures. When all these terms are carefully stipulated, then we have what we might call a reduced science. I think the whole thing probably carries on by a varying allegiance to the Hempel-Oppenheim model of explanation. Given this fact, I see here a vast source of ambiguity, if not simple contrariety, because the historian and philosopher are talking about different things when they use the term reduction. The historian is admittedly simplistic, perhaps naive, but he is trying to be honest to past situations. I personally do not believe philosophers really seek this same degree of truth to historical circumstance. The historian has in mind quite specific terms toward which he will reduce a phenomenon. Unfortunately, he usually takes a view of his reducing science which makes that science more or less an unvarying entity. The philosopher, in turn, focuses upon necessary conditions and operations for a logically sound reduction and, let the reducing science be what it may, one really does not care. This leads to a view-probably disputable but nevertheless widespread-that the philosophy of science is a

217

COMMENTARY-lI

normative endeavor to reconstruct the disharnony or perhaps the illogicalityof our scientificendeavor. What does the biologist mean by reduction? Here is the real morass; listening to all the various uses of this word here, I do not see any particular or consistent system in terms of which the biologist will express himself with regard to reductionism. Usually we can say it is a term of opprobrium. It is an abusive term that you fling at your enemy. On reconsideration of this, it seems to me that the biologist is probably closer to the historian's view; that is, he is taking possibly defensive measures against the disappearance of his discipline in physics and chemistry. But the defensive measures are largely explained (excusing the nervous qualities surrounding them) by the kind of physics and chemistry which is being specified. Clearly then the whole business of reduction is quite meaningless without total specification of the reducing science. It too is a variable. Not only is the science to be reduced varying, developing and shifting, but so is the science toward which we hope to reduce it. When we look at the instance cited by Dr. Grobstein-with the amino acids in a base pair system-we now have a new chemistry toward which we can direct our biological explanation, and we should focus on this word "new." If our meeting were being held in 1938, we would not have had this chemistry. One need only look at Joseph Needham's book of 1936, Order of Life, to see that discussion then focused mainly on induction and organizers, Speman and Driesh, and that Needham's prescience included a full discussion of structural chemistry. This was not, of course, the chemistry of DNA, but it was the chemistry which then seemed alone able to handle problems of development. We ought to ask as well what is to be gained by the biologist if reduction were to be achieved? Is it useful, is it satisfying, that is to say, why should we be pleased with ourselves if we can state this biological problem in terms of physics and chemistry? To use the terms "useful" and "satisfying"might just be postponing the analysis, or it might very well be telling us far more about biologists' actual practice and let normative requirementshere be damned. There is an obvious gain in some areas-certainly molecular biology; there are other areas, however, in which the traditional view of reduction has, I believe, simply no relevance. Among these are, for example, Beckner's escaping anchovyleaping throughthe water, a sheepdogherding its sheep, the destructionof life in the later mesozoic. The relevant point is this: the biologist as biologist (and I hope the phi218

Commentary-II losopher of biology too) faces given problems, and in facing these he need not always go back to first causes, those which implicitly he believes complete reduction to physics and chemistry would provide. But he should use the reductive standard when the phenomena require it. I see no evidence of argument which forbids biology from being reduced to physics and chemistry. I see no reason whatsoever for saying that this could not take place. We could, for example, take the loose definition Putnam gave me for reductionism-anything which does not contravene physical laws. This opens the door to practically any approach we want. In sum, then, I fail to see the value of a reduction to physics and chemistry, unless one's problems and one's understanding and definition of these problems are so cast as to benefit by such reduction. If reduction becomes merely an interesting exercise-but always a threatening challenge by its radical advocates-reductive failure need not seriously distress us. The whole question of reduction is one that revolves around a problem which the biologist is not alone in facing. Some problems can be better resolved this way (reduction) to the satisfaction of biologists; other problems simply cannot. But there is a converse to this which should be emphasized by the philosophers. I see no real grounds for the alarms which are raised by biologists when someone says that reductionism is moving in on what they might call "their"terrain. Perhaps this is because I fail or refuse to grasp the bearing of arguments which claim biology is autonomous; I do not think that it is so. A final point. Such a view as this certainly is a statement that is based on the progressive total reduction of biological penomena to a reducing science. Does this not carry the implication that "biology" and "biological problems" are really defined by the phenomena which the biologist confronts and not by the reducing science? Again, who really cares what the reducing science may be? The biologist is concerned with particular problems, and this is what it is that sets him apart. The question is one of his interests. A biologist will ask a particular set of questions about gastrulation, and physics and chemistry may come into it; indeed, any other reducing science may come into it; he will ask other questions of the bearers of hereditary information; he will ask still different questions of the relation between fossil-mammals in the different cenozoic strata; he is going to ask still further questions of the relationship between the faunae on various islands in a large archipelago. One of our participants, Gould, pointed out that there is a large aesthetic element which comes into

219

COMMENTARY-Il

play the moment a biologist turns to a problem. Is it not being somewhat uncharitable to the biologist to say that he will not use any and every tool at his disposal to solve the problem which he kept for himself? Some of these will be physicalchemical, some comparative, others will be experimental in the sense of behavior study. I think that no biologist is going to leave any particular option closed which would help him toward the kind of explanation which he is seeking. Obviously, reductionism is going to be granted a privileged place in this understanding, and of course again we have to specify what we are to mean by reduction. But again, is this all biology? Here I make a major turn in my argument and give you more systematic bias. It does not seem to me that the reductionist analysis really gets to the core of what is interesting in biology. In fact, rigorous reductionist analyses force us to overlook what is really the salt of biology, and perhaps of all science, and that is those particular factors which lead to individual intellectual persuasion and insight. And indeed, I think the level of personal persuasion is more than hopelessly idiosyncratic; it leads on to a conviction on the part of the scientist that he has a particular scientific generalization which he would call explanation. Certainly this was Driesch's view as we see it in the presentation Churchill gave us. There is what we would like to say, preserving our proprieties, an extra-scientific element here. Nevertheless, it is an element which is introduced into scientific discourse. You are perfectly free to call this poor science and obscurantism. But can we really and fairly say that such elements are not primary components of a man's scientific endeavor? I think it is right at this point that the historian and philosopher of science diverge. It is not that they are contradictory, but they are looking for different things and they go in quite different directions. I will even go further and ask, Why aren't these idiosyncratic elements precisely the critical ones in their proponents' enunciations of scientific explanation? That is, although the empirical bases here may be disputable, and the logical structure may be faulty, there is in the mind of the concerned scientist a conviction that he is right, and let his phenomena go as they will. The philosophy of science seeks to be normative after the event and tries to ignore these abundant and concrete instances of prejudice and presupposition in the delineation of a scientific theory. The result, at least from the historian's point of view, is that you get a very tidy structure but one which is quite irrelevant to what we are trying to do. In so

220

Commentary

II

doing, it destroys the very human context of all scientific discovery and probably practice. I do insist that a scientist is a complete man and that when he is trapped in this larger generalization he is caught up with a great many motives and thoughts other than those provoked by phenomena pressing upon him. Perhaps this final statement really evades the issue, for the philosophy of science is concerned with systematization of knowledge and not with discovery. Nevertheless, men do behave as they do and, because they do not follow the usual tenets of logical exposition, I do not see how this necessarily reduces the explanatory power of their work, particularly to those individual men who are involved. At issue, in short, is this word "explanation." We seek its meaning both in abstract terms and in the no less pronounced persuasive power it carries with it. I believe that biologists and philosophers and historians may very well be at cross-purposes on just this point. We are looking for different things, we certainly are finding different things, and it is notorious how little the historian and philosopher of science are in contact with one another.

221

Organism, Environment, and Intelligence as a System JOHN PLATT Mental Health Research Institute University of Michigan, Ann Arbor

Someone has said that biology can be exhaustively subsumed under three headings: evolution, morphogenesis, and thought. Evolution and morphogenesis have already been discussed and so it is a pleasure to turn to the subject of thought. What I propose to do is tell about some findings, attitudes, and insights that come out of work in our laboratory. These are largely theoretical, and they largely deal with models, but it seems to me they have considerable application to the relation of an organism, or more exactly of "a sensory-motor decisionsystem," to its environment. LIMITATIONS OF AMPLIFIER-INPUTS The first aspect of such a system which I want to emphasize is that inputs into the sensory network involve amplification and selection. For example, the visual receptor cells, the rods and cones of the eye, are sensory amplifiers. It is now believed that the rods probably operate at the "one-quantum level." This means that a single quantum of light coming into a rod can be detected, producing an amplification, which is of the order of 107 times (in electrical voltage), in the first nerve impulse that goes out into the retinal network and then is transmitted down the optic nerve. It is interesting to make a comparison between this amplifier and a little electric photo. multiplier pulse amplifier. The rods and cones are very small, only some 50 microns long and I to 3 microns in diameter, and this factor of 107 in amplification in such a small system is a very impressive performance. But these amplifiers, like all other sensory and electronic amplifiers, have necessary limitations on what they can detect about the external world. The limitations come from the fact that they are macroscopic, that they have a time constant,

225

JOHN PLATT

and that they have some noise background so they are not absolutely reliable. I would also emphasize that they are selective, and that they are unidirectional in their operation. Input into the photocathode or a phototransducer molecule (rhodopsin) is amplified in one direction down the system, and then produces an output which comes out at the other end. This chain cannot be reversed by putting in a light quantum at the output end and expecting to get an amplified signal at the input. I suspect that some of the aspects and limitations of perception which are important to us may depend on such limitations in these elementary amplifiers which mediate between us and the environment. SENSORY-MOTORDECISION-NETWORKS Secondly, I want to emphasize some of the new results on the combination of input-amplifiers and output amplifiers into sensory-motor decision-networks, and on learning networks in particular.

ACTION] Environment

AWAR

9

Proprioceptison

~~~~~~~~~~~~~~~~System

/

Growth TrioI Lerning

Fig. 1. A sensory-motor decision-network interacting with the environment.

Figure 1 is a schematic diagram of an organism detecting its external world. It is an attempt to diagram some of the essential relations for an organism as we now know them. In fact, what I have in mind is specifically an artificial organism, or "automaton," which is capable of surviving in certain types of environments. I am interested in the parallel between

226

Organism, Environment, and Intelligence as a System such an automaton and a small elementary nervous network which has certain properties such as learning processes. One should not belittle the value of such diagrammatic representations. The specification of models of this sort has played a great role throughout history in providing insights into the nature of man, from the Cartesian automatons to Wiener's feedback models. When one wants to study the brain, for example, a good model can be illuminating, not because it is exactly like the brain but because it suggests the kinds of questions to ask and the kinds of relations and details to look for. Wiener's cybernetics, or feedback model, to my mind is one of the two or three most important ideas of the twentieth century in terms of its implications for all sorts of fields. What we have to have now is a model which includes not only the essential elements of these useful models of the past, but some important new elements as well. In particular, this diagram is intended to represent a parallel-processing cybernetic system. All the millions of sensory input cells are symbolized by three little circles on the left side of the diagram which are on the boundary between the organism and the external world. From these sensory cells, in this schematic model, there emerge first-stage neurons (below left). These may be connected to second-stage and thirdstage neurons, and so on, and there may be cross-connections as shown at the bottom of the diagram. There may be other features, such as proprioceptive feedback loops, and there may be some kind of growth of new connections (shown at the bottom) as one goes to higher-order organization of events. All these neuron chains lead eventually to motor outputs (right side of Fig. 1). In fact, unless we do micro-electrode studies, essentially all we can learn about the internal workings of such a decision system is in terms of its motor outputs. These motor outputs feed back into the environment and lead to various environmental consequences (top of the diagram) which may or may not feed back into input in such a way as to modify this input. REAFFERENT STIMULATION The reason for emphasizing this particular parallel-processing network model, in which the inputs lead to motor outputs, which lead back to inputs, is because it is a good representation of the important phenomenon of "reafferent stimulation." I think it was von Holtz who first suggested the term, but recently the work of Richard Held and his colleagues in the

227

JOHN PLATT

Psychology Department of M.I.T. has shown the great importance of this process. What reafferent stimulation means is a change of the input field as a result of the motor output of the organism or animal itself. Held's work has shown that reafferent stimulation is essential to the organization or adaptation of the visual field. This had already been shown in a number of small ways. For example, Ditchburn and Riggs and their coworkers showed fifteen years ago that continual motion of the eyeballs is necessary for visual perception. They measured the motions by putting contact lenses on the eye, with a little mirror on the side of the lens to reflect a beam of light. They showed that when the subject thinks he is fixating most steadily on a point source of light, its image is in fact moving around over the fovea, which is a region about 10 across at the back of the eyeball. The movement consists of a series of drifting motions on which is superimposed a tremor motion of about a minute of arc, combined with occasional jumps or "flicks." These motions of the image over the retina can be canceled by an ingenious mirror arrangement, so that spots you are looking at move just as fast as your eye moves. Or they can be canceled even more successfully by mounting the little spot of light, or a microscope lens and a target pattern, on the contact lens itself, so that it all moves with the eye. The lens combination has no appreciable weight, so it does not significantly affect the motion of the eye, and it is not uncomfortable physically. The remarkable thing is that with such a "stabilized image," vision ceases. For example, if the fixed-pattern apparatus is put on in the dark and if the light is turned on, the pattern can be seen for a fraction of a second and then it fades out. Only when some motion of the image is allowed is it seen again. What this implies is that at this eyeball-level, motion is necessary for visual perception. Presumably the stabilized image fades out for physical-chemical reasons because the cells adapt, or come to a "steady state." FUNCTIONAL GEOMETRY I have used this principle myself in an attempt to solve what I think is a most interesting and classical visual problem: how do we see straight lines? I do not think my own solution is widely accepted in the field, but nevertheless I think it has some important properties which I would like to bring to your attention. The problem of how we see straight lines with a highly

228

Organism, Environment, and Intelligence as a System variable cellular retina is a difficult one by all past methods of approach. I think it is almost impossible to understand if we simply think of a static image of the outside world falling onto the retinal array, because the retinal arrays are genetically different from one person to the next. But even if they were not, it would be quite surprising if during childhood development, every cell came to lie in exactly the same micro-relation to the other cells, or if any micro-relation like this should be individually predictable to a high enough accuracy. It is worth remembering that the accuracy of our perception of straightness, at least in detecting the lateral displacement of a line segment, is of the order of 2 seconds of arc-much smaller than the diameter of a retinal cell. Certainly when we look at the retina through a microscope, the cell positions look very irregular. It therefore seems difficult to believe that a static map on the retina could be interpreted by you and by me to indicate the same object pattern, such as straightness, because the image will fall on one set of cells in your eye and a different set of cells, differently arranged, in my eye. But if we use reafferent stimulation-that is, a dynamic mechanism in which we move the images back and forth on the retina-the problem suddenly becomes much more simple. For example, a way in which we can tell whether the top edge of a blackboard is straight is to look at the left side of the edge and then to look at the right side. If the image of the left side falls on a certain set of cells in your eye, and if it continues to fall on the same set of cells as you move your eye across to the right side, then the line is straight. For no S-curve or general irregular curve can continue to fall on the same array of cells in your eye as you move it. In short, I am suggesting that the simplest way to organize the geometrical space of the external world is by looking for invariances under displacement. I call this the principle of "functional geometry," meaning a dynamic approach to geometry rather than a static approach in terms of mapping. The operation of "invariance under displacement" can be used by any eye and retina to single out as unique such important geometrical pattern elements and regularities as straight lines, parallel lines, and equidistant lines, as well as things like uniform circular arcs (provided you can rotate your eyeball about the optic axis, as indeed you can). It is interesting to see that this method of extracting pattern regularities is very similar to the methods used for generating mathematically perfect surfaces, such as spherical and plane

229

JOHN PLATT

surfaces, in high-precision optical work. John Strong, the designer of high-precision ruling engines, has described in detail the methods of grinding and lapping for generating perfect spheres, perfect planes, perfect helixes, perfect equidistant spacings of a ruled grating, and perfect gears with high-precision angular spacings of the gear teeth. Basically, his methods do not involve measurements, but simply involve sliding and lapping one surface, such as a spherical surface, over another which just fits it. In this kind of displacement-invariance method, you do not use any rulers, and you do not have any centers! The spheres define their own centers and radii as they are polished, and the numerical values are not even measured. If you try to specify the center, it tums out that you damage the approach to sphericity because you are "over-determining" the system. It has to determine itself. In commenting on this elimination of measurement and extemal centers, Strong says, "the methods of highest precision are all primitive methods." Does it not seem likely that a biological system would need to use some primitive general high-precision method of this sort? For the eye has no x-axes and y-axes for Cartesiancoordinate determination of the pattems of figures. It has no method, as far as I know, for determining the "shortest distance between two points." In fact, the elimination of these classical bases of measurement and construction means that functional geometry gives us a new axiomatic basis for geometry. We suddenly realize that Euclidean geometry is "string geometry;" it is the kind of geometry you get by drawing on the sand using a piece of string. This can be seen in the definitions: a straight line is the "shortest distance between two points"-of the kind that is represented by a stretched string. A circle is the "locus of points equidistant from a given point"-using a length of string as a measure of distance. Cartesian geometry, on the other hand, is "box geometry." It is the geometry of a European house, in which the walls are made of bricks that step up 1, 2, 3, 4, 5 . . . units. The floor is made of tiles that march out 1, 2, 3, 4, 5 . . . units, north and east. (I suspect that the Greeks, with the bowed-out columns and arched floors of their temples, could never have invented Cartesian geometry.) But now we come to functional geometry, which differs from these static, locus-of-a-point, geometries, and which has no axes, no centers, or "origin," no measures of distance, no guarantee that some primordial axes are at right angles to each other- and which gets along extremely well without all this initial apparatus. Incidentally, this even throws a new light on the "parallel

230

Organism, Environment, and Intelligence as a System postulate." In functional geometry, the parallel postulate is immediately provable, because if two lines are invariant under displacement along them-that is, if parallelism is defined by that kind of invariance-then if they do not meet in one region, they will never meet, because they are invariant. Therefore two parallel lines so defined will never meet. What I suspect is that Euclid or his predecessors had a kind of intuitive biological sense of this sort. Perhaps Euclid felt that two parallel lines ought never to meet, and since this cannot be proved by any method such as the "interior-angle" construction from the other axioms, he felt it had to be added as an additional axiom. The kind of discriminations that can be made by functional geometry raises an interesting question. The distinction between a "straight-line" displacement (x- and y-axes of the eyeball) without rotation (about the z-axis), and a "slightlycurved line" displacement with a little bit of rotation (about the z-axis) does not depend on functional geometry, since both patterns may be equally invariant under the right musclemotions. Rather, it depends on the accuracy of the "analog system" monitoring muscle movements, which tells you how much the muscles (of the z-axis) are moving. We thus have a lower-accuracy analog system for detecting some features of the pattern and a high-accuracy invariant array system for detecting other features of the pattern. This is like the boy polishing a spherical mirror, who cannot control the curvature very accurately even though he gets a very accurate spherical surface in every case. We cannot in fact tell the difference between a straight line and a slightly curved one unless we have another line nearby to compare it to (which converts the judgment from an analog judgment to an invariance-judgment). The analog and the invariance mechanisms are then two mechanisms working on top of each other. The activities of the muscles of the eye (analog) do not in themselves define a very accurate axis system for the movement of the eye, and when you are in a darkened room the image seems to drift around in quite an arbitrary way. But when you look at a circle or an object with geometrical regularities, the object itself, by its invariances, can determine its own centers and axes with high precision. PROPERTIESOF INVARIANT-SEARCHINGSYSTEMS I want to emphasize now some properties of this kind of system. If it should be used in perception, it would have several rather fascinating characteristics. For one thing, the

231

JOHN PLATT

perception of regularities by this method does not depend on the location of individual cells on the retina. All you ask is that the array of inputs is the same before and after the displacement-motion. Also, the accuracy of the perception does not depend on whether the image is straight or distortedbecause if the image of the left upper edge of the blackboard falls on an S-curve on your retina when you are looking at it over here, and then the right upper edge falls on the same S-curve when you look over there, then the line is straight! The question is simply invariance, and not the shape of the image. And just as it does not depend on the shape of the image on the retina, so it does not depend on the shape of the image farther back in the brain, on the cortex. The nervous network carries back a kind of distorted retinal map to the cortex, but we know that it is split apart onto the right and left half of the visual cortex. But we see that this split in space is irrelevant, as long as we have an invariance in the network under the scanning motion. The result is that with this kind of mechanism straightness is a property in the external field, not in the internal field. It is a property in a public space, in spite of our private differences in detailed anatomy. Held has therefore described functional geometry as a method which "transcends the differences in individual anatomy." This transcendence of anatomy is obviously extremely important if we are going to be able to have an ostensive public language. Only so can there be public reference, in which I can point to something, and have some assurance that you-or a young baby-may look in the same direction and may see a regularity similar to mine, in spite of the fact that your network and his network is different, each of our arrays of cells is different, and so on. Only by some such physical transcendence-method can we arrive at interpersonal regularities in an external public space, which can then be a basis for ostensive definition and for language. INTERACTION WITH THE EXTERNAL WORLD There are further interesting properties which appear when we consider how such a system interacting with its external world must work. Referring to Fig. 1, it can be seen that in this kind of relationship the external world completes the feedback loop. That is to say, we have half of the feedback loop inside the organism, while the other half is outside the organism. The result is that if we look at an object whose

232

Organism, Environment, and Intelligence as a System input-signals have just begun to come in, we do not see the object unless the eyeballs move, unless our motor system is manipulating the images across the visual field. In this sense, we do not see "an object"; we only see a family of invariancerelationships which are determined in part by self-motion. We might say with Martin Buber, then, that our perception depends on our having an "I-it" relationship to the object. Or to put it another way, we can organize the external world only by manipulating it, only by putting out our own amplified motor outputs in such a way as to alter what is in the field. It would not be wrong to say that in this kind of system awareness depends on action. There is no such thing as absolutely passive observation. Another important aspect of a parallel-processing cybernetic system like this is that it is an amplifying system. I have spoken of the amplification at the inputs, but we also have a further amplification occurring in the motor-operations of the outputs. These amplifications modify the external world; for they are designed or programmed to do that, and it is necessary to do that for the survival of the organism. Such an amplifying organism is a center of change of the environment, creating changes which can be amplified up to the maximum power that the organism can control. One new conjectural theorem for such complex systems is that it is probably not possible to predict the behavior of a given decision system by another decision system of the same order of complexity. The reason is that the second decision system cannot determine the initial state of the first decision system, and cannot know what the detailed quantum inputs are into this first system. This is because, in the first place, the inputs may be private: they operate at the one-quantum level, and one decision system may respond to one or a few quanta that signal, say, the presence of a red light, while there is no reason why the other decision system, in a different region of space, necessarily sees these quanta at all. Similarly, the internal structure of a decision-network is private, for it is very much more complex than the number of sensory detector cells that a similar network has to look at it with. Thus, in the case of human beings the number of input cells is of the order of 108, while the total number of neurons in the human decision system is of the order of 101. And the total number of synaptic connections between the neurons in the brain is of the order of 1000 times more numerous-since the average cell is estimated to have on the order of 1000 synaptic buttons-so that you have some

233

JOHN PLATT 1014 specific or not-so-specific synaptic junctions in the network. The result is that with my 108 little neurons of the eye, even if every one of them were detecting a different neuron or synapse in your brain, I could not even in principle detect the total initial state of your 1011 or 1014 neurons or synaptic connections at a given instant. Possibly there is no law in principle against the "mutual determinability" of two complex decision-systems (although I think there is), but there is certainly a kind of practical "forbiddenness rule" that one decision system cannot know what another decision system of the same size is thinking. Perhaps it may be possible for us with our 108 input-elements to see what a tiny earthworm's network is doing, but not what any network of comparable size is doing.

EPISTEMOLOGYOF DECISION-SYSTEMS There is no such difficulty in physics and chemistry observations. The reason is that the rules and laws of physics and chemistry are "low-information laws" in which we detect the motion of a few stars or atoms or indicator-needles. With our 108 detectors, we can detect redundantly, and with a good deal of invariance-checking, the motion of one billiard ball. And even when we come to the complex things in physics and chemistry, such as the observation of water boiling or of compounds of various sorts, these are still low-information subjects compared to biology. And they are very low-information subjects compared to the structure of a human brain. So I think many of the things that have been said in the past about the epistemology of science are really based on what we can know about low-information physics and chemistry. The epistemology, and our ideas of what can be operationally observed or predicted, are going to have to be modified when we jump from the small thousand-element or million-element systems of physics and chemistry over to the 101" and 1014 element-systems of the human brain. This is such an enormous jump and goes so far beyond the number of sensory elements available to study such systems that it will require a new attitude toward what we can know about each other. Of course there may also be statistical regularities and predictabilities in such decision-systems, and invariance-regularities of the type I have described. But there may also be assassination behavior which occurs only in the distorted brain of one particular person, and which is almost unpredictable in the billion-element conditionalities of its success

234

Organism, Environment, and Intelligence as a System or failure, but whose effects may be amplified up to the total energy that the human race can manipulate. It seems to me that this amplification in manipulating the environment leads us into a future which is branching out from each decision-system in an essentially unpredictable way. As a result, I am more sympathetic than most philosophers are to the recent attack by Milic Capek on what he calls the "myth of frozen passage." He is attacking the traditional Minkowski-type of four-dimensional time-space diagram, in which trajectories of particles in time are represented as being essentially completely predictable from the past. All pasts and all futures are thought of as being simultaneously "present" together in this frozen space of the diagram, which is perhaps not so different from the old idea that they were simultaneously present in "the mind of God," who is thus able to know all future destiny in advance. But the present thoughts suggest a quite different formulation, one in which future trajectories, with high amplification-manipulations of energy and materials, are branching out from this present time at any point where complex decision-systems are operating, in ways which are essentially unpredictable in detail by any other similar decision-system. A decision-system is continually modifying the environment around it. One arrives at a kind of existentialist picture of the operations of a decision-system, for which the only time is "now." As it uses its existing structure to operate on and amplify the existing and incoming environmental influences, it reshapes the world around it into a future of its own decision-determination. A further curious point is that the time and space of such a sensory-motor decision-system in their manipulation-rules come to be rather different from the time and space of physics; in some sense they are not Euclidean. We cannot speak accurately of a time, t, or an instantaneous time for a decision-system, because it takes 30 milliseconds or more for the nerve signal to go through me, so that my output at a "given" t is summatiIng inputs over a range of times before, and in fact it is uncertain just where and when we should say, "this is the output." The time is not the time of a lightning flash, as in physics, but is rather a distribution of times-past, including all those that are stored or reverberating in the structure so as to affect the output. This is true also of space: the inputs are not inputs from a point locus; rather they are inputs from whole figures, from the whole array of inputs. And the outputs are not outputs at a point, but are outputs which fill the room, as my voice fills the room.

235

JOHN PLATT

When is the instant of time? Where is the point of action? There is no point of action. What we are doing is operating on a kind of sensory input-totality and manipulating it into another kind of totality at the outputs. I am not saying that this cannot be reduced and analyzed in detail point by point; and I am not saying I do not believe in differential equations. What I am saying is that the philosophy of operation of this kind of decision-system is not the philosophy of operation of the locus of Euclidean points moving on sharp paths through a three-dimensional space continuum. The result is that this kind of decision-system operates in two worlds, so to speak. The first is the world of external, or objective, science, of the physics of the environment around it. This is the world in which the decision-system "prepares a state," lets the external objects run on without interaction, and then sees what happens. It is the external-environment part of Fig. 1, in the upper half, only weakly interacting with the organism. The world of physics is a world of weak interaction. But the other world, the world of the bottom half of the diagram, is not the world of physics but the world of cybernetics. It is the world of goal-directed behavior, of manipulative, purposive behavior which is trying its hardest to amplify and to interact strongly with the environment. It is the world which is not primarily concerned with scientific and detached observation and information, so to speak, except as these contribute to its personal knowledge, values, purposes, decisions, and acts. This very instrumental attitude toward science is implied, of course, in our basic science textbooks, but implied in an oddly suppressed way so that the students tend to forget their own human operational role-their decision-system role-in creating and evaluating science. THE THREE METHODS OF PROBLEM-SOLVING There are three methods of problem-solving which evolved in the course of evolution. The first is the method of problem-solving by survival. This is the "phylogenetic method" in which all the moths that didn't match the color of the bark got picked off by birds, and what is left is a species which has solved the problem of matching the color of the bark. The second method is problem-solving by individual learning. This is the "ontogenetic method." Whereas species survival involves the selection of chromosomes and DNA, individual learning involves the selection or connection of learning neu-

236

Organism, Environment, and Intelligence as a System rons in a learning nervous system. In this case the beast does not have to fall over the cliff before he is able to draw back from it, because he can see the edge of the cliff in time, or he realizes that he has been in a situation like this before, so he pulls back. The third method is problem-solving by anticipation. This comes only with the development of symbolic manipulation, so that by abstract analysis one knows the laws of nature which can be extrapolated into the future. This is of course the method of science. The difference between these methods is illustrated by the first Sputnik that was sent up. It was not one of thousands of Sputniks shot into space, of which all died except one which happened to go around. Nor was it like the rat finding its way through a learning maze, with the rocket going first too high and then too low until it finds out which orbit is satisfactory. No, the rocket was programmed in advance, with problem-solving by anticipation. An on-line feedback-control program was set up, and the laws of physics were used in the computers and the feedbacks to determine into which orbit it should go; and it went in the right orbit the very first time. The importance of these considerations is that more and more of our problems are problems that we have never met before; and they are problems that will have to be solved by anticipation by science-if they are to be solved at all before they kill us. This anticipatory manipulation of the environment by complex decision-system intelligence is something that has come late in the history of planetary evolution. From a broader point of view, Lederberg has described three great stages of evolution, starting with the preorganic molecules. He calls them stages of "chemogeny," or build-up of complex molecules in the sun's radiation-field; "biogeny," after self-catalysis begins, with natural selection; and "cognogeny." What I have been describing here is the development of systems making a collective transition over to the cognogenic stage of evolution, in which we are doing abstraction of natural laws and symbolic manipulation with enormous power control over the environment. Eventually, of course, cognogeny will include the extensive development of artificial sensory-motor decision-systems, or automata. It is philosophically interesting to inquire whether there may be some kind of convergent evolution-in different solar systems, for example-to the cognogenic stage. One sees the convergence of wings in reptiles, birds, and bats to structures built in different ways but having similar longitudinal struts;

237

JOHN PLATT

surfaces that have a "camber," a similar length-to-chord ratio for animals of a given weight floating in our particular density of air, and so on. There is a convergence to certain inherent or predetermined regularities of wing formation which are necessary for aerodynamic success. One can say the same things about the convergence of eyeforms, which have been invented independently seventeen times in evolution, according to one list. Again they converge to a small number of forms, with all of them in the higher organisms, I believe, having some close-packed directionalmosaic of cells like a retina. We suppose that this happens because this is probably the best way, if not the only way, to see simultaneous external relations in many directionsby using a close-packed parallel-processing input array. The physics of the situation, the environmental necessities, channel what is successful into certain directions. Religious people have interpreted this regularity of efficient form as an "argument from design." And I think a scientist could even agree that the wing of an animal of a certain size is "predestined," at least in a metaphorical or phsyicalnecessity sense, not to look like a club. It is predestined to have certain aerodynamic parameters. I am emphasizing this because I want to go on to ask the question, Is intelligence, and the emergence of the cognogenic stage, also a case of evolution toward a "predestined" set of parameters in the same fashion? That is to say, once complex decision-system organisms have passed a certain threshold in manipulating the environment-which might be a threshold of the use of tools or speech or fire-isn't it almost "necessary," in this same sense, that they go on to develop a manipulating environment-organizing symbol-using brain of the type we have? Could one talk about the possibility of convergent evolution in this sense toward intelligence? I think this is not an absurd or meaningless question, but one which could be solved by theory and experiment. And I think it is a question which biologists have an obligation not to deny or erase, because it is a question that inevitably comes up in discussing evolution, in discussing the meaning of the appearance of man and of intelligence with citizens and religious people and philosophers. COLLECTIVEINTELLIGENCE AND THE COLLECTIVEFUTURE We are on the threshold of a very peculiar future, unlike anything that has ever happened in the world before. It is 238

Organism, Environment, and Intelligence as a System one in which we organisms are all interacting like mad, in reshaping all of our environment. In the last few years we have stepped up by many orders of magnitude: in terms of communication speed, by a factor of 107 over what it was 100 years ago; in terms of travel speed, by a factor of 102 over 100 years ago; in weapons, by a factor of 106 over even 30 years ago, and so on. We have stepped into a new world of resources, of power, and of human interaction, within which evolution will take place on entirely different principles, with entirely different parameters of pressure and change from what has ever been before. As Lederberg has said, "There will be more evolution in the next 50 years than in the last 100,000." This is the end of the era of evolution by natural selection. It is the beginning of evolution by human selection. Every species on the globe is going to be increasingly perturbed in its numbers by our predation, our pollution, our breeding, or our protection. The numbers will not be determined by what we used to call "natural selection" so much as by conscious or unconscious human intervention which is now shaping what the structure of the biosphere is going to be. In such an unpredictable large-scale crisis, we will be able to survive only by making maximum use of all we know about anticipatory problemsolving, with anticipatory control and anticipatory feedbacks. The danger is clear; only in this way can we keep from destroying our biosphere and ourselves, by collective accident. If we do survive, these feedbacks, of course, will necessarily lead to higher degrees of cooperation and lower degrees of conflict in the world. Our increasing interaction-intensity will necessarily lead to a new kind of mutuality in our decisions, a mutuality in which we can no longer treat each other as objects, but must treat each other as co-participants and co-subjects in all our experiments and plans. What this all means is that in the times just ahead, if we survive, we are going to move toward something much more like an aggregate decision-system-more like a single organism-then we have ever been before. Perhaps it was implicit in the biological material and the physical necessities all along-as surely as the eventual development of wings is implicit in an atmosphere.

239

ESSAY REVIEW

What Philosophy of Biology Is Not DAVID HULL Department of Philosophy University of Wisconsin-Milwaukee

Periodically through the history of biology, biologists have tried to do a little philosophy and occasionally a philosopher has turned his attention to biology. In the past decade or so a body of literature has arisen which might legitimately be called philosophy of biology. The purpose of this paper will be to review the contributions made to this literature by philosophers during the past ten or fifteen years. Earlier work will be discussed only if it has proven especially influential. The contributions made by biologists to the philosophy of biology will be touched on only briefly, both because the biological literature is too vast to permit anything like a fair summary in the confines of a short paper and because the strengths and weaknesses of this literature tend to be quite different from the efforts of philosophers.1 In this paper I have not refrained from criticism, though it is sure to result in acrimony. There is too great a discrepancy between what philosophers produce under the guise of philosophy of biology and what philosophy of biology could be or, in my opinion, should be to pass over without comment. One striking feature of the remarks made by philosophers about biology is how frequently they are misinformed. For example, Mario Bunge in a paper on the weight of simplicity in the construction and assaying of scientific theories asks the following question: 1. Both restrictions in the scope of this paper have been ignored when special circumstances seemed to demand it. For example, Woodger began his career as a biologist and published most of his work prior to the time limits set for this paper, but he has produced a body of work in the philosophy of biology too important not to include.

241

DAVID HULL

What gave Darwin's theory of evolution through natural selection the victory over its various rivals, notably creationism and Lamarckism? Darwin's theory was in part logically faulty (remember the vicious circle of the "survival of the fittest"); it contained several false or at least unproven assertions ("Each variation is good for the individual," "Acquired characters, if favorable, are inherited," "sexual selection operates universally"); it had not been tested by observation, let alone by experiment on living species under controlled conditions (the development of antibiotic-resistant strains of bacteria, industrial melanism in butterflies, and a few other processes supporting the theory, were observed one century after The Origin of Species appeared); its explanatory power was clearly smaller than that of its rivals (irrefutable theories have the maximum post factum explanatory power); it had no inductive basis but was, on the contrary, a bold invention containing highlevel unobservables. And, if these sins were not enough to condemn the theory, Darwin's system was far more complex than any of its rivals.2 Critics of evolutionary theory seem evenly divided on the question of whether the survival of the fittest is false or tautological. In spite of the tiresome regularity with which this claim is made, it has little foundation. Leading evolutionists from Darwin to G. G. Simpson and Ernst Mayr have provided excellent explanations of why this principle is neither tautological nor viciously circular.3 Fitness in evolutionary 2. Mario Bunge, "The Weight of Simplicity in the Construction and Assaying of Scientific Theories," Phil. Sci., 28 (1961), 120-149. 3. G. G. Simpson, The Meaning of Evolution (New Haven, Conn.: Yale University Press, 1949); G. G. Simpson, Principles of Animal Taxonomy (New York: Columbia University Press, 1961); G. G. Simpson, This View of Life (New York: Harcourt, Brace & World, Inc., 1963); Ernst Mayr, Animal Species and Evolution (Cambridge, Mass.: Harvard University Press, 1963). In most instances the assertion that certain principles in the synthetic theory of evolution are tautologous stems from an extremely superficial understanding of evolutionary theory and an embarrassing uncertainty over the precise nature of tautologies. For example, Murray Eden, in P. S. Moorhead and M. M. Kaplan (eds.), Mathematical Challenges to the Neo-Darwinian Interpretation of Evolution (Philadelphia: The Wistar Institute Press, 1967), began with the assumption that the tautological nature of certain concepts in evolutionary theory was hardly controversial. Under the onslaught of several biologists present, Eden retreated to the position that perhaps "tautology" was the wrong word. Rather such claims are supposedly vacuous. From here he retreated to the assertion that the basic principles of evolutionary theory did not form a theory, and finally he concluded that since one cannot provide a crucial experiment to check whether or not the synthetic theory is false, it is in

242

Essay Review theory is a relative notion. Certain organisms in a given environment are fitter than others. A higher percentage of those organisms which are nearer the "fittest" end of the scale tend to survive than those at the other end. This scale in turn is ordered at least in part independently of the actual survival of these individuals. Of course, the claim that the fittest tend to survive can be made viciously circular if fitness were determined only by means of actual survival or into a tautology by defining "fitness" exclusively in terms of actual survival, but biologists do neither. This issue will be discussed in more detail later in conjunction with Anthony Flew's attempt to reconstruct Darwin's theory. Bunge contends that Darwin held several false or at least unproven assertions. Regardless of whether they are false or unproven, Darwin held none of them. For instance, Darwin maintained from the first that the variations which result in rudimentary or atrophied organs are not good for the individual. Such organs were imperfect and useless.4 Darwin did believe that "use and disuse seem to have produced some effect" but that there "is not sufficient evidence to induce us to believe that mutilations are ever inherited." 5 Nor did Darwin believe that sexual selection operated universally. He limited sexual selection just to animals with separate sexes, and among them the struggle between the males for possession of the females occurred only in most cases. Further, not all differences between males and females of the same species were due to sexual selection.6 Bunge says that industrial melanism in butterflies was observed a century after The Origin of Species appeared, when actually two long monographs were published by Tutt in 1890 (only thirty years after the Origin) in which he argued that melanism and melanochroism in British Lepidoptera were due to a combination of moisture and smoke under the action of some sense tautological or unfalsifiable. Karl Popper was frequently cited in this discussion as maintaining the last position. Anthony Flew ("The Concept of Evolution: A Comment," Philosophy, 41 [1966]) annihilates similar allegations by A. R. Manser ("The Concept of Evolution," Philosophy, 40 [1965], 18-34), though he himself argues later that the principle of the survival of the fittest must be made into a tautology (Anthony Flew, Evolutionary Ethics [New York: St. Martin's Press, 19671). See also T. A. Goudge, The Ascent of Life (University of Toronto Press, 1961). 4. Charles Darwin, On the Origin of Species: A Facsimile of the First Edition (Cambridge: Harvard University Press, 1966), pp. 450-456; Charles Darwin, The Descent of Man, And Selection in Relation to Sex, 2 vols. (London: Murray, 1871), p. 92. 5. Darwin, Origin, pp. 472, 135. 6. Darwin, Origin, pp. 468, 89-90.

243

DAVID HULL

natural selection.7 And, although it might seem an insignificant point to a philosopher, almost all of the observations were on moths, not butterflies. Finally, Bunge asserts that the theory of evolution through natural selection had no inductive basis. One wonders what Bunge can mean by "inductive basis" if Darwin's thirty years' labor in attempting to support his theory and all the data marshalled in the Origin provided no inductive basis. Only the first and last of the points raised by Bunge are especially important to the philosophy of science, but his obvious disinterest in evolutionary theory as a biological theory is discouraging. It is certain that he would not treat quantum theory in such a cavalier fashion. The differences between mesons and pions are important. The differences between moths and butterflies apparently are not. A second example of what can happen when a philosopher does not have sufficient understanding of the views which he criticizes is provided by Peter Caws when he terms the interbreeding criterion of the biological definition of species "artificial." Although he recognizes that the reason for the similarity of the members of a class of living things is that they have a common ancestry, he adds: "Even in this case, the borderlines between species have at times been very difficult to draw, and some rather artificial criteria (such as the ability to produce offspring together) have been used for judging whether or not a pair of individuals belong to the same species." 8 In the first place, ability to produce offspring together is not a criterion for judging whether or not a pair of individuals belong to the same species. Occasional hybridization between individuals of different species can and does occur. This does not mean that these individuals belong to the same species, not does it invalidate the biological species concept. "Interbreeding" as it appears in the biological definition of species applies to populations, not individuals, and is used to decide which classes of organisms (which taxa) are to count as species, not which individuals belong to the same taxon. Membership in a taxon is determined by phenotypic characters, mainly morphological. Whether or not that taxon is to be considered a member of the species category is determined by various evolutionary relationships, in particular interbreeding habits. 7. J. W. Tutt, "Melanism and Melanochroism in British Lepidoptera," Entomologists' Record, 1 (1890); 2 (1891). 8. Peter Caws, The Philosophy of Science (Princeton, N.J.; D. Van Nostrand, 1965), pp. 40-41.

244

Essay Review The taxon-category distinction which underlies the confusion in which Caws finds himself will be discussed later. Perhaps because of his failure to understand the role of the ability to produce fertile offspring in the biological definition of species, Caws terms this criterion "artificial." Numerous objections have been raised to the biological definition, most frequently that it is too difficult to decide when the criterion of potential interbreeding is fulfilled. According to Mayr, two populations are potentially interbreeding whenever they are prevented from breeding only by geographical isolation. If any isolating mechanism is operative, they are separate species.9 But even the most persistent critics admit that the mechanism of interbreeding is central to the evolutionary development of sexually reproducing organisms. If the ability to produce fertile offspring is an artificial criterion in deciding what is to count as evolutionary units, one wonders what a natural criterion would be like.'0 Numerous philosophers have taken an interest in evolutionary theory because of the consequences which they see in it for man. A book entitled Evolution and Philosophy by Andrew G. van Melsen is typical." Van Melsen's main thesis is that natural science deals only with the "outside" of reality, but there is also an "inside" to reality which the natural sciences cannot touch. This "inside" is manifest in man, who has direct access to the "self," the primordial datum. Anything a physical scientist might tell us about man "is peripheral to this primordial datum." The existence of this "self" is responsible for one of the gaps in the scale of nature, but there are others. Just as there is an abyss between man and (other) animals, 9. Mayr, Animal Species, 91. 10. The biological literature on the species concept is overwhelmingly large. The best discussions of the pros and cons of the biological definition of "species" can be found in Simpson, Principles; Ernst Mayr, Systematics and the Origin of Species (New York: Columbia University Press, 1942), and "The Species Concept," Evolution, 3 (1949), 371-372; also Ernst Mayr (ed.), The Species Problem (Washington: American Association for the Advancement of Science Publication Number 50, 1957); Ernst Mayr, "Isolation as an Evolutionary Factor," Proc. Amer. Philos. Soc., 103 (1959), 165-230, and Animal Species. For two philosophers who take the biological aspects of the species problem seriously, see Morton Beckner, The Biological Way of Thought (New York: Columbia University Press, 1959), and Hugh Lehman, "Are Biological Species Real?" Phil. Sci., 34 (1967), 157-167. 11. Andrew G. van Melsen, Evolution and Philosophy (Pittsburgh: Dusquene University Press, 1965). See also Adolf Portman, New Paths in Biology (1961; English trans., New York: Harper and Row, Publishers, 1964), and Hans Jonas, The Phenomenon of Life: Toward a Philosophical Biology (New York: Harper and Row, Publishers, 1966).

245

DAVID HULL

there is an abyss between the living and the nonliving. Some authors of this bent also mention a hiatus between plants and animals, but for some reason van Melsen plays down this gap. This increasing complexity of externally observable structure is paralleled by a "growing interiority culminating in man's self-consciousness." It is this "interiority" that necessitates the recognition of these different levels of being. What can possibly be said in reply to claims such as these? Of course, human beings are conscious and self-conscious. A biologist can be a mechanist-reductionist-neo-Darwinian without being a simple-minded behaviorist. Such biologists can also be aware of the epistemological distinctions which give rise to these and similar utterances. For example, G. G. Simpson says that "our human universes, the ones in which we really have our beings, depend at least as much on our inner perceptions as on the external, physical facts." 12 But the philosophers who emphasize the distinction between the internal and the external world in connection with evolutionary theory do so in order to assert that the existence of the inner world somehow counts against the sufficiency of evolutionary theory. It is difficult if not impossible to discern why man's consciousness and self-consciousness provide greater problems for evolutionary theory than any other so-called "emergent" characters. Until these authors make their claims a good deal clearer, a biologist would be hard put to decide whether they conflict with, are extraneous to, or are consonant with modern evolutionary theory. As is frequently the case with philosophers who wish to insulate the "self' from all inroads by science, van Melsen finds evolutionary theory below par as a scientific theory. Even today Darwin's explanation of evolution lacks a solid foundation. "The theory of evolution in general is, as we have seen, essentially based on 'belief rather than anything else." But he also discusses the evolutionary views of G. G. Simpson and Teilhard de Chardin as if they were on an equal footing scientifically. If the synthetic theory of evolution is based on belief, what must Teilhard's theory be based on? What kind of proof is necessary before a scientific theory can be accepted? Is there no difference between the evidence adduced in support of modern evolutionary theory and Teilhard's views? The difference between the two theories is not so much in their conclusions but in the evidence and arguments adduced by their authors to support them. Teilhard seems to have no conception of what a scientific theory is, what evidence 12. G. G. Simpson, "The World into Which Darwin Led Us," Science, 131 (1960), 966-974; reprinted in Simpson, This View, 3-25.

246

Essay Review is, or how evidence can be brought to bear on a scientific problem. And finally, we are reminded that "the Aristotelian view remains valid even in the context of an evolutionary world view." The literature of this genre is unfortunately extensive. It also follows a surprisingly rigid script. Characteristically it is argued that evolutionary theory is not "proved." After all, it is "only a theory." Scientists are chastised for being dogmatic and taking too much on faith rather than on the evidence. Rather than keeping an open mind on the matter, they plot to stifle all contrary opinions. The utterances of Pius XII, on the other hand, are in conformity with the best scientific tradition. Teilhard de Chardin is inevitably interjected into the discussion, either to show how his views are superior to those of the run-of-the-mill evolutionist, or in order to tar all evolutionists with him for irresponsible speculation. W. R. Thompson (1966) takes the latter tack, lamenting "the infatuation of Catholic intellectuals for the Teilhardian pseudoscience." 18 But he seems to overdo it a bit when he goes so far as to accuse Teilhard of perpetrating the Piltdown hoax. Medawar (1967) concludes more cautiously albeit more condescendingly that Teilhard "had about him that innocence which makes it easy to understand why the forger of the Piltdown skull should have chosen [him] to be the discoverer of its canine teeth." 14 Also characteristic of this body of literature is the view that, in the last analysis, Aristotle was right. Aristotle's concept of Nature provides a badly needed philosophic dimension to biology. DNA exists. Hence, Aristotle was right. One feature of efforts such as those of van Melsen's to discover what consequences evolutionary theory has for man is that evolutionary theory as a scientific theory plays no role whatsoever in his exposition. All he would have needed to know in order to develop his thesis is that man evolved from other animals and that living creatures developed from nonliving matter. All the intricacy of evolutionary relationships, the difficulties with various mechanisms, the recalcitrant data, the wealth of supporting evidence are passed over. Whatever philosophy of biology might be, this is not it.15 13. W. R. Thompson, "The Status of Species," in Vincent E. Smith (ed.), Philosophical Problems in Biology (New York: St. John's University Press, 1966), pp. 67-126. 14. P. B. Medawar, The Art of the Soluble (London: Methuen, 1967). 15. All of the above assertions can be found expressed variously in two volumes edited by Vincent E. Smith: Philosophy of Biology (New York: St. John's University Press) and Philosophical Problems of Biology, as wel as in Paciftc Philosophy Forum, 6 (1968), 1-99.

247

DAVID HULL

One exception among those philosophers who are primarily interested in evolutionary theory because of its possible consequences for the nature of man is T. A. Goudge.'6 In The Ascent of Life he goes to the trouble of providing a philosophically oriented analysis of modern evolutionary theory. Since his views on evolutionary theory are not especially controversial and depart in no important respect from those of the biologists he cites in his preface (Carter, Dobzhansky, Haldane, Huxley, Mayr, Muller, Simpson, and Wright), one might wonder what point there could be in a philosopher going over the same ground. This perplexity can be dispelled quickly by reading the book. Goudge's whole approach to the subject is different from that of a scientist. He does not approach his exposition according to various kinds of empirical phenomena (such as, kinds of species, isolating mechanism, hybridization, populations, genetic recombination, etc.) but according to the logically important differences to be found among the phenomena (for example, the historical aspects of reconstructing particular phylogenetic sequences, the peculiar nature of historical explanations, the causal aspects of evolution and the systematic explanations made possible by certain evolutionary laws and law-like statements). Only after such an analysis does he turn his attention to the implications of evolutionary theory for man. Goudge is very cautious in his assessment of the place of man in evolution, but the importance he puts on the question is indicated by the title of his book. He is interested in the ascent of life and whether man is at its forefront. Goudge argues that in spite of retrogressive periods, more diverse kinds of organisms are alive today than ever before and a higher percentage of more recent organisms are biologically more efficient than earlier organisms. As far as man is concerned, he is extremely flexible in his adaptiveness. He is a dominant type and the dominant primate. He is also "almost certainly the youngest species of mammal now on earth" and as such "there is a sense in which he is quite literally the highest species." 17 As cautious as these claims are, Goudge goes too far. There is no evidence to indicate that man is the youngest species of mammal and, even if he were, he would hardly thereby become the highest species; perhaps the highest species of mammal, but hardly the highest species period. Man is very efficient, flexible in his adaptiveness, and so on. So are cockroaches. Man is unique. So are cockroaches. Only when Goudge 16. Goudge, Ascent. 17. Goudge, Ascent, 133, 134.

248

Essay Review leaves the realm of such strictly biological features of man does the uniqueness claimed for man begin to carry some weight. Only man has developed culture and has been able thereafter to pass on information by more direct means than can other organisms. Hence, new possibilities and new difficulties have opened up for the future development of man. Enough has been said, I think, to show how unsuccessful contemporary philosophers have been in extracting the consequences of biology for philosophy. What of the other side of the coin? Have philosophical analyses of biology provided any insights into biological phenomena, any clarity which biologists themselves have been unable to provide, a deeper understanding of biological theories? When philosophers have turned their attention to biology they have tended to limit themselves to a few issues-vitalism, teleology, reductionism, and related topics. One thing is obvious from this list. Philosophers have not been motivated in their choice of topics by any concern with issues currently of interest to biologists. From the point of view of contemporary biology, both vitalism and teleology are stone-cold dead. No better proof can be found than that offered by recent attempts to argue to the contrary.18 In support of vitalism the observation is made that living creatures are not just matter but structured matter and that the world exhibits finality because regularities exist. The major problem with this defense of vitalism and teleology is that no materialist or mechanist ever held any differently. Even though much of the heat generated by these controversies was due to misleading formulations, it is hard to believe that the disagreements were entirely verbal. There is substance to these disagreements, and central to all of them is the role of "principles" in science. For example, J. H. Randall in his attempt to rehabilitate Aristotle emphasizes that Aristotle's formal and final causes are principles of understanding.19 18. See Pacific Philosophy Forum, 6 (1968), in which several philosophers argue that the vitalism-mechanism controversy cannot be dismissed as a dead issue. The major paper in this number by Hilde Hein is sufficiently innocuous, but in the others we are once again subjected to such inanities as "the human intellect violates the laws of the irreversibility of space and time"; "no less than the evolutionist's view, the approach of the creationist yields a true and satisfactory philosophy of life"; the fossil remains of so-called missing links between man and other primates are actually the remains of abnormal individuals of the unaltered species. Edward Manier had the unenviable task of replying to these papers. 19. J. H. Randall, Aristotle (New York: Columbia University Press, 1960). Numerous other examples in addition to Randall could be given. For example, Michael Polanyi, in "Life's Irreducible Structure" (Science, 160 [1968], 1308-1312), recognizes various levels of organization and

249

DAVID HULL

They don't enter causally into the world of things. They don't do anything. Living creatures do what they do because of their organization and can be understood only in terms of their organization. All it takes for a world to exhibit finality is for there to be regularities and for the understanding of these regularities to be increased by stipulating a stage in these sequences which can be thought of as "final."On this attenuated view only a totally chaotic world could be nonteleological and un-Aristotelian.To be sure, Aristotle'sformal and final causes are principles of understanding, but in too many instances Aristotle explicitly has them doing things in the empirical world. Randall himself finds it difficult not to slip on occasion and have these principles doing things; for instance, he has nous nousing nous. The views expressed by early materialists and mechanists were certainly overly crude, but to the extent that contemporary knowledge is applicable to the various stages of these controversies, the mechanist-materialistswere right. The only remaining issue which is even vaguely related to these theses is the translation of teleological modes of expression into nonteleologicallanguage and the specificationof the characteristics of those systems whose actions are frequently described teleologically. Both of these tasks are proving a good deal more intricatethan one might at firstexpect.20 The status of reductionism (and related topics) is more equivocal. At a recent conference on the history and philosophy of biology sponsored by the Commission on Undergraduate Education in the Biological Sciences, the discussion from the floor repeatedlyreturned to the question of whether biology was in some sense reducible to chemistry and physics and what such a reduction entailed. Two things were made clear principles which apply to certain levels and not to others, but he also has these higher principles "harnessing" the lower processes. See also Marjorie Grene, The Knower and the Known (New York: Basic Books, Inc., 1966). No biologist would deny the existence of levels of organization or of principles which apply to these levels, but how can these principles do things? What can William E. Carlo ("Mechanism and Vitalism: A Reappraisal," Pacific Philosophy Forum," 6 [1968], 57-68) mean when he says that "there is a principle of organization which takes the matter of a man, shapes it, molds it, and organizes it into a man"? See also P. R. Durbin, Philosophy of Science: An Introduction (New York: McGraw-Hill, 1968), and V. E. Smith, Science and Philosophy (Milwaukee, Wisc.: Bruce, 1965). 20. See Beckner, Biological Way; Hugh Lehman, '"unctional Explanation in Biology," Phil. Sci., 34 (1965), 1-19; J. V. Canfield, "Teleological Explanation in Biology," Brit. J. Phil. Sci., 14 (1964), 285-295; J. V. Canfield (ed.), Purpose in Nature (Englewood Cliffs, N.J.: Prentice-Hall, 1966).

250

Essay Review by these discussions-biologists and philosophers mean very different things by "reduction" and are worried by very different aspects of the problem. There does seem to be at least one way in which the issues of interest to both biologists and philosophers can be joined. During the last thirty years a process has been under way in genetics which fulfills all the requirements of what both scientists and philosophers have in mind when they speak of reduction. A biochemical explanation is being produced for phenomena which have been explained previously in terms of classical Mendelian genetics. Geneticists have not been "reducing" Mendelian genetics to molecular genetics in the sense of providing translation rules. The relationship is too complex for that. For example, classical genetics speaks of things like "the gene for albinism." In molecular genetics there is no such thing as the or a gene for albinism. Any one of several changes in the genetic code (deletions, inversions, and so on) at numerous different loci can result in the failure to produce pigments. What is the molecular version of the Mendelian statement that in man the gene for albinism is epistatic to the gene which produces pigmentation? Do aRl phenomena which were previously explained in terms of recessive epistasis receive the same kind of explanation in molecular biology? Is the molecular explanation for dominant epistasis at all related to that for recessive epistasis? These phenomena are closely related in Mendelian genetics. Do they receive comparable explanations in molecular genetics? For any philosopher who is interested in the question of reduction, a detailed investigation of the inferential and definitional relations between these two theories would seem absolutely necessary, as careful an analysis as that which thermodynamics and statistical mechanics have received.21

Regardless of the outcome of the dispute over reductionism, there is certainly more to the philosophy of biology than whether or not biology can be reduced to chemistry and physics! During the last decade or so, biology has been forced briefly to the center of the stage in the philosophy of science because of the relevance of explanation and prediction in evolutionary studies to a remark made by Hempel and Oppen21. There is a large body of literature devoted to the problem of reduction. Recent discussions of reduction in biology can be found in Grene, Knower; R. T. Blackburn, ed., Interrelations: The Biological and Physical Sciences (Chicago; Scott, Foresman, 1966); K. F. Schaffner, "Approaches to Reduction," Phil. Sci., 34 (1967), 137-147; K. F. Schaffner, "Antireductionism and Molecular Biology," Science, 157 (1967), 644-647.

251

DAVID HULL

heim in their classic paper on the logic of explanation.22 After setting out four conditions of adequacy for scientific explanation, they state that the same formal analysis applies to scientific prediction as well. Although almost every aspect of the Hempel-Oppenheim thesis has been subjected to extensive criticism, it is the symmetry thesis which has received the most sustained attack. The significance of biology for this controversy in the philosophy of science is that biologists frequently set forth what they take to be explanations of the evolutionary development of certain groups when they readily admit that they could not have made much in the way of any reasoned predictions in the matter. As gratified as biologists may be for the attention that one of their theories had at last received from first rate philosophers of science, the actual nature of evolutionary inferences was investigated none too intensively. What biologists actually did in producing certain putative explanations rapidly became peripheral to the issue of whether or not these formulations were truly explanations.23 Was Hempel and Oppenheim's analysis purely description, explication, or legislation? The introduction of evolutionary reconstructions into a controversy in the philosophy of science points up even more strongly the need for a careful and detailed study of the logical structure of modern evolutionary theory and its relation to historical data. Periodically philosophers have discussed evolutionary theory and evolutionary reconstructions, but their treatments have left much to be desired. Abraham Kaplan considers evolutionary theory a concatenated theory in contrast to hierarchical theories like relativity theory, but he says little about the actual structure of concatenated theories.24 A. G. N. Flew on the other hand contends that there is a deductive core to 22. Carl G. Hempel and Paul Oppenheim, "The Logic of Explanation," Phil. Sci., 15 (1948), 135-175. 23. Of all the papers devoted to the relevance of explanation in evolutionary theory to the symmetry thesis, Michael J. Scriven in his "Explanation and Prediction in Evolutionary Theory," Science, 130 (1959), 477-482, is the most concerned with evolutionary theory as a biological theory, rather than as a handy source for an example. See also Michael J. Scriven, "Explanation, Prediction, and Laws," in Herbert Feigel and Grover Maxwell, eds., Minnesota Studies in the Philosophy of Science (vol. III, Minneapolis: University of Minnesota Press, 1962), pp. 477-482; Adolf Griinbaum, "Temporally Asymmetric Principles, Parity between Explanation and Prediction, and Mechanism versus Teleology," in B. Baumrin, ed., Induction: Some Current Issues (Middletown, Conn.: Wesleyan University Press, 1963); Hugh Lehman, "On the Form of Explanation in Evolutionary Biology," Theoria, 32 (1966), 14-24. 24. Abraham Kaplan, The Conduct of Inquiry (San Francisco, Cal.: Chandler Publ. Co., 1964).

252

Essay Review Darwin's argument in the Origin.25 Several attempts have been made to specify precisely what this deductive core is. Flew cites Julian Huxley's formulation26 but finds it "curiously slapdash." His own formulation is scarcely better. He says, for instance: Though the argument itself proceeds a priori, because the premises are empirical it can yield conclusions which are also empirical. That living organisms all tend to reproduce themselves at a geometrical ratio of increase; that the resources they need to sustain life are limited; and that while each usually reproduces after its kind sometimes there are variations which in their turn usually reproduce after their kind: all these propositions are nonetheless contingent and empirical for being manifestly and incontestably true. That there is a struggle for existence; and that through this struggle for existence natural selection occurs: both of these propositions equally are nonetheless contingent and empirical for the fact that it follows, necessarily as a matter of logic a priori, that wherever the first three hold the second two must hold alSo.27 Everything which Flew says is true, important, and needs saying, with one exception. Neither of his conclusions follows deductively from the premises which he presents. All that can be deduced is that not all those organisms which are born will survive. Flew is aware that he has not presented a rigorous deduction. To do that "one would have to construct for all the crucial terms definitions to include explicitly every necessary assumption." 28 Though he himself does not attempt such a rigorous reconstruction, he believes that such an endeavor would be "an exercise which might prove instructive." 29 Anyone who undertakes this exercise will find that it is highly instructive and a good deal more than an exercise. The only step which Flew takes toward such definitions is to identify surviving to reproduce with being the fittest. He observes that if this identification is not made, the deductive argument which he has set out is no longer valid. The question of whether certain basic principles of a sci25. Anthony Flew, "The Structure of Darwinism," in M. L. Johnson, Michael Abercrombie, and G. E. Fogg, New Biology, No. 28 (Baltimore: Penguin Books, 1959), pp. 25-44; Flew, Evolutionary Ethics. 26. Julian Huxley, The Process of Evolution (London; Chatto and Windus, 1953), p. 38. 27. Flew, "Structure," 28. 28. Ibid. 29. Ibid., p. 29.

253

DAVID HULL

entific theory are analytically connected within that theory is neither new nor unique to evolutionary theory, but in order to connect fitness analytically with actual survival, a distinction fundamental to evolutionary theory and to science in general must be ignored. This distinction is the difference between what could happen, given the appropriatelaws, and what actually does happen. For instance, given Newton's laws a planet must revolve around a star in a conic section. Which of these possible paths the planet actually takes depends upon the particular makeup and history of that star system. Similarly, biologists want to retain the distinction between which organisms do in point of fact survive and those which have the greatest likelihood of surviving-and they define "fitness" in terms of the latter notion. Every organism which could survive, given the appropriatelaws, does not survive. "Accidents" do happen and are frequently important in evolution, especially in small populations. The appropriatelaws in this case are those of physiology, ecology, embryology, and so on. It may well be true that in principle all macroscopic phenomena are governed by deterministic laws and that all these laws can be organized into a deductive hierarchy, but biologists do not have these laws. The laws which give substance to the claim that an organism which did not survive was nevertheless exceedingly fit are currently not deducible from evolutionary theory and are formally independent of it. Until the day that biologists can organize all of the relevant parts of biology into one grand deductive hierarchy and know all the relevant antecedent conditions for the evolutionaryphenomena under study, not only are they entitled to retain the distinction between what organisms actually survive and those that are the fittest,they must. Flew can be seen to vacillate on precisely this issue in his comments concerning the survival of the fittest. Sometimes he says that "actual or possible survival is to be construed as the sufficient condition of fitness to survive." Sometimes he says that "actual survival to reproduce is itself within Darwin's theory the sole and sufficient criterion for fitness to survive."Only from this latter assertion is he entitled to conclude that "it is as always the fittest who have survived, the fittest who do survive, and the fittest who will survive."80 When philosophers have turned their attention from evolutionary theory to evolutionary reconstructions, they have also found much to criticize. For example, is it true, as Woodger said many years ago, that phylogenetic explanation is "a 30. Flew, Evolutionary Ethics, pp. 14, 36, 14.

254

Essay Review historical explanation in the strict sense and one which could not be generalized because it would describe a unique series of changes characterizing an evolutionary succession" and that we are "'in possession of no inductive generalization regarding the modus operandi of evolution of such high probability and generality as will justify us in asserting with any confidence what happened in an historical example"? 81 The evolution of Hypohippus was a unique event, but the occurrence of adaptive radiations, the invasion of new ecological niches, cases of convergent evolution, and so on, are not. Such phenomena are generalizable. Whether or not a particular species resulted from one or more of these processes is another story.82 It is impossible, however, in the space of a single paper to summarize all the various criticisms which have been made of evolutionary theory by philosophers. Rather I have chosen to discuss in detail the criticisms of a single philosopher, Marjorie Grene. I have chosen her work to discuss for the simple reason that no criticisms of evolutionary theory in the past decade or so have irritated biologists as much as those set out by Grene in her paper "Two Evolutionary Theories."33 The reasons for this irritation throw some light, I think, on the shortcomings of philosophy of biology. In her paper Grene contrasts the evolutionary theories of G. G. Simpson and 0. H. Schindewolf.34 Such an undertaking is perfectly legitimate and should have proved instructive, since Schindewolf's theory, unlike the musings of Teilhard, is a respectable scientific theory. Although the paper begins by setting out a fairly balanced account of the two theories, it gradually develops into a sustained attack on the orthodox neo-Darwinian views of Simpson. But the synthetic theory has been criticized before. Like all scientific theories it is under constant revision and reexamination, and in many instances the severest critics are found among those who con31. J. H. Woodger, Biological Principles (1929; 2nd ed., London: Routledge & Kegan Paul, 1948), pp. 394, 402. 32. For differing opinions on these views see W. B. Gallie, "Explanation in History and the Genetic Sciences," Mind, 64 (1955), 161-167; R. P. Gould, "The Place of Historical Statements in Biology," Brit. J. Phil. Sci., 8 (1957), 192-210; T. A. Goudge, "Causal Explanations in Natural History," ibid., 9 (1958), 194-202; Beckner, Biological Way; M. J. S. Rudwick, "The Inference of Function from Structure in Fossils," Brit. J. Phil. Sci., 15 (1964), 27-40. 33. Marjorie Grene, "Two Evolutionary Theories," Brit. J. Phil. Sci., 9 (1958), 110-127, 185-193. 34. Simpson, Meaning; 0. H. Schindewolf, Grundfagen der Palaontologie (Stuttgart: Schweizerbart, 1950).

255

DAVID HULL

sider themselves advocates of the theory. Why then did Grene's particular criticisms appear so offensive? Grene's attack on the synthetic theory is divided into two parts. First she tries to establish that, as far as the evidence is concerned, the two theories are about on a par. She claims that Simpson and Schindewolf "disagree seldom, if at all, about the 'facts'." The two theories merely "provide alternative frameworks for understanding the data." Although for some details one point of view is preferable, for other details the other point of view is preferable. Thus, "it seems to be purely a matter of choice which we prefer . . . Perhaps what we need, then, is a more inclusive theory, which will assimilate adequately both sides of the ambiguity." 35 After establishing to her own satisfaction the equality of the two theories as far as the evidence is concerned, she proceeds to argue that Schindewolf's theory is more adequate than Simpson's on epistemological grounds. For example, Simpson claims not to make use of types in an epistemologically significant sense; i.e., he may occasionally refer to types, but he claims that in no instance does he suppose that natural kinds can or must be defined in terms of necessary and sufficient conditions.36 The reason that Schindewolf's theory is more adequate than Simpson's is that he admits types and makes them a part of his theory, whereas Simpson makes use of them surreptitiously, though they are incompatible with his theory. Grene does not claim that "Schindewolf's typetheory is adequately explanatory. Only that it is not selfcontradictory and so is at least a possible starting-point for for giving philosophical asking philosophical questions-not answers, as my critics suggested I mean to do." 87 What is to be said of Grene's argument? In the first place, almost all biologists disagree with her assessment of the evidence. Taking the two theories on a whole, the vast majority of the evidence supports the synthetic theory, and it is in just those cases where Schindewolf departs most radically from the synthetic theory that the evidence is most decidedly against him.38 35. Grene, "Two Evolutionary Theories," 185-186. 36. G. G. Simpson, "Types in Modern Taxonomy," Amer. J. Sci., 238 (1940), 413-431. 37. Marjorie Grene, "Two Evolutionary Theories: A Reply," Brit. J. Phil. Sci., 14 (1963), 152-153. 38. Of course, not all biologists agree with all aspects of the synthetic theory or that all the data are favorable to it. Almost any biologist will have some reservation about one or another of the tenets of the synthetic Graham Cannon, Paul G. theory, and Grene cites several examples-H. 'Espinasse, Ronald Good, Adolf Portman, W. R. Thompson, and C. H. Waddington. What Grene fails to mention is that these scientists disagree

256

Essay Review For example, Grene cites Schindewolf's belief that basically new types or patterns of organization have a "sudden origin." There were no feathered creatures. Then there were. The fossil evidence happens not to be decisive on this issue, but given what we know of genetics and physiology, it is unlikely that such macro-phenotypic changes resulting from mutation could occur in the space of a single generation and the results be viable-even once, let alone in the origin of every new type. It does not help in the least to say that the gross phenotypic changes were due to micromutations early in development. The magnitude of the change in the genetic makeup is not at issue, but the magnitude of the resulting phenotypic change. According to Schindewolf, one basic plan of organization must be changed into another in the space of a single generation. Most biologists find this unlikely.39 with different aspects of the synthetic theory and with each other. It is not the case that a unified, viable alternative to the current orthodox theory has been brought forth. Grene herself makes passing reference to looking at "phylogeny as an ontogeny writ large, at the history of groups as expressing a fundamental rhythm still, in its intimacy, unknown to us, but analogous to the rhythm of individual development." Instead of species becoming extinct by the various mechanisms of the synthetic theory, "this pattern of living-simply played itself out" (Marjorie Grene, "The Faith of Darwinism," Encounter, 13 [1959], 52, 55, reprinted in Grene, Knower, 185-201). Biologists have investigated this possibility, most recently, Lawrence S. Dillon, "The Life Cycles of the Species: An Extension of Current Concepts," Systematic Zoology, 15 (1966), 112-126. The idea even occurred to Darwin Journal of Researches into the Natural History and Geology of the Countries Visited during the Voyage of H. M. S. Beagle Round the World [2nd ed., revised, London: Murray, 1845]). Unorthodox views do get a hearing. Unfortunately, the weight of evidence has proved to be against this particular unorthodox view. The fact that most biologists, after listening to these dissident views, remain unconvinced leads Grene to label them dogmatic. If refusal to go against the weight of evidence is dogmatism, then these biologists are being dogmatic. 39. Although Goldschmidt was a strong proponent of macroevolution by macromutation, he was aware that the two positions were independent of each other and said so explicitly: Continuing the line of argument derived in the foregoing chapters, we must find out further whether the development system is capable of being changed suddenly so that a new type may emerge without a slow accumulation of small steps, but as a consequence of what we called a systematic mutation. Such an analysis may be carried out in complete independence from the detailed conceptions which we developed concerning the architecture of the germ plasm and its changes. It does not make any difference whether a single macroevolutionary step is caused by a major change within the chromosomal pattern, a systematic mutation, or by a special kind of gene mutation with generalized effect, if such is imaginable. The decisive point is the single change which affects the entire reaction system of the developing organism simultaneously, as opposed to a slow accumulation of small additive changes" (R. B. Goldschmidt, The Material Basis of Evolution [New Haven: Yale University Press, 1940], p. 251).

257

DAVID HULL

Thus, with respect to the first part of Grene's argument, most biologists think that Grene is factually in error, and it was these factual errors which elicited much of the negative response from biologists.40 Grene, however, is not primarily interested in these factual disputes but in her second point that Schindewolf's theory is somehow epistemologically more adequate than Simpson's theory. To be specific, Schindewolf freely admits certain concepts which Simpson pretends to do without, but introduced surreptitiously. One of these concepts is the concept of "type." Traditionally the notion of type entails that membership in a natural kind is determined by one set of characters which are severally necessary and jointly sufficient. If Grene intends to be referring to the essentialist notion of type, then she could be making either of two claims -either the weak claim that Simpson thinks he is not using an epistemologically significant concept of type when actually he does, or the strong claim that not only does he use such a type concept, but also he must. On the first count, Grene is probably right. Even though Simpson, Mayr, and other evolutionists have repeatedly emphasized that evolutionary theory is incompatible with the essentialist-type concept, this point of view is quite difficult to maintain in the midst of complex lines of reasoning.41 Simpson may well have slipped on occasion. The strong claim, however, is another matter. Essentialism is a philosophic position of long standing, but if Grene is to resurrect it, she owes the reader some explanation of why this particular notion is a necessary element in any adequate epistemology. In the article in question, she provides none. In a later paper, she retums to this theme but this time she equates the concept of type with "gestalt-idea." But it is one of the key features of a gestalt that one or more of the elements can be replaced or altered without affecting the gestalt, in direct opposition to the traditional notion of type.42 40. Walter J. Bock and Gerd von Wahlert, "Two Evolutionary Theories: A Discussion," Brit. J. Phil. Sci., 14 (1963), 140-146; Leigh Van Valen, "On Evolutionary Theories," ibid., 146-152; G. S. Carter, "Two Evolutionary Theories, by M. Grene: A Further Discussion," ibid., 345-348; Marjorie Grene, "Two Evolutionary Theories: Reply to Dr. Carter," ibid., 349-351. oriented criticism, One philosopher who attempted a philosophically Edward Manier ("The Theory of Evolution as Personal Knowledge," Phil. Sci., 32 [1965], 244-252) was met with a harsh rebuff (Marjorie Grene, "Discussion: Mr. Manier's Theory of Evolution as Personal Knowledge: A Quasi-reply," Phil. Sci., 33 [1966], 163-164). 41. Ernst Mayr, "Agassiz, Darwin and Evolution," Harvard Library Bulletin, 13 (1959), 165-194; Mayr, Animal Species; Simpson, Principles, and This View of Life. 42. Marjorie Grene, "Positionality in the Philosophy of Helmuth Plessner," Rev. Metaphys., 20 (1966), 258.

258

Essay Review It seems to be this confusion over exactly what Grene intended to be arguing in her two evolutionary theories paper that occasioned much of the controversy. A later paper in which she criticized Fisher's mathematical version of evolutionary theory did not meet with any strong objections, since in this paper Grene explicitly states what it is she is attempting to establish and provides careful arguments in support of her thesis.43 In the original paper, biologists read her to be claiming that a false scientific theory could be preferable to a highly confirmed theory because of some preemptive epistemological considerations. They failed to see what these epistemological considerations could be or how they could outweigh considerations of empirical truth. To give Grene a run for her money, however, one might also suggest still another, more subtle possibility. The leading proponents of the synthetic theory of evolution are also antireductionists. It would be easy to understand why an essentialist would be antireductionist, but given the synthetic theory of evolution, it is difficult to see why these biologists are equally opposed to the possibility of reduction.44 From what has been said thus far, the reader might infer that philosophers have very little to contribute to biology. On the contrary, there are many aspects of the scientific endeavor to which philosophers could contribute. My complaint is that by and large they have not. A classic example of how ineffectual philosophers have been in communication with biologists is provided by the taxon-category distinction. J. H. Woodger spent most of his life doing what he called "turning the BooleFrege searchlight upon statements in biology." The differences between individuals, classes, and classes of classes is one instance in which such an effort could have resulted in considerable benefit for biology. The logical tools were available. The time was right. Just when Woodger was beginning his work in mathematical logic, biologists were beginning to subject the species concept to intensive reexamination.45 All it takes to appreciate how greatly the work of these biologists would have benefited from having more than an intuitive grasp of the differences between defining the name of a particular species (a class) and defining the name of the species category itself (a class of classes) is to read the literature of the period. But Woodger did provide just such an analysis in his book 43. Marjorie Grene, "Statistics and Selection," Brit. J. Phil. Sci., 12 (1961), 25-42. 44. This interpretation was suggested by Hugh Lehman and is borne out by certain comments in Grene, Knower, p. 185. 45. Theodosius Dobzhansky, Genetics and the Origin of Species (New York: Columbia University Press, 1937); Mayr, Systematics.

259

DAVID HULL

The Axiomatic Method in Biology.46 Unfortunately, his account went unnoticed. Finally, ten years later, when biologists did begin to point out the significant differences between what they called taxa and categories, the increase in clarity was substantial. It would be pleasant to be able to say that philosophers played a large role in the recognition and dissemination of this important logical distinction, but such does not seem to be the case. Biologists had to work out the appropriate distinctions and terminology for themselves. At this same time philosophers were producing parallel accounts.47 I have asked several of the biologists involved whether they found these accounts helpful. They replied in the negative. What was the reason for this failure in communication? Two factors combined to keep the work of these philosophers from having any significant impact on biology. In setting forth these factors, I do not mean to imply that all parties were equally guilty in every respect or that some of the blame cannot be laid at the feet of biologists. I really am not so much interested in fixing blame as in discovering the causes for this failure in communication. First of all, philosophers tend to exhibit what can only be described as disdain for the issues and distinctions which biologists find important. For example, in his 1950 paper John R. Gregg argues that species are classes, not individuals, and that the relation between an individual organism and the species to which it belongs is membership, not the partwhole relation.48 These issues were raised because two biologists had advanced independently the notion that species are as much concrete, spatiotemporal things as are individual organisms. On this score I think Gregg is right, but in his arguments Gregg seems almost willfully blind to the reasons these biologists might have had for making such an assertion. The point that they were trying to make was that species are not just sets, just collections of isolated individuals like the class of all things smaller than a breadbox. The members of a species are interrelated in numerous biologically significant ways, among which is spatiotemporal proximity. The ontologi46. J. H. Woodger, The Axiomatic Method in Biology (Cambridge [Eng.] University Press, 1937). 47. John R. Gregg, "Taxonomy, Language and Reality," Amer. Naturalist, 84 (1950), 419-435; John R. Gregg, The Language of Taxonomy (New York: Columbia University Press, 1954); J. H. Woodger, Biology and Language (Cambridge [Eng.] University Press, 1952); Beckner, Biological Way. 48. Gregg, "Taxonomy."

260

Essay Review cal questions of whether a class can be identified with its members, whether the class of all cells that compose an organism is identical to the class of all molecules that compose that organism, or, for that matter, whether the whole universe can be viewed as an organism, are irrelevant to the issues raised by these biologists. Gregg says that this problem "is a pseudo-taxonomic one which is resolved by reference to the semantic structure of language, and upon which no purely biological evidence (geographical distribution, interbreeding relations, etc.) has the slightest bearing whatsoever." 49 If so, then Gregg has misidentified the problem. The second factor which has contributed to the failure of communication between these philosophers and biologists is their method of doing philosophy of science-the formal reconstruction of biological statements in the notation of mathematical logic. There is a continuum with respect to the degree to which these "formalists" utilize mathematical logic. At one end such notions are used just for clarity and consolidation in the presentation of definitions. At the other end are those works in which the entire presentation is set-theoretical. This method of doing philosophy of biology has two drawbacks. The obvious one is that few biologists are familiar with the notation. One reason why Woodger's work has had so little impact on biology is that biologists cannot read most of his later work. But isn't this the fault of biologists? Isn't it up to them to learn set theory or symbolic logic so that they can reap the benefits of this large body of literature? The only answer that I can honestly give to these questions is "No." Formalists such as Woodger and Gregg have made some biologically significant points in their work, but few that could not have been made just as clearly without extensive use of these notations. Perhaps the discovery of certain logical distinctions was aided by the use of these techniques, but the results need not have been communicated in these same terms. The taxon-category distinction is a case in point. Too often the applications of mathematical logic to problems in biology give the impression that more or less commonplace ideas have been expressed in tiresome exactitude when they could have been conveyed more easily and more directly in a few sentences of plain English. The second drawback of the formalist method is that more often than not the method becomes the message. The substantive problem in biology which occasioned the formalization is forgotten as special, technical problems arise in the for49. Ibid., p. 426.

261

DAVID HULL

malism. The numerous papers which have been written to solve what has been called Gregg's paradox provide an excellent case in point. In his 1954 monograph, Gregg set himself the task of providing a set-theoretical reconstruction of the Linnaean hierarchy within the confines of extensional logic. (Woodger in 1952 had attempted a similar reconstruction.)50 The problem is that biologists make use of intensional notions in constructing their classifications. In fact, the distribution of the characters of the organisms being classified is a primary consideration in constructing biological classifications. Some biologists argue that it is the only consideration. As might be expected, Gregg's purely extensional reconstruction gave rise to paradoxes. Thereafter, a long series of papers appeared in which various devices were used to eliminate these paradoxes (or antinomies) without introducing the notion of intensions.51 But why? Even if one or more of these devices works, what relevance will this reconstruction have for biology? Taxonomists will continue to make their decisions on the basis of the distribution of characters among their specimens. The need for reconstructing the Linnaean hierarchy in intensional or modal logic will remain. All the effort expended in attempting to reconstruct the Linnaean hierarchy within the confines of extensional logic may be first-rate logic. It has little claim to being philosophy of biology. A second instance in which philosophers could have been of some service to biologists is afforded by the question, what is a character? This question has plagued genetics since the days of the one gene-one character hypothesis. Closely associated with are two inthis problem is the question of homology-when stances of a character to be considered instances of the same character and in what sense same? After the advent of evolutionary theory the answer has been that two characters are homologous if they are similar because of origin from a common ancestor. A decision as to whether or not two characters are homologous in an evolutionary sense requires recourse to all available evidence and to numerous scientific theories, including 50. Gregg, Language; Woodger, Biology and Language. 51. A. F. Parker-Rhodes, "Review of Gregg, The Language of Taxonomy," Philos. Rev., 66 (1957), 124-125; A. Sklar, "On Category Overlapping in Taxonomy," in J. R. Gregg and F. T. C. Harris, eds., Form and Strategy in Science (Dordrecht-Holland: D. Reidel, 1964), pp. 395-401; Leigh Van Valen, "An Analysis of Some Taxonomic Concepts," in Gregg and Harris, Form, pp. 402-415; C. J. Jardine, N. Jardine, and R. Sibson, "The Structure and Construction of Taxonomic Hierarchies," Mathematical Biosciences, 1 (1967), 173-179; John R. Gregg, "Finite Linnaean Structures," Bull. Math. Biophys., 29 (1967), 191-200.

262

Essay Review evolutionary theory. Because of the intricacy of the inferences involved in such decisions and the frequent paucity of evidence, some biologists have suggested that a different, more basic notion of homology-operational homology-should be substituted for the notion of evolutionary homology. At first, operational homologies were supposed to be something observed directly with no recourse to inferences or scientific theories. Gradually biologists have come to see that such units were quite ephemeral and of little scientific use and have expanded the concept to permit inferences on the basis of certain scientific theories-in fact, any scientific theory except evolutionary theory. When evolutionary interpretations are put on operational homologies, the result is evolutionary homologies. As the reader may have detected, the efforts of biologists to clarify the ideas of character and homology have been excursions into pure epistemology. Within the scope of a decade, they have relived the history of phenomenalism, operationism, and logical atomism. There is really no need for biologists to remake all the old mistakes and to explore every blind alley.652 Another place at which philosophers could have helped clarify the issues in a dispute between biologists is the disagreement over the problem of whether higher categories evolved first and then later diversified into lower categories, or whether species evolved first and then only later higher categories. Goldschmidt and Schindewolf incline toward the first view. Simpson and Mayr subscribe to the latter view.53 The major difficulty with this controversy is the logical crudity with which it is frequently expressed. Both Marjorie Grene 52. Robert R. Sokal and P. H. A. Sneath, Principles of Numerical Taxonomy (San Francisco: W. H. Freeman and Company, 1963); Donald H. Colless, "An Examination of Certain Concepts in Phenetic Taxonomy," Syst. Zool., 16 (1967), 6-27; Donald H. Colless, "The Phylogenetic Fallacy," ibid., 289-295; N. Jardine, "The Concept of Homology in Biology," Brit. J. Phil. Sci., 18 (1968), 125-139. Woodger had provided a similar analysis some years earlier ("On Biological Transformations," in W. E. le Gros Clark and P. R. Medawar, Essays in Growth and Form Presented to D'A. W. Thompson [Oxford: Clarendon Press, 1945]). Even though Woodger in Biological Principles, provided an excellent summary of the problems inherent in phenomenalism, a few biologists continue to argue that sense data or the like are the fundamental subject matter of science; for example, J. S. L. Gilmour, in Julian Huxley, ed., The New Systematics (London: Oxford University Press, 1940), pp. 461-474; J. S. L. Gilmour and S. M. Walters, "Philosophy and Classification," Vistas in Botany, 4 (1964), 1-22. 53. R. B. Goldschmidt, "Evolution as Viewed by One Geneticist," Amer. Scholar, 40 (1952), 84-98; Goldschmidt, Material Basis; Schindewolf, Grundfagen; Simpson, Meaning; Mayr, Animal Species.

263

DAVID HULL

and T. A. Goudge discuss this controversy at some length.54 Though they themselves do not make the fairly straightforward logical error at the bottom of the confusion, neither does anything to clarify the situation. The confusion lies primarily in the modes of expression of Goldschmidt and Schindewolf. On neither view can higher taxa (the term now in use) evolve first and lower taxa such as species evolve later. Perhaps Archaeopteryx evolved in one fell swoop. Perhaps its basic organization is so novel that it must be recognized as a new higher taxon regardless of any future developments. Even so, a new higher taxon has in no way evolved before a new lower taxon. Given the principles of classification agreed upon by both sides, it is logically impossible for a new phylum or family to evolve without a new species evolving simultaneously.55 The two sides are in disagreement, but this is not it. Their disagreement concerns one matter of fact and one matter of taxonomic strategy. The factual disagreement over the existence of large, abrupt changes in phenotypic makeup has been discussed previously. With respect to their differences on taxonomic strategy, Goldschmidt and Schindewolf want to classify entirely on the basis of over-all similarity and phenotypic gaps, regardless of the number and diversity of taxa which eventually exhibit this type of organization. Novelty alone guaranties a taxon a highcategory assignment. Simpson and Mayr classify on a multiplicity of principles. To maintain some kind of balance in a classification while still retaining a systematic relationship to phylogeny, principles of vertical classification must be tempered with those of horizontal classification. Thus a taxon with numerous included taxa is likely to be classified at a higher category level than its degree of divergence alone might warrant. Finally, no review of the philosophy of biology could possibly omit the two recent works devoted entirely to problems in the philosophy of biology-the Foundations of Biology by Felix Mainx and Morton Beckner's The Biological Way of Thought. As might be expected from the fact that his monograph is a contribution to the International Encyclopedia of Unified Science, Mainx emphasizes the verifiability criterion of meaningfulness and the unity of science. The errors and conceptual dangers which he most frequently points out in 54. Grene, "Two Evolutionary Theories"; Goudge, Ascent. 55. Amazingly, two authors who should know better are guilty of exactly the same confusion: Felix Mainx, Foundations of Biology (Chicago, Ill.: University of Chicago Press, 1955); Lancelot Hogben, Science in Authority (New York: W. W. Norton, 1963).

264

Essay Review biological works are attempts to pass off tautologies and metaphysical claims as empirically meaningful statements, and a tendency among biologists for conceptual realism. Both of these tendencies are worth bringing to the attention of biologists, but unfortunately for Mainx's treatment, he fails to reflect the increased sophistication of the positivist position which had occurred since its inception. For instance, he sees tautologies everywhere because he accepts a rather facile notion of the relationship between operations used to test the applicability of a term and the definition of that term. He says in one place, for example: If in the statement "The positive phototactic reaction of a Euglena is proportional to its light-requirement" the concept "light-requirement" is only testably defined by means of the establishment of the behavior under the stimulus of light, this is a tautologous statement of the above kind.56 Behavior under the stimulus of light is certainly neither logically nor physicaUly the only way of "testably defining" the concept "light-requirement." Hence, the statement is not tautologous. The same is true for most of the examples which Mainx gives. Mainx would have done well to have read Carl G. Hempel's earlier contribution to the same series on the foundations of concept formations in empirical science.57 A careful reading of this earlier monograph might also have suggested to him that the basic distinction which pervades his book serves only to frustrate his efforts to provide an adequate explication of the foundations of biology. This distinction is between order-analytic statements, which express the coexistence of characters, and causal-analytic statements, which express a succession of states in time. As time-honored as this distinction is, it just will not do as a characterization of the relationship between concept formation and theory construction in science. For example, Mainx recognizes three different viewpoints in biology-the morphological, the physiological, and the genetical. Although he warns the reader that these three viewpoints overlap somewhat, he makes it sound as if a morphologist could analyze an organism into organs and tissues independently of any knowledge of physiology, genetics, or evolutionary descent. Further, a taxonomist supposedly can then erect a classffication by means of the covariation of these characters, and this classification is equally 56. Mainx, Foundations, p. 5. 57. Carl G. Hempel, Fundamentals of Concept Formation in Empirical Science (University of Chicago Press, 1952).

265

DAVID HULL

neutral as far as scientific theories are concerned. Only afterwards can such theoretical considerations as the functions these structures perform or their evolutionary derivation be brought into play. This is the impression his exposition gives although he says that it must be "remembered that even in these elementary descriptive statements the beginning of hypothesis construction must be recognized." 58 Currently, biologists are carrying on an extensive debate over just these issues.59 Must the sequence of events in an actual scientific study be the same as those in Mainx's epistemological reconstruction? There are good reasons for maintaining that they cannot be. As Beckner has pointed out,60 the morphological characteristics which are used to produce a supposedly neutral classification are not purely morphological characters. The definition of "kidney," for example, necessarily presupposes knowledge of physiology. Perhaps a biologist might begin with visual clues, but what makes a structure a kidney is determined as much by its function as by its structure. A similar story can be told for the evolutionary derivation of a character, though the relevant evidence is more difficult to obtain. The gills of a fish and the gills of a crayfish are not the same character even though they perform the same function, both because of their structural differences and because of their differing phylogenetic histories. These various aspects of scientific teminology are interconnected in very complicated ways in the formation of concepts in science. One school of biologists is presently arguing that all these different features in biological terminology must be disintangled, especially any assumptions about evolutionary development. This task is, to say the least, ambitious. Reading Mainx's overly simplistic treatment will help these biologists appreciate the difficulties which they are likely to encounter very little. The story is quite different with respct to Beckner's book. The attention which Beckner's explication of organismic biology continues to receive from both biologists and philosophers is well-deserved. (The Biological Way of Thought has recently been reissued in paperback by the University of California Press.) What Beckner says is important and expressed only one major and pervasive exception. The clearly-with 58. Mainx, Foundations, p. 3. 59. Hogben, Science; Sokal and Sneath, Principles; Ernst Mayr, "Numerical Phenetics and Taxonomic Theory," Syst. Zool., 14 (1965), 73-97; Robert R. Sokal, J. H. Camin, F. J. Rohlf, and P. H. A. Sneath, "Numerical Taxonomy: Some Points of View," Syst. Zool., 14 (1965), 237-243; Colless, "Examination"; Colless,"Fallacy." 60. Beckner, Biological Way, pp. 110-131.

266

Essay Review main purpose of Beckner's book is to show that biology is an autonomous discipline with concepts and laws peculiar to itself. In general he tries to show that organization, directiveness, and historicity play a more crucial role in biology than in other sciences. Specifically, he attempts to show that three classes of concepts-polytypic, historical, and functional concepts-are characteristically biological in one sense and fully unique in biological theory in another. Beckner's explication of polytypic concepts affords a rare instance in which a philosopher has actually contributed significantly to the development of a biologically important notion, that of polythetic definitions.6' Biologists had been using the notion for some time and had discussed it with varying degrees of clarity, but Beckner also provided a general explication and a philosophical justification for it. The one glaring fault in Beckner's presentation is his introduction of two technical terms, "W-definition" and "E-definition," which he claims will be helpful in his exposition. A biologist attempting to read Beckner's book is confronted in the second chapter by a series of semiformal definitions. Most readers, if they go on at all, are tempted to skip this chapter and, as it turns out, with no great loss. His brief discussions of polytypic, historical, and functional concepts are expanded in later chapters devoted exclusively to them, and the notions of W- and E-definitions play almost no role in the ensuing pages. To be sure, Beckner periodically uses these technical terms in his subsequent discussions but to little consequence, since in most cases a term cannot simply be W-defined or Edefined. Instead they can only be E-defined with "preestablished criteria of adequacy for any W-definition," and Beckner leaves the nature of these criteria completely unexplicated. At the risk of seeming totally opposed to the formalist method of doing philosophy of science, I must point out that once again the extensive logical machinery which Beckner introduces serves to hinder rather than aid his exposition. Formalization may be an excellent way of working out problems in the philosophy of science. It is not a very good way of communicating the results of these endeavors. However, in spite of this flaw in presentation, Beckner's book remains the single major contribution of a philosopher to biology in over a decade. 61. See also, Woodger, Biology and Language; J. H. Woodger, "Taxonomy and Evolution," Nuova Critica, 3 (1961), 67-78; J. H. Woodger, "Biology and the Axiomatic Method," Ann. New York Acad. Sci., 96 (1962), 1093-1104; Douglas Gasking, "Clusters," Australas. Rev. Psych., 38 (1960), 1-36; Hogben, Science.

267

DAVID HULL

In conclusion, there are many things that philosophy of biology might be. A philosopher might uncover, explicate, and possibly solve problems in biological theory and methodology. He might even go on to communicate these results to other philosophers, to scientists, and especially to biologists. He might show what consequences biological phenomena and theories have for other sciences and for philosophy or to show what consequences other sciences and even philosophy have for biology. These are some of the things which philosophers of biology might do. With rare exception, they have not. What philosophy of biology is not? It must be admitted that thus far it is not very relevant to biology, nor biology to it.62 62. I wish to thank Ernst Mayr, Helen Heise, and Hugh Lehman for reading and commenting on this paper. Special appreciation is owed Marjorie Grene for doing the same, even though several of her own ideas are severely criticized. This paper will also appear in Synthese, vol. 20 (1969-70). It was presented at the 1968 Biennial Meeting of the Philosophy of Science Association, Pittsburgh, Pennsylvania, October 12, 1968; Philosophy Department, Washington University, St. Louis, Missouri, Otcober 20, 1968; Philosophy Department, University of Guelph, Guelph, Ontario, Canada, February 6, 1969. The paper was prepared under grant GS-1971 of the National Science Foundation.

268

Special Historical Note

Darwin's Questions About The Breeding of Animals (1839) PETER J. VORZIMMER Temple University, Philadelphia, Pennsylvania

The eight-page quarto pamphlet Questions About the Breeding of Animals (see Appendix to this essay) is a rare and, until recently, unpublished work of Charles Darwin.' Only a single copy, that in the Library of the British Museum (Natural History), was known until recently, when a copy was located in a collection of documents at the University Library, Cambridge, England. The privately printed pamphlet, distributed by Darwin himself, bears no printer's mark-only the signature "C. Darwin-12, Upper Gower Street, London." It consists of twenty-one numbered paragraphs which contain forty-eight questions.2 The printed sections (of 9 cm width) have been placed to the left of each page, presumably to leave the even wider margins created for the inclusion of annotated responses. The pamphlet bears no date. Dating of the Questions: One of the only references to the Questions that can be found is that made by Darwin in the following extract from his autobiography, in which he records that, upon opening his first transmutation notebook in July, 1837, "I worked 1. The present paper was begun in the Fall of 1967, finished that winter, and submitted in the Spring of 1968. I am gratefully indebted to Dr. Richard B. Freeman of University College, London, for first providing me with a copy of the Questions-as well as a good deal of relevant informationwithout which this paper would not be possible. A Facsimile of the Questions, edited and with an Introduction by Sir Gavin de Beer, has been published as the Sherborn Fund Facsimile No. 3, December 1968, Society for the Bibliography of Natural History-Ed. 2. Though only forty-five question marks appear in the text, it can be clearly seen, from both context and syntax, that a total of forty-eight questions has been asked. The three unmarked questions are the first questions in paragraphs 8, 10, and 18.

269

PETER

J. VORZIMMER

on true Baconian principles, and without any theory collected facts on a wholesale scale, more especially with respect to domesticated productions, by printed enquiries, by conversation with skilful breeders and gardeners, and by extensive reading." 8 Two facts allow the bracketing of the pamphlet's publication between two specific dates. First, as its final page bears the printed address "12 Upper Gower Street"-Darwin's home as of December 31, 1839-this is likely to be the earliest possible date of its publication. Second, as one set of replies to the Questions was clearly dated May 5, 1839 by one respondent and the recently-found second copy of the pamphlet is dated by its recipient "10 May 1839", it may be safely concluded that the pamphlet was printed not later than the opening days of May, 1839.4 A glance at Darwin's personal "Journal" covering the chronology of his activities in 1839 reveals no period of time before the end of the first week in March sufficient for the preparation of the Questions. The entire month of January was spent in preparations for his wedding on the 29th, nearly all of it in Shrewsbury and Maer. At the beginning of February the journal records the recommencement of the Coral Reefs paper. The next entry reads "First week in March Earthquake paper then a little work on Species & then Coral paper.", which is followed by "End of March & nearly all April Coral paper." The middle two weeks of March and the "little work on Species" are undoubtedly those in which he conceived of the idea of the broader method of soliciting factual information represented by the Questions. Reference to Darwin's fourth "Transmutation" notebook (covering enteries from October, 1838 to July, 1839) reveals that the first entry on the subject of animal cross-breeding in 1839 appears after the date " March 5".5 Numerous notes on the crossing of varieties and species are found on the following eight pages, ending with the following passage dated March 12: Varieties are made in two ways-local varieties when whole mass of species are subjected to the same influence & this would take place from changing country: but greyhound race3. F. Darwin, ed., The Life and Letters of Charles Darwin, 2 vols. (New York: Appleton, 1898), I, 68. Italics mine. 4. Personal communication from P. J. Gautrey, Manuscript Librarian, University Library, Cambridge. Details of these and other items found in the Robin Darwin Deposit at the Library will appear in a paper by P. J. Gautrey and R. B. Freeman in the Journal of the Society for the Bibliography of Natural History, London. 5. "Darwin's Notebooks on Transmutation of Species: Part IV-Fourth Notebook (October 1838-July 1839)," Edited and with an introduction by Sir Gavin de Beer, Bull. Brit. Museum (Nat. Hist.), Historical Series, II, no. 5,p. 171.

270

Darwin's Questions About the Breeding of Animals (1839) horse & poulter Pidgeon have not been thus produced, but by training & crossing & keeping breed pure-and so in plants effectually the offspring are picked & not allowed to cross.Has nature any process analogous( ? ) -if so she can produce great ends-But how (? )-even if placed on Isld, if &c, &c,make the difficulty apparent by cross-questioning."6 It should be mentioned here that the transmutation notebooks up to this date show that Darwin had sought much information on various subjects via a number personal letters querying more than a dozen fellow naturalists. It therefore seems likely that an increasing number of questions on this particular subject at this particular time convinced Darwin that a printed list of queries would be a handier method of solicitation. Thus, on or about March 13 Darwin composed the Questions and sought out a private firm for their printing. As Darwin left for Maer on April 26, not to return to London until the May 20 (finding the replies of May 5 and 10 awaiting him), it would seem that his pamphlets must have arrived from the printers some six weeks after his ordering them, on about April 24, and that at least a number of them were dispatched by him in the two days before he left on his holiday. How the Questions Fits into the Schedule of Darwin's Activities (from the landing of the Beagle in October, 1836, to the first 'Species Sketch' of June, 1842): After his reunion with family and the disposal of his collections, Darwin's principal task upon his return was the writing up of his observations into a publishable Journal of Researches. Though he had made attempts at this writing during his stay in Cambridge, Darwin did not begin a really determined effort to write until he moved down to London in March 1837. This task occupied his time almost exclusively to mid-July. It was during this latter period that detailed consideration of the phenomenon of adaptation and its causes fortified the early glimmerings of a belief in the transmutation of species. Though still far from "convinced" that such had taken place-and, at that time, without any notion of a process acting towards that end-Darwin opened his first transmutation notebook. It was, as he said, "to record any fact which might bear on the question." Between the time he opened this first notebook in July 1837 and concluded the fourth and last in July 1839, two significant ideas served to change his original approach. Darwin had begun the first notebook to simply record relevant facts with the leading 6. Ibid., p. 173.

271

PETER

J. VORZIMMER

idea, borrowed from Lyell, that the key to past biological change might be found in whatever sources of change could be presently observed. It was in following this line of thought in his initial pursuit of facts that Darwin first came to see that the change effected through selection, as practiced by plant and animal breeders, was the key to past change. It was the realization of this concept of change that gave an added dimension to his fact gathering. The second alteration to Darwin's original Baconian approach came through his reading of Malthus (September 28 to October 3, 1838). Malthus provided the basis for Darwin's seeing the relevance of factors leading to the struggle for existence to his concept of transmutation by selection. Thus, from the opening to the closing of the notebooks, Darwin had expanded his view from a belief in the fact of transmutation, to the addition of a means by selection, to, finally, the requisite force of competitive struggle. By the addition of this last element, his hypothesis (paraphrasing Darwin's own words) "had at last become a theory by which to work." After closing the last notebook, however, there remained several other obligations to be attended to-principally his impending marriage and the unfinished zoology and geology of the Beagle-before he could spread his fact-gathering net more widely. The now-resolved theory would require the collecting of relevant information along more specific, and thus more formal, lines. It would seem that it was just such an attitude that precipitated, in 1839, the Questions on the Breeding of Animals. How the Questions Fits into the Development of Darwin's Theory (to the first 'Species Sketch' of 1842): Quite clearly, the Questions should be looked upon as more than a formalized interrogation of experts along certain lines. For the actual queries are nearly all leading questions. That is, they are framed in such a way as to elicit specific responses which may or may not confirm an existing hypothesis. This is borne out by the conditionalinterrogative ("If this . . . will this ... ?") syntactic arrangement of nearly all the queries. Thus, the questions can be seen as very much the deductive manifestation of Darwin's own view of how the phenomena of hereditary transmission might bear upon his transmutation hypothesis. As such, the Questions proves a valuable document in analyzing the development of Darwin's evolutionary thought during that period between his first clear apprehension of the theory of natural selection, and his first attempt to demonstrate its action in writing. While the period from Darwin's landing with the Beagle (Octo-

272

Darwin's Questions About the Breeding of Animals (1839) ber 1836) to his reading of Malthus (October 1838)7 is of far greater significance in the actual formulation of Darwin's theory, the interval between Malthus and the first "Species Sketch"- a period of nearly three years-remains of considerable interest. This is partly because we are bound to ask ourselves the question why, if the process of natural selection was at last clearly formulated in Darwin's mind after reading Malthus in 1838, did he delay in writing up his account of how that process effected the transmutation of species? From within his writings, Darwin provides us with but a single clue: "Here then [after seeing the relevance of Malthus' argument for his selection hypothesis] I had at last got a theory by which to work; but I was so anxious to avoid prejudice, that I determined not to write even the briefest skech of it." 8 The prejudice seems to be the bias that Darwin felt a theory might impose upon the impartial collection of all relevant facts, for or against the theory. To gather facts first, before beginning the detailed elaboration of any theory which is to transcend them, was Darwin's concept of true scientific method. One should, as he said of himself, "work on true Baconian principles, and without any theory collect facts on a wholesale scale." 9 It remains, however, that Darwin did have a clear theory and, notwithstanding a desire for impartiality of method, this dictated the very specific nature of the classes of facts that he was seeking. A good measure of the reason for his delay lay elsewhere. Despite his obvious satisfaction in at last having a clear idea of his transmutative process, Darwin had been troubled by the diverse and contradictory implications of the material he had thus far collected on breeding. The notebooks clearly show Darwin's intention of making his case for evolution in the organic past by projecting back into it those presently observed processes of modification. As a result, he directed his fact-collecting almost exclusively along empirical lines. It had also become obvious to him that any would-be critic could equally apply any presently observable limitations against his account as to what might have transpired in the past. Darwin had come across many such cases, and the notebooks are crammed with such anticipated difficulties: the projected limit to variation; the tendency toward constancy of form seen as a natural function of generation; the tendency of new and hybrid forms to revert to original type; the assumption that old, estab7. win, 8. 9.

For a detailed account of this period, see Peter J. Vorzimmer, "DarMalthus, and the Theory of Natural Selection," J. Hist. Ideas, 1969. F. Darwin, Life and Letters, I, 68. Ibid.

273

PETER J. VORZIMMER

lished forms dominate over any newer ones; the problems attached to a blending inheritance. All of these had been gleaned from the existing literature. As a result of the more than two years of pooling information Darwin had, by the second half of 1839, a clearer and more detailed view of how selection acted, and a rough conceptual picture of the process of hereditary transmission on which it so manifestly depended. It appeared to Darwin that the stamp of real change upon an existing form must come about through the cumulative addition of small increments of individual variation. Larger increments seemed to either precipitate reversion, reduce fertility, decrease fecundity, or were somehow more susceptible to the prepotent effects of more well-established characters of the same kind. Though he recognized prepotency (as dominance was called) as an obvious and common exception to blending, both blending and reversion remained for him two phenomena the effects of which seemed to run counter to permanent modification of specific type. Darwin had to acknowledge a general tendency in the organic world toward the maintenance (and therefore relative constancy) of specific type. For him, however, there was a big difference between absolute and relative fixity of species. In other words, it may be difficult to permanently modify a species, but not impossible. So it was that Darwin came to feel that, under certain conditions, species could be permanently modified. This process he saw as beginning with individual variation; creating, as it were, a new subvariety in a single individual. The compounding of this first and single degree of difference, under the proper conditions of time and circumstance, to that which would mark a significantly large population as a new specific form, was speciation as Darwin envisioned it. Darwin saw this process of compounding variation as taking place somewhere deep within the individual. In the early notebooks there is no suggestion of an actual anatomical site where such change takes place (and is recorded for future transmission). He does, in several places, refer to the "blood" in this capacity, but this cannot, at this time, be taken in a literal sense. The blood thus became for Darwin the figurative repository for the germinal "constitution" of an organism. As such, it functioned as the basis for hereditary transmission. Reversion itself (as well as the case of a waning of prepotency) indicated that some sort of storehouse of unit characters persisted through one or more succeeding generations. It was not to be until many years later that Darwin would address himself to the question of

274

Darwin'sQuestionsAboutthe Breedingof Animals (1839) a possible anatomical location for the germinal constitution. By the second half of 1839, however, Darwin did have a limited hypothesis regarding the nature of inheritance. Beginning in the second notebook, we find Darwin talking of variations that "become more & more impressed in blood with time",'0 and how a "variety when long in blood gets stronger and stronger, so that though by great effort one unlike can be produced." 11 By the opening of the third notebook he offered the maxim "What has long been in blood, will remain in blood,converse, what has not been, will not remain." 12 But he must have quickly seen the decidedly negative plications that such would have held for transmutation and, a few pages later, he altered this to "An animal in either parent cannot transmit to its offspring any change from the form which it inherits from its parents stock without it be small & slowly attained. N.B. The longer a thing is in the blood, the more persistent any amount of change & shorter time less so." 13 Darwin continued, giving his view of the transmission of characters involved in a cross: When two animals cross, each sends his own likeness & the union makes hybrid, in fact the parents beget child like themselves. Expression of countenances, organic diseases, mental disposition, stature, are slowly obtained & hereditary; if the change be congenital (that is most slowly obtained with respect to that individual) it is more easily inherited,-but if the change be in blood long, it becomes part of [the] animal, -by a succession of generations, these small changes become multiplied, & great change be effected.'4 By the end of this third notebook we find Darwin applying the above view to a case involving a decidedly new (that is, nonderivative) character-and in the face of blending inheritance. If A & B be two animals which have some peculiarity for the first time, & if all their offspring inherit the same peculiarity in lesser degree & theirs again in lesser degree-now if the second race both have this peculiarity strongly; they transmit with same force as first pair, but to this tendency is added the 3rd tendency from first pair. -Now if two of third pair of same peculiarity breed they will have the same influence as first pair 10. "Darwin's Notebooks ...," 1. 12. 13. 14.

N.B. II:33.

Ibid., II: 136. Ibid., III:13. Ibid., III:16. Ibid.

275

PETER J. VORZIMMER

+ tendency they inherited from second pair, + the influence they themselves inherit.'1 Confusing, perhaps, but not entirely unclear. It seems that, for Darwin, not only could the new variation itself be transmitted, but also the original predisposition producing such a variation ab initio. At first glance it would appear that Darwin is here bringing to bear a rather ad hoc assumption to counter the implied effects of a blending inheritance. And this may well be the case if one looks for some contemporary justification. However, it is quite understandable if one sees that Darwin felt that not only the variation, but that unique set of constitutional peculiarities which predispose any organism bearing them to produce such a variation, was also inherited. The important difference between Darwin's and the modem view is that Darwin looked upon the two as not only separate but additive-and therein lies the real distinction. Such was the shape of Darwin's view of inheritance relative to transmutation up to the time he decided to seek the further information requested in the Questions. Despite the apparent orderliness of their presentation, the forty-eight questions set forth in the pamphlet were not grouped together in a systematic fashion. The following table provides a more systematic outline of the nature of the inquiry: I.

II.

Darwin seeks factual information on basically three types of crosses: Species x Species; Species x Variety; and Variety x Variety. The following considerations (with regard to the nature of the actual crossing elements) super-imposed on the above crosses, add several permutations: Male x Female; Wild x Domesticated; Established x New or Recent; Closely Related x Disparate Forms. The information sought: 1. Is the cross sterile or fertile and/or is fecundity affected? 2. If fertile, is the result a dominance or blending of characters? 3. If dominance, which parent form and which sex? 4. If blending, in subsequent generations is intermediate forn maintained, or does it revert to one of the originals? 5. Second inter-cross (F2) generation: 1-4 above? 6. Second back-cross generation: 1-4 above? 7. Are the following inherited? a. fertility and/or fecundity?

15. Ibid., III: 169.

276

Darwin's Questions About the Breeding of Animals (1839) b. new and continuous variations? 16 c. new and discontinuous variations? d. predisposition to variation? e. direct effects of conditions? f. effects of habit or use? g. effects of disuse, mutilation, or abuse? h. instincts or behavior? 8. If inherited, in above crosses: 1-4 above? All of the questions, save the 47th in paragraph 20, can be seen to fit somewhere in the above scheme. As such they represent quite a comprehensive interrogation on the subject of crossing and the transmission of characters involved. Darwin's own continuing readings and later researches, together with his assimilation of the new information which he received, not only provided the very basis of his theory of natural selection but, in the years after the Origin, were to result in his publishing a theory of inheritance of his own.17 Thus, the Questions about the Breeding of Animals is one of the earliest of Darwin's writings to give some indication of the direction in which he was moving.

Appendix QUESTIONS ABOUT THE

Breeding of Animals 1. If the cross offspring of any two races of birds or animals, be interbred, will the progeny keep as constant, as that of any established breed; or will it tend to return in appearance to either parent? Thus if a cross from the Chinese and common pig be 16. For Darwin, the term "continuous" was applied to a variation which was slight and discernible as qualitatively relating to pre-existing structure; "discontinuous" referred to a saltation or a qualitatively unrelated structure. 17. For further information see R. C. Olby, "Darwin's Manuscript of Pangenesis," Brit. J. Hist. Sci., 1:258-269 (1963), and P. J. Vorzimmer, "Charles Darwin and Blending Inheritance," Isis, 54:371-391.

277

PETER J. VORZIMMER

interbred, will the offspring have a uniform character during successive generations, that is, as uniform a character, as the purebred English or Chinese ordinarily retains? Thus, again, if two mongrels, (for instance of shepherd dog and pointer) which are like each other, be crossed, will the progeny, during the succeeding generations retain the same degree of constancy and similarity, which might have been expected from pure-bred animals? Is it known by experience, that when an attempt has been made to improve any breed by a cross with another, that the offspring are apt to be uncertain in character, and that unusual care is required in matching the descendants of the halfbred among themselves in order to keep the character of the first cross? Always please to give as many examples as possible, to illustrate these and the following questions. 2. If by care, the character of half-bred animals (mongrels or hybrids) be preserved through some two, three, or more generations, is it then generally found, that the character becomes more permanent, and less care is required in matching the offspring? If this be so, how many generations do you suppose is requisite to forn a mixed race, into what is ordinarily termed a permanent variety or well-bred race? 3. Supposing some new character to appear in a male and female animal, not present in the breed before, will it become more permanent, and less likely to disappear, after it shall have been made to pass through some successive generations, by picking out and crossing those of the offspring, which happened to possess the character in question? 4. In crossing between an old-established breed, or local variety, which from time immemorial has been characterized by certain peculiarities, or the animal in its aboriginal state, with some new breed, does the progeny in the first generation take more after one than the other? or if not so, is the character of one more indelibly impressed on the successive generations than that of the other? Or, which is the same question, is the breed of the parents of more consequence, when a breeding animal is wanted, than when merely a fine animal is wanted in the first generation? The effect should be observed both in a female of the old race crossed by the new, and a female of the new crossed by a male of the old; for otherwise the greater or less preponderance of the peculiarities in the progeny might be attributed to the power of the sex, thus characterized in transmitting them; and not to the length of time the breed had been so characterized. Thus to take an extreme example, we may presume that an Australian Dingo is an older breed than a pug-dog: if both were crossed with Spaniel bitches, would the litter in the one case more

278

Darwin's Questions About the Breeding of Animals (1839) resemble the Australian, than in the other case the pug: and however this may be, would the pug, or Australian character be most persistent under similar circumstances in successive generations? How would this be in the various breeds of cattle? Thus if a Bull (or a cow) of a breed which had long been known to have been white with short horns, were crossed with a black cow with long horns, (or Bull, if the first were a cow) which had accidentally sprung from some breed, not thus characterized, would there be any marked leaning in the character of the calves to either side; or would successive generations have a stronger tendency to revert to one than the other side? Please to mention in detail any instances you may be acquainted with. 5. What would the result be, in the foregoing respects, in crossing a wild animal with a highly domesticated one of another species, supposing the half cross to be fertile? Thus if a fox and hound were crossed with pointer-bitches, what would the effect be both in the first litter and in the successive ones of the halfbred animals? To form a judgment on this latter point, the subsequent crosses in each case should be relatively the same; thus the half-bred fox and half-bred hound should be recrossed with the pointer, or with some other, but the same breed. 6. Where very different breeds of the same species are crossed, does the progeny generally take after the father or mother? 7. When two breeds of dogs are crossed, the puppies of the same litter occasionally differ very much from each other, some resembling the bitch and some the dog. In the mule between the ass and horse, this great variation does not appear commonly to occur. Do you know any cases, where two varieties have been often crossed, and mongrels have been uniformly produced similar to each other within small limits, and intermediate between their parents? And on the other hand, do you know of hybrids, between such animals as are generally considered distinct species, varying in this manner? 8. When breeds extremely different (as the grey-hound and bull-dog, the pouter and fantail-pigeon,) are crossed, are their offspring equally prolific, as those from between nearer varieties (such as from the grey-hound and shepherd-dog). Is the halfbred Chinese pig as prolific as the full-bred animal? Does a slight cross increase the prolifickness of animals? 9. Do you know of instances of any character in the external appearance, constitution, temper, or instinct, appearing in halfbred animals, whether mongrels or hybrids, which would not be expected, from what is observable in the parents? 10. In those rare cases, where hybrids inter se have been productive; have the parent hybrids resembled each other; or have

279

PETER J. VORZIMMER

they been somewhat dissimilar, partaking unequally of the appearance of their pure-bred parents. Also, what has been the character of the progeny of such hybrids? 11. When wild animals in captivity, cross with domesticated ones, is it most frequently effected by means of the male or female of the wild one? 12. Amongst animals (especially if in a free, or nearly free condition,) do the males show any preference, to the young, healthy, or handsome females? or is their desire quite blind? 13. Where a female has borne young to two different breeds or kinds of animals, do you know of any instances, of the last born partaking of any part of the character of the first born, and to what extent? 14. When a female of one breed has been crossed by a male of another breed several times, do the last-born offspring resemble the breed of the father, more than the first-born, and therefore are they more valuable in those cases, where the peculiarity of the father is desired? 15. Do you know instances of any peculiarities in structure, present for the first time in an animal of any breed, being inherited by the grand-children, and not by the children? It cannot be said to be inherited without it appear in more than one of the grand-children, or without it be of an extremely singular nature; for otherwise it ought to be considered as the effect of the same circumstances, which caused it to appear in the first case. 16. What are the effects of breeding in-and-in, very closely, on the males of either quadrupeds or birds? Does it weaken their passion, or virility? Does it injure the secondary male characters,-the masculine form and defensive weapons in quadrupeds, or the plumage of birds? In the female does it lesson her fertility? does it weaken her passion? By carefully picking out the individuals most different from each other, without regard to their beauty or utility, in every generation from the first, and crossing them, could the ill effects of inter-breeding be prevented or lessened? 17. Where any animal whatever (even man) has been trained to some particular way of life, which has given peculiarity of form to its body by stunting some parts and developing others, can you give any instances of the offspring inheriting it? Do you know any such case in the instincts or dispositions of animals? If an animal's temper is spoilt by constant ill usage, or its courage cowed, do you believe the effect is transmitted to its offspring? Have any cases fallen under your observation, of quadrupeds (as cats or pigs, &c.) or birds (fowls, pigeons, &c.) born in this country from a foreign stock, which inherited habits

280

Darwin's Questions About the Breeding of Animals (1839) or disposition, somewhat different from those of the same variety in this country? If removed early from their parents, there are many habits, which we should be almost compelled to believe were inherited, and not learnt from them; and if transmitted to any half-breed we should feel sure of this. 18. Can you give any detailed account of the effects on the mind, instincts or disposition of the progeny, either in the first or in the succeeding generations from crossing different breeds, (for instance carrier and tumbler pigeons, grey-hounds and spaniels) or different species (as fox and dog.) Do they show an aptness to acquire the habits of both parents? Or do they partake strongly of the habits of one side, (if so, which side?) with some peculiarity showing their hybrid origin? Or do they entirely follow one side? 19. Can you give the history of the production in any country of any new but now permanent variety, in quadrupeds or birds, which was not simply intermediate between two established kinds? 20. Do you know any cases of different breeds of the same species, (as of dogs &c.) being differently affected by contagious or epidemic diseases, and which difference cannot be attributed merely to a greater vigour in the one breed than in the other? In countries inhabited by two races of men, facts of this kind have been observed. 21. All information is valuable, regarding any crosses whatever, between different wild animals, either free or in confinement, or between them and the domesticated kinds; equally so between any different breeds of the same species, especially the less known kinds, as Indian with common cattle, different races of Camels, &c. Please to state all or any particulars, for what the cross was made and whether it is habitually made; whether the female had offspring before; whether she produced as many of the half-breed at one birth, (if more than one be produced) as she probably would have done of the pure breed; whether the progeny were fertile inter se, or with their parents whether they resembled one stock more than the other and in what respects, and which; and whether the favoured side was the male or female. State, if known, whether the progeny differ when stock (A) is the father and (B) the mother, and from what it does where (A) is the father and (B) mother. If the half-bred are fertile, inter se or with the parent stock, describe the offspring whether like their parents and all like each other, or whether they revert to either original stock, or whether they assume any new character? C. DARWIN 12, Upper Gower Street, London.

281