I was once asked to be the commentator at a Philosophy of Science Association meeting on a paper by a famous epistemologist on the difference between theory confirmation in physics and in evolutionary biology. At first I was worried that I would not understand his paper, but after I read it, what worried me was that I did understand it.

Biology has been an embarrassment to the philosophers of science and so, for a very long time, most of them pretended it did not exist, or denied that it was real science at all. The great guru of modern philosophy of science, Sir Karl Popper, declared in The Poverty of Historicism that despite the claim of its practitioners to be scientists, the theory of evolution is not a science at all, but what he elsewhere called a “metaphysical research program.” He has since changed his view to some extent, but the unease that the study of biology creates in the hands of an older generation of philosophers remains. During the last twenty years the philosophy of science has become more and more, in the hands of a younger generation of philosophers egged on by a few of their more insightful elders like Marjorie Grene, the philosophy of biology. The historical irony is that with cries of “No Poppery!” they have nearly all chosen to study evolutionary biology.

At the heart of the philosophers’ preference for physics over biology is the question of uncertainty. Science is supposed to be a study of what is true everywhere and for all times. The phenomena of science are taken to be reliably repeatable rather than historically contingent. After all, if something happens only on occasional Tuesdays and Thursdays, popping up when one least expects it like a letter from the IRS, it is not Science but History. So, philosophers of science have been fascinated with the fact that elephants and mice would fall at the same rate if dropped from the Tower of Pisa, but not much interested in how elephants and mice got to be such different sizes in the first place. In terms of the formal calculus of propositions, the statements of science are supposed to be so-called “universally quantified” statements of the form—

For all x, if x is A then x is B

—rather than historical statements, which are only existentially quantified—

There exists an x such that x is B.

The point, Popper tells us, is that the first kind of statement can always be falsified, by finding a single example that does not obey the rule, while we can never disprove the second kind because we may have accidentally missed the cases that agree with it. So the first kind of statement is what characterizes a science, while the second kind is just storytelling.

Philosophers have not been alone in claiming that science must be about universals. In a fit of severe physics envy, biologists, even evolutionary biologists concerned with the history of life on earth, have tried to construct their science as a set of universal “laws,” acceding to the general intellectual disdain for the merely particular. Beginning with Darwin’s notion that there is a universal “struggle for existence” that is the engine of evolutionary history, through R.A. Fisher’s “Fundamental Theorem of Natural Selection” of 1930 that claimed (over-optimistically, as it turned out) to be an equation for making general predictions, to the presentday assertion that virtually all DNA evolves at a constant clock-like rate, evolutionists seeking fame and fortune have invented their own versions of the law of gravity. Wonderful Life is Stephen Gould’s powerful reassertion of the importance of historical contingency in evolutionary explanations. As its subtitle suggests, it is not the history of nature that is at issue in Gould’s book since that is the concern of all evolutionary biology; the question is whether that history is the necessary unfolding of universal laws of life, whether the race is to the swift, the battle to the strong, and riches to the wise, or whether, as Gould argues, time and chance happeneth to all. In the process, Gould shows again that he is unique among evolutionists in the subtlety and depth of his thinking about the history of life.

To understand the problem of historical contingency in evolution, we must distinguish between two kinds of statements, both of which appear at first sight to be universal assertions. Consider two favorites of philosophers:

All planets move in ellipses


All swans are white.

The first of these is truly universal because it applies not only to all planets that actually exist but to all planets that could conceivably exist, because the shape of planetary orbits is a necessary consequence of the relations of the motions of objects in a gravitational field. All physicists are agreed upon this, and that is why philosophers of science have so loved physics.


On the other hand, the assertion that all swans are white. Some biologists would say that if all swans are white, it is because the laws of development or the exigencies of natural selection make it impossible, in principle, for black swans to exist. In this respect, swans are like planets and biology is like physics. Other evolutionists would say that “swans” is simply the name of a finite collection of historical objects that may happen all to be white, but not of necessity. The black ones have just not come along yet, but could perfectly well do so at any time.1 Finally, we might simply say, “If it isn’t white, it isn’t a swan,” and leave the matter as a purely taxonomic one. While the last ploy may seem a mere trick, it plays an important role in the story of the Burgess Shale, although Gould barely mentions it in his book.

The history of evolutionary theory has been the history of the struggle among these various views of swans. On one side are arch-adaptationists who see every pimple on the body physical and the body politic as a necessary consequence of natural selection, leading to optimal adaptation of the organism to the problems of life. For adaptationists, human intelligence, social organization, and control of external nature were inevitable consequences of long-term natural selection forces and could have been foresee by an intelligent Martian watching terestrial evolution from afar. The argument is that natural selection favors those characteristics that increase the average rate of offspring production, and so eventually, traits that maximize flexibility and the ability to manipulate nature will appear. Influential works such as E.O. Wilson’s Sociobiology and Richard Dawkins’s The Selfish Gene are hautes vulgarisations of this major determinist tradition on evolutionary theory. According to Gould, this element in evolutionary explanation has grown in the last half century, in a “hardening of the evolutionary synthesis,” to become the dominant mode, replacing an earlier, more pluralistic theory.

The problem with the determinist view of evolutionary necessity is that it does not tell us why all animals (and plants, for that matter) are not as intelligent and well organized socially as we are. Bad luck? But then there are important contingencies in evolution, and swans do not have to be white after all. Curiously, over the half century during which Gould sees the evolutionary synthesis as being “hardened” around optimizing natural selection, there has been a proliferation of technical literature of alternative explanations that do not depend on direct natural selection for traits. They all emphasize historical contingency.

First, real populations of organisms are finite, and as a consequence no generation is a precise statistical copy of its parental generation. Variable traits fluctuate in frequency from generation to generation and eventually one or another form comes to characterize the species by sheer chance. In the literature of probability theory, this is known as the “drunkard’s walk,” by analogy with the drunk who emerges from the bar in the middle of the block and who staggers randomly left and right until he falls off the curb at one end of the block or the other.

A second consequence of the finiteness of population size is that most newly arisen mutations are lost to the species within a few generations, even mutations that are favored by natural selection. Had we but world enough and time, the favorable mutations would arise again and again until finally one took hold. But no species lives forever. More than 99.99 percent of all species that have ever existed are already extinct, and the average length of life of a genus of carnivores had been only about ten million years, a fraction of 1 percent of the history of life.

Third, there is more than one way to skin a cat. Identical selective forces do not lead to identical ends but depend on the actual genetic state of the species when selection begins. It is a well-known property, even of “deterministic” physical systems, that one cannot predict where a system will end up if one does not know where it started. The day before I started to write this article, I traveled eighty-five miles west, then seventeen miles north, and then again eight miles west, but from that information alone, you cannot guess that I wrote it in Marlboro, Vermont. So it is probable that one and two-horned rhinoceroses exist, not because it is better to have one horn in India and two in Tanzania and Sumatra, but because the same force of natural selection produced different solutions to the same problem, beginning with somewhat different genetic materials in the three parental species.

Fourth, genes that influence traits are organized on physical bodies, the chromosomes, that tend to be inherited as a single unit, so that different genes do not have independent fates in inheritance. If natural selection favors a particular trait and increases the incidence of genes that influence the trait, other genes that are by chance mutation on the same chromosome will also be increased despite their total irrelevance to the selective forces. This “genetic hitchhiking” results in chance selection of a trait without selection for it.


All of these phenomena, and the list is not exhaustive, point to the role of chance and contingency in the outcome of evolutionary processes, even when the deterministic force of natural selection is operating. It follows that if evolution could be run all over again from the beginning, a completely different set of organisms would arise. We do not know of totally rejected evolutionary experiments, which is what made Karl Popper despair of evolution as a science, but we occasionally get glimpses of long-extinct assemblages of organisms that appear to be very different from any others that we know. Evolution occasionally gets off on a different track, and as with God’s failed attempt to produce a lasting set of angels the first time around, there is a general housecleaning. The Burgess Shale is a rock formation that contains a record of these fallen angels and provides the ammunition for Gould’s struggle with the determinist forces.

The Burgess Shale is a deposit of fossil-bearing rock about 530 million years old in Yoho National Park in the Canadian Rockies.2 In this one very small quarry, indeed in a layer only seven to eight feet thick, is a collection of invertebrate fossils that shows more diversity of body plan than is encompassed by all living marine invertebrates. One of the puzzling features of the fossil record is that all the basic body plans of the invertebrate animals now known appeared almost instantaneously (geologically speaking) at the beginning of the Cambrian Period, about 550 million years ago. Before this “Cambrian explosion” of forms, there was practically nothing. The Burgess Shale organisms encompass all these body plans, and they include a number that are no longer seen and are not known from other deposits of fossils. The accompanying picture illustrates some of the “weird wonders,” as Gould calls them, that have been reconstructed from their squashed remains in the rock. Among the most shocking to zoologists are Opabinia with its five eyes and front hose, a sort of vacuum cleaner with paddles, Wiwaxia, made up of overlapping plates from which issue long spines, and Anomalocaris, whose circular mouth, jointed front arms, and leaflike appendages make it unlike anything else ever seen. Yet other undiscovered fossil beds may show us yet a new set of types still unsuspected. All the organisms now alive show only a fraction, and perhaps a small fraction, of the diversity of organic form produced during the early history of organisms. Moreover, much of the diversity that has been lost disappeared almost as soon as it arose, more than 500 million years ago.

The progressivist ideology of the nineteenth century produced a rather Whiggish picture of the history of life. The evolution of organisms was seen as a movement from simple to complex, from less adapted to more adapted, from organisms at the mercy of every perturbation of a hostile and uncontrolled nature, to those that resist and even dominate nature, from the lowly amoeba to you know who. As Gould points out, the iconography of organic evolution reflects this in the “tree of life,” which begins with a narrow range of types at the bottom of the trunk and slowly diversifies into more and more branches and subbranches as the tree grows. This spreading out is often ascribed to the slow “discovery” and “exploration” of new ways of making a living that had previously been only potential until they were realized by evolving organisms in their ceaseless drive for expansion and diversification.

Yet the Burgess Shale, and indeed the entire Cambrian explosion, contradicts this metaphor. All the main branches of the “tree” of invertebrate life seem to have emerged at ground level like a lilac bush and some, perhaps most, were nipped in the bud at an early stage. Since that early attrition, however, there has been neither a major new diversification nor a further loss of basic body plans. Plus ça change, plus c’est la même chose, at least for the last half billion years. Of course, a lot of very important things have happened more recently, like the emergence of the vertebrate animals and higher plants, and their occupation of the land. But even here, we get the impression of a series of rapid diversifications as each new major form arose, followed by a weeding out, rather than a slow and steady increase in variety.

It might be argued that the loss of the alternative body plans of the Burgess Shale creatures during evolution is precisely a demonstration of the inexorable operation of natural selection, replacing less fit forms by better alternatives. But the “weird wonders” could not have been all that unfit. They lasted for tens of millions of years and swam side by side with the ancestors of our modern invertebrates. Nor were their body plans aberrant in any sense except that they were different from what we know. While they passed from the evolutionary stage rather early in the piece, they likely would have proved most successful had they stayed on. The Burgess Shale fauna tell us that the animals now living need not have been, but owe their present existence to an historical contingency. If no animal now has five eyes, it is for a very different reason than that no planet moves in straight lines, for once upon a time Opabinia lived and thrived. Extinction is the inevitable fate of every species.

Tennyson was right. Nature is careless of the type. She cares for nothing. All shall go. Nor is there any indication that in the process of evolution, species are learning by experience. The average geological lifetime of species that appear late in the evolutionary history of a lineage is no longer than of those that appear earlier. This observation has given rise to Leigh Van Valen’s “Red Queen Hypothesis” that species, like the chess queen in Through the Looking Glass, are constantly running just to keep up with a world that is ever changing in the wrong direction. Species go extinct because the world from which they make a living changes faster than the species can adapt, and it appears to be matter of chance which species meet their eventual mortality first. Since major groups of organisms and the blueprint they embody go extinct when all of the species of which they consist go extinct, the historical survival of classes of shape and organization has no sensible or predictable pattern. If life exists elsewhere in the universe it will not look anything like life on earth, for its evolution will have been another of the many potential realizations of history.

In Gombrich’s Art and Illusion, there is a juxtaposition of an English lithograph of cows in Derwentwater with a Chinese rendering of the same scene. The English countryside in the hands of the visiting Chinese painter winds up looking as we might suspect it would, just like a rocky landscape on the Li River. One could not imagine from these remarkably dissimilar paintings that they are not of two countries with very different kinds of plants and mountains, differences that would be instantly obvious in photographs. We have long since been taught by such as Gombrich that painters paint, not nature, but other paintings. We know that historiography cannot escape its own history. But when it comes to science, we still naively believe in Ruskin’s “innocent eye.” Gould takes it as a major task of Wonderful Life, to strip us of our innocence.

The discoverer of the Burgess Shale in 1909 was the prominent paleontologist Charles Doolittle Walcott. His reaction to the Burgess Shale fauna was to see them as a collection of conventional organisms. In Gould’s often-repeated metaphor, Walcott “shoehorned” the fossils into the standard classification of animals. Given the assumption that all the kinds of organisms that had ever lived had already been seen, this “shoehorning” was not all that hard to do because the fossils were badly squashed and distorted or had missing parts.

Beginning in 1970, the Burgess remains were subjected to a careful reexamination by Harry Whittington at Cambridge University, who painstakingly stripped apart the squashed layers of the fossils and succeeded in making three-dimensional reconstructions. The result was that by 1975, with his reconstruction of the five-eyed Opabinia, Whittington began to see the Burgess fauna as radically different from what had been previously known. Whittington, together with two of his students, Simon Conway Morris and Derek Briggs, then spent the next ten years redescribing and radically revising the conservative view of the animals that Walcott had created. A minor revolution, perhaps, but one that has considerable influence on our view of evolutionary possibilities.

For Gould, the lesson to be learned from this bit of the history of ideas is the overwhelming influence of ideology in science. Walcott is described as having a “deeply conservative and traditional perspective on life and morality,” a perspective that naturally extended to his view of dead animals. Nor, Gould implies, could we expect anything else from someone who was the head of the Smithsonian Institution, president of the National Academy of Sciences, founder, together with President Wilson, of the National Research Council, and generally the intimate of America’s upper-class elite. How could Walcott, revealed as a jingoist, anti-Semitic, right libertarian, be expected to see anything in fossils but a confirmation of what established science had already long determined to be true? In contrast, Whittington, Conway Morris, and Briggs are regular guys. (They soon become “Harry,” “Simon,” and “Derek” and they are, of course, pals of Steve’s.) Indeed, Simon was a young Turk and social radical of the 1960s, who by temperament and ideology would try to overturn established truth. It is not only the title of Wonderful Life that is taken from the movies. “The story…is complex, involving the collective efforts of a large cast. But three paleontologists dominate center stage.” Not three, but four, if we include the villain of the piece.

There are real problems with this analysis, however. Harry Whittington, although the child of a skilled worker, is described as a “quiet, conservative, moderately busy man,” hardly the stuff from which revolutions are made. For that matter, no one was a more important establishment figure than Charles Darwin when he began his work on the origin of species. He was a man of leisure whose income came from investments in railway shares. Yet he quite deliberately stirred rather more trouble than Harry, Simon, and Derek ever dreamed of. The problem is that ideological analysis has been applied at the wrong scale. Certainly people see nature through a glass molded by social experience. Darwin’s theory of evolution by natural selection is obviously nineteenth-century capitalism writ large, and his immersion in the social relations of a rising bourgeoisie had an overwhelming effect on the content of his theory. One cannot even imagine such a theory arising in the fourteenth century. Class matters, but it is not determining for individuals. As the inventor of the modern theory of class reminded us in the third thesis on Feuerbach: “The doctrine that men are the products of circumstances and upbringing…forgets that it is men that change circumstances.”

The difference between the work of Walcott and that of Harry and friends is to be found at a different, although no less general, level. It arises from the fact that professional scholarship is a way of building a life, and that successful careers are constructed in ways constrained by the social structure of professional life.

In scientific scholarship, there are two patterns for a high status life. One is to be the discoverer and inventor of something that reverberates through the current structure of knowledge, or at least to make consistent, significant contributions to the fundamental structure of problems of the field. The second, sometimes flowing from the first, but not always, is to become part of the political hierarchy of the field, influencing bureaucratically what one no longer can or never has influenced by the substance of one’s scholarship.

Charles Doolittle Walcott, at the time of his discovery of the Burgess Shale, was already at or near the peak of his professional status. His heavy administrative schedule as a wheeler and dealer in the national politics of science, documented at length by Gould, had two inevitable effects on his treatment of the Burgess Shale. First, he simply did not have the time or psychic energy to carry out the minute dissection of the squashed fossils that would have been necessary to make a correct reconstruction possible. Even Harry Whittington had to work for four years before he came across a case, Opabinia, that could not be fit into the old view.

Second, Walcott seemingly had no motive for pressing his inquiries further. Status and success were already his and would continue throughout his life irrespective of any scientific discovery he might make. The career problem facing Harry, Simon, and Derek was rather different. Unlikely by circumstances and temperament to rise in science by purely political means, they chose the path more honored in song and story. They rebelled against the orthodoxy of their predecessors and said something new. In evolutionary biology, in particular, revisionism is a common mode of career building. We are regularly treated to “new” theories about evolution that claim in one way or another to renovate Darwinism without actually overturning it. Self-proclaimed “non-Darwinian” mechanisms of evolution abound, and their anti-eponymous advertisements are a mode of self-help. After all, if what I say is “non-Darwinian,” then perhaps it will come to be known as “Lewontinian.” None of us is quite free of the ambition, but only a few have cashed it out (including the author of Wonderful Life, who has made his own contribution to the “non-Darwinian” corpus in the theory of punctuated equilibrium). Gould tells us that the young revisionists of the Burgess Shale “have since built brilliant careers.” One has no need of a theory of ideology for this bit of history.

Much of Gould’s argument for the dramatic effects of contingency in evolution rests on the failure of most Burgess Shale organisms to fit into the classification of present day forms and their fossil ancestors. So, for example, among the arthropods (“joint footed” animals like shrimp, crabs, insects, spiders, and the like), there are four major subdivisions based on how their multiple appendages and serial segments are arranged from front to back. Four arrangements characterize modern arthropods and their ancestors while an additional twenty arrangements with no modern representatives can be seen in Burgess Shale organisms. But the failure to be able to classify the Burgess organisms into modern groups is a consequence of the contingency of classification, which is, in turn, only partly reflective of the contingency of evolution.

It is said that at a testimonial dinner for Bill Clem, the famous big league umpire, a speaker rose to praise Clem by saying, “He always calls them the way he sees them.” At this, a second speaker rose to disagree. “No, he always calls them the way they are.” Clem then got up and replied, “You’re both wrong. Until I call ’em, they ain’t nothin’.” Clearly there can only be white swans if we decide in advance that “if it ain’t white, it ain’t a swan.” So the seeming strangeness of most Burgess arthropods arises from the use of particular characters by taxonomists to erect the boundaries of arthropod classes. Had more of the Burgess animals left descendants, a very different classification would have been created, one that may easily have included the extinct lines as only minor variants. Gould dispenses with this problem in a footnote and so misses the opportunity to consider a deep problem in biology: How do we decide what is similar to what? The purpose of classification of organisms is to reconstruct their ancestry and relationships—to write the “begats” of all of life. But relationship is judged from similarity of shape and behavior.

Unfortunately, there is no biological theory that enables us to decide on some independent scale how similar two organisms are, based only what they look like. Biologists judge similarity by surveying the range of known organisms and using an intuitive notion of similarity that depends upon what organisms they have actually seen. So, for example, systematists do not use similarity of size as a character on which to classify organisms because size has been found to vary immensely among organisms that are thought, on other grounds, to be closely related. The process is not entirely circular, because organisms classified according to one set of characteristics often, but not always, have a very similar classification when judged from a different set of characteristics. To be sure, characteristics like size that do not agree are tossed out as unreliable. The consequence of this methodology is that organisms that do not fit our longstanding classification scheme are regarded as “weird wonders,” although we have no independent criterion of their weirdness. The appearance of oddness depends a great deal on the range of what we have seen and on conventions of those who are in the know. To make the point, I have included in the picture of the Burgess Shale organisms on page 4 one, Ottoia, that may seem very odd to most of the readers of The New York Review, but is a conventional priapulid worm whose descendants today look pretty much the same.

Precisely because biology has no general theory of what organisms ought to look like, we have nothing but our accumulated historically contingent experience about what they do look like as a guide for ordering the world. The divisions that we regard as basic are simply the result of what has been left over or dug up from the extinctions of the past. Suppose, for example, that an archaeologist of a long distant future uncovered a fragmentary copy of the Manhattan telephone directory and found among the “S” entries only those that began with, say, “Sh” and “Sp.” This would seem a fundamental distinction and she would classify the Shapiros into one group and the Spiros into another. A later, luckier, archaeologist, finding a more complete directory that included also “Sa,” would then regard the Sapirs and Safers as radically different from what was previously known, although the names are, in fact, variants of a basic root and their possessors may well be second cousins once removed. So evolution may be contingent only in a superficial and uninteresting way. The exact forms that have left descendants visible in fossil remains may indeed be accidental variants of a historically accidental process. But they may all be distinctions without a difference, superficial orthographic variants of a deep structure whose rules we have yet to uncover. A description of all the organisms that have ever been cannot decide the issue. It may be that all swans are not white, but that there is some other nontrivial attribute that they all necessarily possess. We cannot know the answer unless we have a theory of biological form that is deduced from some general principles of biological organization, rather than inferred from the collection of objects. Or it may be that no such principles exist, and that in this broadest sense, life has no meaning.

This Issue

June 14, 1990