Dimensions of Darwinism
iEDITED BY

MARJORIE GRENE
CAMBRIDGE UNIVERSITY PRESS

4
The hardening of the
modern synthesis
STEPI IEN JAY GOULD
Pluralistic and hard versions

In 1937, just as Dobzhansky published the book that later generations would laud as the foundation of the modern synthesis, the American Naturnlist published a symposium on "supra-specific variation in nature and in classification." Alfred C. Kinsey, who later became one of America's most controversial intellectuals for his study of basic behaviors in another sort of WASP,1 led off the symposium with a summary of his extensive work on a family of gall wasps, the Cynipidae.
In his article, Kinsey strongly advocated the central theme of the developing synthesis: Evolution at all scales, particularly macroevolution, could be explained by the genetic mechanisms observed in laboratories and local populations. He first complained that some geneticists and naturalists were still impeding a synthesis with their insistence upon causal separation of levels: "Just as some of the geneticists have insisted that the laboratory genetics may explain the nature and origin of Mendelian races, but not of natural species, so others indicate that the qualities of higher categories must be explained on bases other than those involved in species" (1937, p. 208). He then defended the central postulate:

On the contrary, recent taxonomic studies and such experimental analyses as have been made of natural species indicate that Mendelian genetics provides all the hereditary mecha

1. Apologies to readers unfamiliar with American slang, but I coulelil't resist the allusion. Yes, Kinsey the celebrated wasp taxonomist is the same man who later published the famous "Kinsey Reports" on sexual behavior in thehuman male and female - the first dispassionate studies, based on sufficiently large samples, of what people really do do. WASP is a colloquial American acronym for "white Anglo-Saxon Protestant," still, despite all our ethnicity, the largest group in our country.
72 Stephen J Gould

nism necessary for the evolution of species as well as for laboratory races; and I shall undertake to show that the same genetics is all that is involved in the origin and development of the characteristics of higher categories. [1937, p. 208]
Yet in his major monograph of 1930, the first "Kinsey Report" if you will, Kinsey defended a position that would put him outside the synthesis by modern definitions (and he continued to hold these views in 1937). In his section on "mutation" in the origin of Cynips species, he contrasted two points of view: the first, supported by "laboratory studies" held that "changes in genes occur . . . as mutations which are complete as soon as they have occurred"; and the second, defined as the "neo-Darwinian conception of 'fluctuating variations' which, by being accumulated over many generations and bent in a given direction by the force of natural selection, would gradually give rise to the characters of new species" (1930, p. 25). Kinsey was not merely talking about the origin of new features by sudden mutation, but of new species. Later in the paragraph, he said so explicitly and again contrasted his position with Darwinism:

Under this latter [neo-Darwinian] interpretation there may be incipient species of the sort conceived by many taxonomists in their definitions of varieties and subspecies; under the genetic interpretation the first mutant individual embodies all that the new species will contain, and is the new species as soon as it is given an opportunity to perpetuate its mutantcharacters through a population of individuals. [1930, p. 25]

Let me emphasize that the "mutations" Kinsey defended for the origin of species were generally of small - though discontinuous morphological effect, entirely comparable to (indeed homologous with) standard laboratory mutations. Kinsey did not ally himself with the true de Vriesian saltationists who refuted the central syntheticist postulate by contrasting large-scale mutations that make species all at once with small-scale variation that differentiates local populations. Instead, Kinsey supported a unified genetics for all levels. His "macro" mutants were relatively small, not discombobulations of entire genetic systems a la Goldschmidt's "systemic mutants." They did not constitute a different genetics acting above the level of processes that continually occur within populations (the antisyntheticist claim). Rather, they represented the standard kind of change that, according to Kinsey, occurred within populations and also made new species.
Kinsey then explicitly defended the "genetic" view for speciation in Cynics (1930, p. 34) and promptly committed a second, and consequent, apostasy from later conceptions of the synthesis: He denied


Hardening of the modern synthesis 73

that most of the morphological expressions of species-forming mutations had any adaptive significance:

There seems no basis for believing the shortened wings or any of the concomitant variations of any adaptive value to any of these insects. The short wings are not confined to warmer or colder climates, and long- and short-winged forms of various species are active at the same season in the same localities. The field data suggest nothing as to the survival value of these outstandingly basic modifications of structure. [1930, p. 34]

This brief synopsis of Kinsey's views raises a dilemma: How could a man who seemed so much "in" the synthesis (as later defined) by his defense of genetic continuity in causation, be so far out of it in his explicitly antiadaptationist views and his defense of discontinuous origin (however small the steps) for new species. Consider, for example, Mayr's definition (1963, p. 586; but see also 1980, p. 1, to appreciate his long advocacy of this account): "The proponents of the synthetic theory maintain that all evolution is due to the accumulation of small genetic changes, guided by natural selection, and that transpecific evolution is nothing but an extrapolation and magnification of the events that take place within populations and species."
This dilemma can be resolved (ironically, since professional students of evolution must face it) by recognizing a property of science that monolithic, textbookish interpretations ignore at their peril: theories evolve, too. I shall argue that the synthetic theory underwent a major change in intent between its formulation in the 1930s and its hardening in the late 1940s. Most biologists are unaware of this change because we only read the last editions of famous works (Dobzhansky, 1951; and Simpson, 1953 for example) and ignore as superseded historical curiosities the original presentations (Dobzhansky, 1937; and Simpson, 1944, in these cases). Kinsey's work lay within the original boundaries for its defense of ordinary genetics as the source of all evolutionary change, while it clearly fell beyond the pale of the later, hardened version in its non-Darwinian approach to the roles of adaptation and natural selection in speciation.
The change involved intent even more than content. During the 1930s, the main concern of the primary builders (for a movement as yet unnamed) was methodological. They knew the great Darwinian hope of the late nineteenth century: The mechanism of heredity, when elucidated, would vindicate Darwin's views on small-scale variation and the creative action of natural selection working in the cumulative mode. They remembered the crushing of this hope during
74 Stephen J. Gould

the early twentieth century when Mendelism, in its first evolutionary incarnation, contributed yet another anti-Darwinian theory in the form of de Vriesian saltationism. They rued the resulting anarchy of incommensurate, competing theories that had driven so many great naturalists to the despair of nearly total agnosticism, as in Bateson's famous declaration: "Less and less was heard in genetical circles and now the topic is dropped. When students of other sciences ask us what is now currently believed about the origin of species we have no clear answer to give. Faith has given place to agnosticism" (1922, p. 57).
As genetics had sown confusion, so too (they hoped) might it eventually breed concord by setting a unified causal base for all evolutionary change. Advances in genetics had established the ground for such a hope. De Vries' systemic saltations had proven to be, quite literally, a primrose path; a genetics of smaller-scale change might now prove sufficient for all levels of evolution, from the lowest (by observation and experiment) to the highest (by extrapolation). Fisher had proven that continuous variation could have a Mendelian underpinning. Morgan had abandoned his commitment to saltational change and had, late in life, accepted the idea that micromutations underlay evolutionary change. The time was ripe for causal unity.
Yet, just as geneticists were weeding out their confusion and settling upon a consensus about how heredity worked, many evolutionists refused to recognize the unifying potential of this emerging agreement, and continued to insist upon a separation between smallscale (and, by implication, unimportant) evolution produced by these known genetic mechanisms and a larger-scale evolution produced by other genetic processes as yet unknown. Kinsey, as a synthesist of the first period, expressed his frustration at this defeatist argument in ridiculing (see quote on p. 71) those evolutionists who liked Mendelism for lab stocks but not for natural populations, or for change within population but not for transspecific evolution.
The original version of the synthesis, therefore, sought to remove this idea of a separate but unknown genetics for large-scale changes and to render all of evolution by known genetic mechanisms that could be studied directly in field and laboratory. It was primarily a plea for knowability and operationalism, for a usable and workable evolutionary unity. It did not attempt to crown any particular cause of change, but to insist that all permissible causes be based on known Mendelian mechanisms. In particular, it did not insist that adaptive, cumulative natural selection must underlie nearly all, or even most, change - though many synthesists personally favored this view. Any theory of change would be admitted, so long as its causal base lay in

Hardening of Synthesis 75

known Mendelian genetics. In the 1930s, for example, genetic drift was often granted a predominant role in phenotypic change, not only at the level of demes, but also in the origin of many species. (Dobzhansky's 1937 book may favor selection slightly, but it comes far closer to asserting this role for drift than his later work would lead one to expect.) Such views about the relative importance of drift (and consequently diminished role for selection) were admitted to the original synthesis because they invoked only known and small-scale genetic change. Likewise, the original version admitted Kinsey's view that the primary phenotypic changes of speciation were generally nonadaptive. For Kinsey vigorously supported the key methodological claim of the original synthesis: Small-scale genetic changes within natural populations are the stuff of all evolution. I have called this original version "pluralistic" because it admitted a range of theories about evolutionary change, Darwinian and otherwise, and insisted only that explanations at all levels be based upon known genetic causes operating within populations and laboratory stocks.
This pluralistic version was slowly and subtly altered, primarily during the 1940s (and perhaps with the 1947 Princeton conference as a focal point), as the intent of explanation by known genetics shifted to the content of one particular theory - neo-Darwinism and its insistence that cumulative natural selection leading to adaptation be granted pride of place as the mechanism of evolutionary change. The synthesis hardened by elevating one theory to prominence among the several that supported the primary methodological claim of the original version - and eventually (particularly among many secondstring votaries) by insisting to the point of dogma and ridicule that selection and adaptation were just about everything. The hard version had no more room for Kinsey, who was off doing something else by then anyway.
In the rest of this essay, I shall illustrate this shift by discussing the increasing commitment to adaptationist explanations by three key figures: Dobzhansky, who was certainly primus inter pares of the whole movement; Simpson, who brought the most disparate field of paleontology under the aegis of the synthesis; and Wright, the great geneticist who did so much to create the original version and then largely stood outside its later hardening (his own shift to adaptationism is especially interesting in this light). If this subject strikes some readers as a bit arcane or unimportant, I can only assert an insider's view, admittedly partisan (Gould and Lewontin, 1979; Gould and Vrba, 1982): The rather strict Darwinism of the hard version established a research program that has directed (and in some ways restricted) the field for 30 years. Adaptationist commitments extend
76 Stephen J. Gould

from work on the origin of life to sociobiology. I am also painfully aware that the "triumph of adaptationism" is an exceedingly complex historical subject. I do not wish to fall into the typological traps for history that all evolutionists are trained to avoid for nature. Variation abounded at all times, and it was as unconstrained as human intellects are broad. I speak of no lock step affecting all evolutionists together as an imposed essence upon a field of variation, but only of general tendencies that, like evolutionary trends generated by species selection within clades, often sneak up on you and get established while virtually no one notices what is happening.

Three examples of increasing commitment to adaptation 2

1. Increasing emphasis on selection and adaptation between the first (1937) and last (1951) edition of Dobzhansky's Genetics and the Origin of Species.

Dobzhansky's original probe (1937) toward synthesis was more a methodological claim for knowability than a strong substantive advocacy of any particular genetic argument - though his general Darwinian preferences are clear enough. It held, contrary to his own mentor Filipchenko in Russia or H. F. Osborn in America, that the methods of experimental genetics can provide enough principles to encompass evolution at all levels, but it did not play favorites among the admitted set of legitimate principles. It did not, in particular, proclaim the pervasive power of natural selection leading to adaptation as. a mechanism of evolutionary change. Macroevolution, it stressed, is not a thing apart, unknowable in principle from experimental work on laboratory and natural populations and requiring different (and perhaps unfathomable) modes of genetic change. Dobzhansky writes cautiously, emphasizing the hope for complete knowability based on microevolutionary genetics:
Experience seems to show, however, that there is no way toward an understanding of the mechanisms of macroevolutionary changes, which require time on a geological scale, other than through a full comprehension of the microevolutionary processes observable within the span of a human lifetime and often controlled by man's will. For this reason we are compelled at the present level of knowledge reluctantly to put a

2. I have written the Dobzhansky and Simpson examples in other contexts before (Gould, 1980 and 1982a) and much of the material in this discussion is taken from my previous work. William Provine knows so much more about Sewall Wright than anyone else does that I have based my discussion upon his excellent analysis in the manuscript cited in the references.

Hardening of the Modern synthesis 77

sign of equality between the mechanisms of macro- and microevolution, and, proceeding on this assumption, to push our investigations as far ahead as this working hypothesis will permit. [1937, p. 12]

Some inkling of the chaotic and depressed state of evolutionary theory before the synthesis can be glimpsed in a simple list of previously popular arguments that Dobzhansky regarded as sufficiently important to refute - claims that denied his hope for synthesis by suggesting that Mendelian processes observed in the laboratory do not represent the genetic style of "important" evolutionary change in nature. Dobzhansky rebuts explicitly the following arguments: Continuous variat,ion in nature is non-Mendelian and different in kind from discrete mutational variation in laboratory stocks (p. 57); Mendelian variation can only underlie differences between taxa of low rank (races to genera), while higher taxa owe their distinctions to another (and unknown) genetic process (p. 68); chromosomal changes are always destructive and can only lead to degeneration of stocks (p. 83); differences between taxa of low rank are directly induced by the environment and have no genetic or evolutionary basis (p. 146); Johannsen's experiments on pure lines showed the ineffectiveness of natural selection as a mechanism of evolutionary change (p. 150); selection is too slow in large populations to render evolution, even in geological time (p. 178); genetic principles cannot account for the origin of reproductive isolation (p. 255).
Dobzhansky's fifth chapter, on "variation in natural populations," stresses the pluralism of the early synthesis. Observable genetic phenomena are the source of all evolution; we find continuity from studies in the laboratory, to variation within natural populations, to formation of races and species:

It is now clear that gene mutations and structural and numerical chromosomal changes are the principal sources of variation. Studies of these phenomena have been of necessity confined mainly to the laboratory and to organisms that aresatisfactory as laboratory objects. Nevertheless, there can be no reasonable doubt that the same agencies have supplied the materials for the actual historical process of evolution. This is attested by the fact that the organic diversity existing in nature, the differences between individuals, races, and species, are experimentally resolvable into genic and chromosomal changes that arise in the laboratory. [1937, p. 118]

But what forces shape and preserve this variation in nature? Dobzhansky stresses natural selection (p. 120) as he does throughout the book, but he does not grant it the dominant role that later "hard" 78 and 79 are missing!
80 Stephen J. Gould

leopard than to those occupied by wolf, coyote, and jackal. The feline adaptive peaks form a group different from the group of the canine "peaks." But the feline, canine, ursine, musteline, and certain other groups of peaks form together the adaptive "range" of carnivores, which is separated by deep adaptive valleys from the "ranges" of rodents, bats, ungulates, primates, and others. In turn, these "ranges" are again members of the adaptive system of mammals, which are ecologically and biologically segregated, as a group, from the adaptive systems of birds, reptiles, etc. The hierarchic nature of the biological classification reflects the objectively ascertainable discontinuity of adaptive niches, in other words, the discontinuity of ways and means by which organisms that inhabit the world derive their livelihood from the environment. [pp. 9-10]

Thus, Dobzhansky renders the hierarchical structure of taxonomy as a fitting of clades into ecological spaces. Discontinuity is not so much a function of history as a reflection of adaptive topography. But this cannot be, for surely the cluster of cats exists primarily as a result of homology and historical constraint. All felines are alike because they arose from a common ancestor shared with no other Lade. That ancestor was well adapted, and all its descendants may be. But the cluster and the gap reflect history, not the current organization of ecological topography. All feline species have inherited the unique cat Bauplan, and cannot deviate far from it as they adapt, each in its own particular (yet superficial) way. Genealogy, not current adaptation, is the primary source of clumped distribution in morphological space.

2. The shift in G. G. Simpson's explanation of "quantum evolution" from drift and nonadaptation (1944) to an exemplar of strict adaptation (1953).

Although Simpson, probably more than Dobzhansky, personally favored selectionist arguments in the first version of his seminal work (1944), he was equally pluralistic and nonrigid. Indeed, he developed an explicitly nonadaptationist theory to resolve his problem, and he considered this theory of "quantum evolution" as the crowning achievement of his book.
Like Dobzhansky, Simpson viewed consistency of all evolutionary change with principles of modern genetics as his primary claim. The major challenge to unity and consistency arose from the famous "gaps" or discontinuities in the fossil record - the appearance of new Bauplane without fossil intermediates. He wrote:

Hardening of the modern synthesis 81

The most important difference of opinion, at present, is between those who believe that discontinuity arises from intensification or combination of the differentiating processes already effective within a potentially or really continuous population and those who maintain that some essentially different factors are involved. This is related to the old but still vital problem of micro-evolution as opposed to macro-evolution . . . If the two proved to be basically different, the innumerable studies of micro-evolution would become relatively unimportant and would have minor value to the study of evolution as a whole. [1944, p. 97]
To explain discontinuities, Simpson relied, in part, upon the classical argument of an imperfect fossil record, but he concluded that such an outstanding regularity could not be entirely artificial. He also recognized that his favored process of gradualistic Darwinian selection in the phyletic mode would not suffice, and he, therefore, framed the hypothesis of quantum evolution. He was clearly quite pleased with this formulation, for he ended the book with a twelve-page defense of quantum evolution, calling it "perhaps the most important outcome of this investigation, but also the most controversial and hypothetical" (p. 206). Faced with the prospect of abandoning strict selection in the gradual, phyletic mode, he framed a hypothesis that stuck rigidly to his more important goal - to render macroevolution by genetical models operating within species and amenable to study by neontologists. Thus, he focused upon the one major phenomenon in the literature of population genetics that granted control of direction to a phenomenon other than selection - Sewall Wright's genetic drift.
He envisaged the major transition as occurring within small populations (where drift might be effective and preservation in the fossil record virtually inconceivable). He chose the phrase "quantum evolution" because he envisioned the process as an "all-or-none reaction" (p. 199) propelling a small population across an "inadaptive phase" from one stable adaptive peak to another. Because selection could not initiate this departure from the ancestral peak, he called upon drift to carry the population into its unstable intermediary position, where it must either die, retreat, or be drawn rapidly by selection to a new stable position. Simpson felt that with quantum evolution he had carried his consistency argument to completion by showing that the genetical models of neontology could encompass the most resistant and mysterious of all evolutionary events. Quantum evolution, he wrote, "is believed to be the dominant and most
82 Stephen A. Gould

essential process in the origin of taxonomic units of relatively high rank, such as families, orders, and classes. It is believed to include circumstances that explain the mystery that hovers over the origins of such major groups" (p. 206). Simpson could, therefore, conclude: "The materials for evolution and the factors inducing and directing it are also believed to be the same at all levels and to differ in megaevolution only in combination and in intensity" (p. 124).
When pressured for a new edition of Tempo and Mode, Simpson realized that too much had happened in the intervening ten years to permit a reissue or even a simple revision. The field that he pioneered had stabilized and flourished: "It was [in the late 1930s] to me a new and exciting idea to try to apply population genetics to interpretation of the fossil record and conversely to check the broader validity of genetical theory and to extend its field by means of the fossil record. The idea is now a commonplace" (1953, p.ix). Thus, Simpson followed the outline of Tempo and Mode, but wrote a new book more than double the length of its ancestor - The Major Features of Evolution, published in 1953.
The two books differ in many ways, most notably by Simpson's increasing confidence in selection within phyletic lineages as the only important cause of change. Consider this addition to the 1953 book, a speculative comment on trends in titanothere horns and too rapid a dismissal (in my view) of the venerable argument that incipient stages of useful structures may have no evident function themselves:

This long seemed an extremely forceful argument, but now it can be dismissed with little serious discussion. If a trend is advantageous at any point, even its earliest stages have some advantage; thus if an animal butts others with its head, as titanotheres surely did, the slightest thickening as presage of later horns already reduced danger of fractures by however small an amount. [pp. 270-1]

But the most dramatic difference between the two books lies in his demotion to insignificance of the concept that was once his delight and greatest pride - quantum evolution. It had embodied the pluralism of his original approach - reliance on a range of genetical models. For he had advocated genetic drift to propel very small populations off adaptive peaks into an ultimately untenable inadaptive phase. And he had explicitly christened quantum evolution as a mode different in kind, not only in rate, from phyletic transformation within lineages. But now, as the adaptationist program of the synthesis hardened, Simpson decided that genetic drift could not trigger any major evolutionary event: "Genetic drift is certainly not involved in all or

Hardening of the modern synthesis 83

in most origins of higher categories, even of very high categories such as classes or phyla" (p. 355).
In an "intermediate stage" - his presentation to the Princeton conference - Simpson (1949, p. 224) had emphasized the dominance of selection in quantum evolution, but had not denied other factors. But by 1953, he had completed his own transition. In 1953, quantum evolution merits only four pages in an enlarged final chapter on modes of evolution. More importantly, it has now become what Simpson explicitly denied before - merely a name for phyletic evolution when it proceeds at its most rapid rates, a style of evolution differing only in degree from the leisurely, gradual transformation of populations. Quantum evolution, he now writes, "is not a different sort of evolution from phyletic evolution, or even a distinctly different element of the total phylogenetic pattern. It is a special, more or less extreme and limiting case of phyletic evolution" (p. 389). On page 385, quantum evolution is listed as one among the four styles of phyletic evolution - and all four are characterized by "the continuous maintenance of adaptation." The bold hypothesis of an absolutely inadaptive phase has been replaced by the semantic notion of a relatively inadaptive phase (an intermediary stage inferior in design to either ancestral or descendant Bauplalt). But relative inadaptation is no threat to the adaptive paradigm, for it matters little that an intermediate is not so well suited for its environment as its ancestor was for a different habitat (because they are not in competition). And it matters even less that the intermediate is not so well designed as its descendent will be (for there is even less opportunity for competition here!). In short, relatively inadaptive populations are fully adaptive to their own environments in their own time - and quantum evolution moves comfortably under the umbrella of the adaptationist program. Simpson even suggests that quantum evolution may be more rigidly controlled by selection than other modes of evolution (though he still invokes inadaptation for the initial trigger): "Indeed the relatively rapid change in such a shift is more rigidly adaptive than are slower phases of phyletic change, for the direction and the rate of change result from strong selection pressure once the threshold is crossed" (p. 391).

3. Sewall Wright's early change on the role of adaptation in the formation of species.

Sewall Wright, when interviewed today (as both W. Provine and I can attest), complains bitterly that his views on the evolutionary role of genetic drift, once called the "Sewall Wright effect," have been consistently misrepresented. Since genetic drift is incontestably about
84 Stephen J. Gould

stochastic change in gene frequencies via sampling error alone, one might assume (as many have, given the lamentable "scholarly" tradition of pronouncement without reading) that Wright had advocated a radically non-Darwinian approach to evolutionary change by demoting selection and adaptation in favor of accident. Not so, Wright insists. He argues, with evident justice apparent to anyone who studies his works of the past thirty years, that his theory of "shifting balance," which does specify an important role for genetic drift, is strongly adaptationist - but that adaptation arises at a level higher than the traditional Darwinian focus on individuals.
In brief, Wright believes that most evolutionary change within species occurs by the differential success of some demes over others (rather than by strictly individual selection within a panmictic population). Wright views this "interdemic selection" as adaptive and argues that his theory lies within Darwinian traditions because it identifies (higher-level) selection as the cause of evolutionary change, and adaptation as its result.
But an appeal to higher-level or demic selection poses another problem: How does a species get divided into demes in the first place? In other words, what process supplies the raw material (differences among demes) that higher-level selection must utilize to produce adaptive change? Wright invokes genetic drift to resolve this problem, and argues that drift acts as a generator of higher-level variability among demes, not as an agent of evolutionary change.
Consider the founding deme of a new species. It occupies one adaptive peak of the potential landscape. Other peaks are available for habitation, but how can they be populated? Other peaks cannot be reached by selection alone, because organisms descending the slopes of the original peak - a prerequisite to any climbing of a new slope - will be eliminated by selection, even though adjacent peaks may be higher than the original habitation. Wright invokes genetic drift as a mechanism for permitting small demes to descend slopes and cross adaptive valleys. When a small deme has, by accident, entered a valley and approached another slope, selection can then draw it up to the new peak. In this way, many peaks can be inhabited and the raw material for interdemic selection can be generated.
Wright's theory is rooted in the conceptually unfamiliar notion of hierarchy (see Gould, 1982b). Indeed, Wright told me that he originally thought of naming his shifting balance theory the "two-level" theory for its emphasis on the concerted interaction of ordinary Darwinian selection upon organisms and the higher-level process of interdemic selection. In this context, we can understand the major
Hardening of the modern synthesis 85

source of the misrepresentation that Wright so bitterly deplores: Few evolutionists are used to thinking in terms of hierarchically superposed and interacting levels of selection, for the strict Darwinian tradition advocates nearly complete explanation by selection acting upon organisms within a population. If evolutionists tend to translate everything they hear into their favored terms of organisms within populations, then what else can genetic drift be but a mechanism of evolutionary change, for it indisputably alters gene frequencies within demes - and this is the stuff (by extension) of all evolution. Genetic drift, in this limited view, must, therefore, be a non-Darwinian agent and opposed to selection. Hierarchical thinking is a prerequisite to understanding Wrigsht's true intent, for drift can then act (as he believes) as an agent of variation, supplying raw material in the form of differentiated demes to the higher-level process of interdemic selection.
But we must also acknowledge a different and historical (or ontogenetic) reason for the confusion. Wright's later shifting balance theory is adaptationist as he claims and does invoke drift only as a source of variation. But Wright, though in many ways estranged from the developing synthesis, followed the same trend toward increasingly exclusive emphasis upon adaptation in evolutionary change. The version of shifting balance that he has advocated for 30 years did not arise by sudden creation, complete in its final form. It had antecedents of differing tenor in Wright's earlier work, and these articles, written during the pluralistic phase of the synthesis, granted a much greater role to nonadaptation in evolutionary change. In short, Wright did sometimes invoke drift as a non-Darwinian agent of change in articles written during the early pluralistic phase of the synthesis.
William Provine, who is writing a complete scientific biography of Wright, has catalogued Wright's ambiguities and multiple intents during the crucial period 1929-32. The later selectionist view is already in the wings, but several key passages also advocate the nonadaptationist role for drift that Wright would later reject. Wright wrote in 1931 (p. 158) that shifting balance "originates new species differing for the most part in nonadaptive respects." In the following year, he stated (1932):
That evolution involves nonadaptive differentiation to a large extent at the subspecies and even the species level is indicated by the kinds of differences by which such groups are actually distinguished by systematists. It is only at the subfamily and family levels that clear-cut adaptive differences
86 Stephen J. Gould

become the rule. The principal evolutionary mechanism in the origin of species must then be an essentially nonadaptive one.
[pp. 363-4]
Provine concludes "The careful reader in 1932 would almost certainly conclude that Wright believed nonadaptive random drift was a primary mechanism in the origin of races, subspecies, species, and perhaps genera. Wright's more recent view that the shifting balance theory should lead to adaptive responses at least by the subspecies level is found nowhere in the 1931 and 1932 papers" (ms., p. 58).

Was hardening a parochial American phenomenon?

The community of American evolutionists, although large enough, was itself sufficiently focused and hierarchical that the changed opinion of a few key leaders might have swayed an entire field for no particular good reason beyond authority. We must, therefore, ask whether the hardening of the synthesis was simply a parochial American phenomenon, traceable to one or a few key people, and without general significance.
England provides a good test for the generality of hardening because adaptationist traditions have been so strong in Darwin's own land (though Darwin himself was a pluralist). Perhaps the synthesis was "hard" in Britain from-the first, and all the change I have documented merely represents a few recalcitrant Americans finally falling into line. After all, England has produced a long line of hyperselectionists, from Wallace in Darwin's own day, through the purity of R. A. Fisher's adaptationism, to the convictions of E. B. Ford and the Cepaea school launched by A. J. Cain.
Yet we can demonstrate that hardening also affected the work of several prominent British evolutionists. Consider, for example, Julian Huxley who, after all, gave the synthetic theory its name (1942). Huxley, to be sure, and as a good Englishman (but unlike his famous grandfather), was always more committed to adaptation than, say, the early Dobzhansky. Yet his early views had the pluralistic cast so characteristic of the 1930s and early 1940s, and he undertook what many biologists regard as his most important research in a largely nonadaptationist context. Throughout the 1920s, and culminating in his book of 1932, Huxley developed and largely founded the study of allometric growth expressed by power functions. Though this theme could have been developed in a strictly adaptationist context (see next paragraph), Huxley took a radically different approach and used the allometric insight primarily to argue that major features of

Hardening of the modern synthesis 87

an organism's phenotype are probably not adaptations ill se, but complex nonadaptive consequences of inherited growth tendencies, when selection works on one feature and "drags along" a large set of correlated traits.
I say this with personal diffidence because I am, in some measure, responsible for the reinterpretation of this work along hard adaptationist lines (Gould, 1966). In one of my first papers, written with all the fervor of a graduate student asked to compose a review article, I argued that virtually all allometries should be recast as adaptations because parameters of power functions are as subject to selection as the morphology of phenotypes. I well remember, in my unquestioning adaptationist commitment of the time, feeling that Huxley had somehow "betrayed the cause" in departing from the straight and narrow of strict Darwinism when an adaptationist interpretation for the same phenomena was available. My general contention about selection on growth rates was not wrong, but I now rather suspect that Huxley's original perspective is the more important theme for allometricians.
Huxley continued his active pluralism into the book that gave the synthesis its name (Huxley, 1942). He emphasized adaptation and the revival of Darwinism, but held throughout that differences among taxa of lower rank were largely nonadaptive and that Wright's genetic drift supplied an explanation. Provine (ms.) has counted the references and documented that Huxley gives about equal time to Fisher's strict adaptationism and to the nonadaptationist interpretation of Wright's genetic drift.
Yet, by the 1950s, Huxley had adopted the hard version of strict adaptationism. At the major symposium of the 1959 Darwin centennial - a three volume paean to the hard version (Tax, 1960) - Huxley presented the keynote address (Huxley, 1960) and defended panselectionism with an explicit renunciation of his former view that nonadaptation regulates small-scale diversification. He quotes some lines of Darwin (slightly out of context I think), and then comments: "The first sentence refers to small-scale processes and makes intelligible the omnipresence of detailed adaptation" (1960, p. 11), thus recalling Weismann's famous line, the rallying cry of strict Darwinism at the turn of the century, about the omnipotence (Allmacht) of selection.
Hardening is also reflected through the several versions of what may be the most celebrated empirical study of the time - David Lack's work on the Galapagos finches. (I thank Malcolm Kottler for calling my attention to this example and for documenting it in a long letter of May 4, 1981.) Lack published his first monograph on the finches
88 Stephen J. Gould

in 1945, though a first draft had been written just before war broke out in 1939. In this work, Lack strongly supported the common view that small scale differences were largely nonadaptive. Kottler writes:

In this first monograph, Lack took the position that all subspecific differences and almost all differences between closelyrelated species were non-adaptive. The major exceptions in the latter case were the beak differences. But at the time Lack explained them entirely in terms of reproductive isolation. He explicitly rejected the view that closely-related coexisting species had to be ecologically isolated and denied that their beak differences were correlated with significant dietary differences.

In his major work of 1947 (Darwin's Finches), Lack continued to defend nonadaptation for many small-scale differences, but now advanced an adaptive interpretation for several others. He also introduced the famous, and now well-established, claim that beak differences are adaptive responses to diet and act to prevent competition.
In 1960, as the Darwin centennial marked the heyday of the hard version, Harper Torchbooks reissued Lack's 1947 monograph. In the major point of a one-page preface, Lack explicitly renounced his nonadaptationist view, citing the general change in concept that I have labeled as the hardening of the synthesis:
The reader may therefore be reminded that this text was completed in 1944 and that, in the interval, views on species-formation have advanced. In particular, it was generally believed when I wrote the book that, in animals, nearly all of the differences between subspecies in the same genus, were without adaptive significance. I therefore specified the only exceptions then known and reviewed the various ideas as to how nonadaptive differences might have been evolved. Sixteen years later, it is generally believed that all, or almost all, subspecific and specific differences are adaptive, a change of view which the present book may have helped to bring about. Hence it now seems probable that at least most of the seemingly nonadaptive differences in Darwin's finches would, if more were known, prove to be adaptive.

Conclusion

After all this documentation, I fear that the conclusion will seem a bit flat. I now arrive at the point where I should be giving a conclusive and erudite explanation of why the synthesis hardened. Yet truly, I do not know. Some guidelines and factors seem clear enough. First

Hardening of Modern Synthesis 89

of all, the community of evolutionary biologists is sufficiently small and sufficiently stratified - a few lead and many follow, as in most human activities - that we need not seek some deep and general scientific or societal trend to render a change in opinion of so many evolutionists of different nations. A change by a few key people, themselves in close contact and with mutual influence, might trigger a general response.
Empirical aspects certainly influenced the hardening. Once the pluralistic version had reemphasized classical Darwinism as a respectable alternative, the search for actual measures of selection and adaptation in nature intensified and succeeded. The British panselectionist scho;ol, headed by Ford and Cain (see Ford, 1963, for example), presented many examples from butterflies and snails. More importantly, Dobzhansky, the key figure in the transition, discovered that his favorite example of potential nonadaptation needed to be reinterpreted in selectionist terms. In 1937, he had attributed differences in inversion frequencies within natural populations of Droso phila to genetic drift, but he then discovered (see 1951) that these frequencies fluctuate in a regular and repeatable way from season to season, and decided (with evident justice) that they must be adaptive. Still, we surely cannot attribute such a major change as the hardening of the synthesis entirely to induction from a few empirical cases. After all, no pluralist had doubted that selection regulated many examples, so the elegant display of a few should not have established a generality. After all, the arch antisynthesist Richard Goldschmidt argued forcefully (1940, see Gould, 1982c) that most changes within populations had a selectionist base (that could not be extrapolated to explain higher-order evolution).
More importantly, most evolutionists brought up wholly within the generation of the synthesis do not know that an "official" (though admittedly not exclusive) viewpoint among taxonomists during the early 1930s identified as nonadaptive the great majority of phenotypic differences separating subspecies and species. (Kinsey's position, cited in the first section, reflects this professional consensus; the highly influential book of Robson and Richards, 1936, advanced this view as its primary contention.) Deference to experts is an important phenomenon in science - all the more so in a diffuse and largely nonexperimental field like evolutionary biology, where no one can have direct experience in all subdisciplines, where simple laboratory repetition cannot validate many conclusions, and where deference to expert opinion, therefore, plays a larger role than usual. One cannot read Wright and Huxley, in particular, during their pluralistic period without getting the definite impression that they personally rooted
90 Stephen J. Gould

for a stricter form of Darwinism, but hesitated to advance it because taxonomists proclaimed so certainly that the required frequency of adaptation did not exist at low levels in nature.
As the theoretical basis for neo-Darwinism grew, a new generation of systematists (Mayr, 1942, in particular) asserted a much wider role for selection and adaptation, and the earlier view of Robson and Richards eventually faded. One might give two different interpretations to such a scenario. The "heroic" version holds that the prevalence of adaptation is an empirical truth whose recognition had been impeded by an inadequate theoretical base. The supply of proper theory opened the eyes of systematists to an empirical reality that, in turn, reinforced the theory. The more "cynical" version holds that we still do not have a proper assessment of the relative importance of adaptation in small-scale differences. Since the world is so full of a number of things, cases of both adaptation and nonadaptation abound, and enough examples exist for an impressive catalogue by partisans of either viewpoint. In this light, historical trends in a science might reflect little more than mutual reinforcement based on flimsy foundations - in this case, an assertion of neo-Darwinian theory might lead taxonomists to emphasize the cases of adaptation that manifestly exist, and theorists might then feel sufficiently strengthened to assert a harder version on the false assumption that its empirical base had been independently and securely established.
Questions of relative frequency are among the hardest to resolve in science, and natural history abounds with them. No crucial experiment is possible, nor can we (given the multitude of potential cases) establish an adequate relative frequency by simple enumerative induction (What, after all, is a "random" sample in a world of protean taxonomic diversity?) We still do not know the relative frequency of adaptation in small-scale differences. The adaptationist context of current theory leads us to focus upon established examples, but how often do we find any discussion, or even acknowledgment, of the hundreds of other unexplained differences that separate most taxa?
Finally, a methodological point in conclusion. Darwin's assertion of evolution was an event of such unrivaled importance in the history of science and human society that we tend to view it as a watershed for all concepts in biology - as though everything should be discussed primarily in terms of before or after evolutionary theory. Such a perspective is inadequate, for many traditions of thought persist through evolutionary theory, emerging with a reinterpretation of causality to be sure, but intact nonetheless. In particular, certain "national styles" persisted from the eighteenth century, through Darwin's era, and into our own time. Views on adaptation provide a good example.

Hardening of the modern synthesis 91

I have said nothing about German biology because it has generally held a view of adaptation outside the scope of this essay (Rensch and other synthesists notwithstanding). Adaptation is seen as real but superficial, a kind of jiggling and minor adjustment within a Bauplarl evolved by some other mechanism - not, in any case, a general mechanism (by extrapolation) for evolutionary change at higher levels. This viewpoint is firm in the nonevolutionist transcendental morphology of Goethe and many of the Naturphilosopilen. It affects the entire "laws of form" tradition, underlies the pre-Darwinian evolutionism of Etienne Geoffroy de Saint-Hilaire, and persists to our time among such German evolutionists as Remane, Schindewolf, and Goldschmidt. ;
The adaptationist tradition, on the other hand, has been an English pastime for at least two centuries. If continental thinkers glorified God in nature by inferring the character of his thought from the laws of form linking his created species, or incarnated ideas (as Agassiz maintained), then Englishmen searched for him in the intricate adaptation of form and function to environment - the tradition of natural theology and Paley's watchmaker. Darwin approached evolution in a quintessentially English context - by assuming that adaptation represented the main problem to be solved and by turning the traditional solution on its head. Few continental thinkers could have accepted such a perspective, since adaptation, in their view, was prevalent but superficial. The centrality of adaptation among English-speaking evolutionists in our own times, and the hardening of the synthesis itself, owes much to this continuity in national style that transcends the simple introduction of evolutionary theory itself. One might say that adaptation is nature's truth, and that we had to overthrow the ancient laws-of-form tradition to see it. But one might also say that twentiethcentury panselectionism is more a modern incarnation of an old tradition than a proven way of nature. It may now be impeding a more pluralistic account that the early synthesists groped toward and then let slip away. The only honest answer at the moment is that we do not know.

References
Bateson, W. 1922. Evolutionary faith and modern doubts. Science 55:55-61.
Crampton, H. E. 1916. Studies on the Variation, Distribution, and Evolution of the Genus Partula, the Species Inhabitinlg Tahiti. Publ. No. 228. Washington, D.C.: Carnegie Inst. of Washington.
1932. Studies on the Variation, Distribution, and Evolution of the Genus Partula. The Species Inhabiting Moorea. Publ. No. 410. Washington, D.C.: Carnegie Inst. of Washington.
92 Stephen 1. Gould

Dobzhansky, Th. 1937. Genetics and the Origin of Species. New York Columbia Univ. Press.
1951. Genetics and the Origin of Species, 3rd ed. New York: Columbia Univ Press.
Ford, E. B. 1963. Ecological Genetics. London: Methuen.
Goldschmidt, R. 1940. The Material Basis of Evolution. New Haven: Yale Univ. Press.
Gould, S. J. 1966. Allometry and size in ontogeny and phylogeny. Biol. Rev. 41 :587-640.
1980. G. G. Simpson, Paleontology and the Modern Synthesis. In The Evolutionary Synthesis, ed. E. Mayr and W. B. Provine, pp. 153-72. Cambridge, Mass.: Harvard Univ. Press.
1982a. Introduction to T. Dobzhansky, Genetics and the Origin of Species, pp. xvii-xli. The Columbia Classics in Evolution Series, ed. N. Eldridge and S. J. Gould. New York: Columbia Univ. Press.
1982b. Darwinism and the expansion of evolutionary theory. Science 216:380-7.
1982c. The Uses of Heresy: An Introduction to Richard Goldschmidt's "The Material Basis of Evolution," pp. xiii-xlii. New Haven and London: Yale Univ. Press.
Gould, S. J., and R. C. Lewontin. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc. Roy. Soc. Lond. B 205:581-98.
Gould, S. J., and E. S. Vrba. 1982. Exaptation - a missing term in the science of form. Paleobiology S(1):4-15.
Huxley, J. 1932. Problems of Relative Growth. London: MacVeagh.
1942. Evolution, the Modern Synthesis. London, Allen & Unwin.
1960. The Emergence of Darwinism. In Evolution after Danvin, Vol. 1, ed. S. Tax, pp. 1-21. Chicago: Univ. of Chicago Press.
Kinsey, A. C. 1930. The gall wasp genus Cynips, a study in the origin of species. Indiana Univ. Studies, Vol. 16.
1937. Supra-specific variation in nature and in classification, from the viewpoint of zoology. Amer. Natural. 71:206-22.
Lack, D. 1945. The Galapagos finches (Geospizinae). Occasional Papers of the California Academy of Sciences, No. 21.
1947. Darwin's Finches. Cambridge: Cambridge Univ. Press.
1960. Darwin's Finches, 1960 reprint with preface. New York: Harper Torchbooks.
Mayr, E. 1942. Systematics and the Origin of Species. New York: Columbia Univ. Press.
1963. Animal Species and Evolution. Cambridge, Mass: Belknap Press of Harvard Univ.
1980. Some Thoughts on the History of the Evolutionary Synthesis. In Tile Evolutionary Synthesis, ed. E. Mayr and W. B. Provine, pp. 1-48. Cambridge, Mass.: Harvard Univ. Press.
Provine, W. B. Ms., "Adaptation and mechanisms of evolution after Darwin:

Hardening of Modern Synthesis 93

a study in persistent controversies." Paper presented at conference "Persistent controversies in evolutionary theory," Univ. of Chicago, March 1982.
Robson, G. C., and O. W. Richards. 1936. The Variation of Animals in Nature. London: Longmans, Green.
Simpson, G. G. 1944. TenlA70 and Mode in Evolution. New York: Columbia Univ. Press.
1949. Rates on Evolution in Animals. In Genetics, Paleontology and Ez101lition, ed. G. L. Jepsen, E. Mayr, and G. G. Simpson, pp. 205-28. Princeton: Princeton Univ. Press.
1953. The Major Features of El1olutiorl. New York: Columbia Univ. Press.
Tax, S. 1960. Evolution after Darwin. Chicago, Univ. of Chicago Press, 3 vols.
Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97-159.
1932. The'Roles of Mutation, Inbreeding, Crossbreeding, and Selection. Proceedings of the Sixtil International Congress of Genetics, Vol. 1 pp. 35666.