The Structure of Evolutionary Theory

by Stephen Jay Gould

  This work also led Goldschmidt to what many scientists regard as his most enduring contribution — the naming and characterization of “phenocopies.” If genes affect timing, then experimental manipulations of tempera­ture and chemical environment might induce changes identical with those found in mutants, thus confirming the rate hypothesis. Goldschmidt pro­duced such “mutant” phenotypes without mutations and christened them “phenocopies.” He maintained great fondness for this subject and for his dis­coveries in this area. Indeed, the very last words of his posthumously pub­lished autobiography (1960, p. 326) do not proffer cosmic advice, but merely state: “It is my greatest intellectual happiness that I can still work in my labo­ratory and even make interesting discoveries in the field of chemically in­duced phenocopies.”

  Goldschmidt's later apostasy on macroevolution may be traced to this personal source in his early work on development, particularly to his early recognition that small genetic changes, operating early enough in ontogeny, may engender cascading effects towards a large phenotypic jump in a single ge­netic step.

  3. The very term “hopeful monster,” and the form of evidence adduced by Goldschmidt in support, establishes the developmental theme as primary. Why did Goldschmidt use such an odd term at all, an apparently flippant phrase (and, therefore, a poor rhetorical strategy for pushing heterodox views) al­most guaranteed to generate rebuke from upholders of orthodox (and dull) academic prose? In part, of course, the term began in whimsy, and then flowed too far on winds of circumstance. Goldschmidt recalled in his autobi­ography (1960, p. 318): “I spoke half jokingly of the hopeful monster in my first publication on the subject.” But, in another sense, the term could not have been more apt or appropriate once one recognizes that ontogenetic de­velopment — not systemic mutation — undergirds the concept.

  What makes a monster hopeful? Goldschmidt identifies two necessary and sufficient conditions:

  (1) The mutant must, by good fortune, be well fitted for a particular, previously unexploited environment in its vicinity — the Darwinian, or functional theme. A mutant rat with fused tail vertebrae is just a monster; a proto-bird, with feathers better positioned for flight as a fortuitous consequence of a sim­ilar fusion, becomes a hopeful monster. An ordinary nektonic teleost fish with [Page 461] both eyes on one side of the head is only a monster; a benthic flatfish with both eyes on the head's upper surface, with better scanning of surroundings as a lucky result, becomes a hopeful monster. A short and bow-legged dog is merely a monster; a frankfurter that can drag badgers from their holes is a hopeful monster.

  (2) More importantly, hopeful monsters must pass a developmental crite­rion. The vast majority of mutations with large phenotypic effects are le­thal — that is, just monsters. However, certain rare mutations will produce extensive, but viable, phenotypic changes because they operate within the confines of a well regulated developmental system. Such changes yield work­able organisms (which may thrive if they become lucky enough to find a wel­coming environment), rather than inviable hodge-podges of unintegrated sys­tems in varying phases of ontogeny. The fecund macroevolutionary monster becomes potentially “hopeful” when all phenotypic effects unfold in a coor­dinated manner within a regulated developmental system. In his autobio­graphical statement, just before admitting that he had originated the term “hopeful monster” in partial jest, Goldschmidt linked his concept firmly to the developmental theme: “What addition to Darwinism was needed in order to account for the macroevolutionary processes? The solution was the exis­tence of macromutations, which, in rare cases, could affect early embryonic processes so that through the features of embryonic regulation and integra­tion at once a major step in evolution could be accomplished and fixed under certain conditions” (1960, p. 318).

  Invoking the classical formalist theme of constraints and channels, Goldschmidt argues that a knowledge of developmental systems and their norms of reaction can specify the range of perturbations that might yield hopeful monsters — a clear invocation of “developmental constraint” in its positive mode of enabling (see Chapter 10): “Within a constant genotype the potenti­alities of individual development may include a range of variation of the same phenotypic order of magnitude which otherwise characterizes large evolu­tionary steps based upon changes in the genotype. The norm of reaction thus shows that paths are available for changes in the genotype (mutations in the broadest sense) without upsetting normal developmental processes” (p. 260).

  Goldschmidt designates this creative constraining force in the last phrase — “without upsetting normal developmental processes.” If shifts to alternate pathways discombobulate development, then any resulting monster must be hopeless. Many intricately complex systems simply fall apart or change in injurious ways under the impact of major perturbations. But the regulation of organic systems has evolved to accommodate impacts and to integrate changes into canalized and viable pathways. Goldschmidt's famous phrase transcends whimsy or nonsense — once we grasp the intended developmental theme. Goldschmidt granted hope to his monsters because regulation can integrate certain large alterations of phenotype into viable systems of devel­opment.

  4. The origin and subsequent ontogeny of hopeful monsters (both the term and the concept) reveal a “smoking gun” for centrality of the developmental [Page 462] theme. Evolutionary biologists should honor world's fairs, despite their hoopla and crass commercialism — for Goldschmidt's work provides a second example of their spur to scientific progress. C. O. Whitman presented his most cogent defense of orthogenesis in pigeons (see pp. 383–394) in an ad­dress delivered at a meeting held in conjunction with the St. Louis fair of 1904 (also the source of the ice cream cone, several Scott Joplin rags, and the song “Meet me in St. Louis, Louis”). Richard Goldschmidt christened the term “hopeful monster” in an address at the AAAS meeting of 1933, held in conjunction with Chicago's World's Fair to celebrate a “century of progress.” Goldschmidt, representing the “Kaiser Wilhelm Institute for Biology, Berlin-Dahlem,” spoke on “some aspects of evolution.” In his closing paragraph (1933, p. 547), he coined the fateful term in summarizing his entire paper:

  I chose ... first, an aspect where I had to express skepticism in regard to well-established beliefs. I tried to show on the basis of large experimental evidence that the formation of subspecies or geographic races is not a step towards the formation of species but only a method to allow the spreading of a species to different environments by forming preadaptational mutations and combinations of such, which, however, always remain within the confines of the species. The second aspect, which I dis­cussed, was one where I felt again optimistic. I tried to emphasize the im­portance of the methods of normal embryonic development for an un­derstanding of possible evolutionary changes. I tried to show that a directed orthogenetic evolution is a necessary consequence of the embry­onic system, which allows only certain avenues for transformation. I fur­ther emphasized the importance of rare but extremely consequential mu­tations affecting rates of decisive embryonic processes, which might give rise to what one might term hopeful monsters, monsters that could start a new evolutionary line if fitting into some empty environmental niche.

  Two features of this citation (and of the whole article) clinch my argument. First, the article's structure provides an epitome for the book that Gold­schmidt would publish seven years later, and that would seal and symbolize his apostasy. The Material Basis of Evolution must have been written as an expansion of this outline — a two-part structure, with the first half (Gold­schmidt's self-styled “skepticism”) on the Darwinian character, but macro-evolutionary inefficacy, of adaptive differentiation within species; and the sec­ond half (Goldschmidt's proclaimed “optimism”) on a different style for macroevolution based on occasional saltation in a rare but viable mode, as embodied in the slightly whimsical phrase “hopeful monster.” Goldschmidt wrote the following statement in 1933, but he could not have composed better jacket copy for his 1940 book:

  A
t the beginning of this lecture I said that my mind, like that of many geneticists, is oscillating between skepticism and optimism with regard to the views on the means of evolution as derived from genetical work. I [Page 463] have now presented to you examples of both states of mind: First, a bit of skepticism with regard to the role which the formation of geographic races or subspecies may have played in evolution; and then a bit of opti­mism in trying to show that the physiological system underlying orderly development, on the basis of the genetic constitution, allows some of the larger steps in evolution to be understood as sudden changes by single mutations concerning the rate of certain embryological processes (1933, p. 546).

  This quotation would work as full jacket copy for Goldschmidt's later book — except for one omission. The quotation contains no statement at all, about systemic mutations or the attempt to construct a revolutionary, holistic genetics by denying the corpuscular gene. In other words, Goldschmidt devel­oped the full intellectual framework of his argument for the strict separation of micro- and macroevolution, and for the saltational basis of macroevolution, by invoking the developmental theme alone — that is, before he initiated his campaign for a revolutionary genetics (beginning in the late 1930's, and then continuing and intensifying to his death). The developmental theme en­joys both temporal priority and complete sufficiency. Goldschmidt devised the hopeful monster (both the term and concept) before he ever formulated his radical genetics. Moreover, the developmental theme can carry the argu­ment for saltational macroevolution all by itself. This conclusion, I think, re­solves the puzzle of textual confusion in The Material Basis of Evolution. Goldschmidt had constructed his outline by 1933, based on the developmen­tal theme alone. He began to formulate his radical genetics later, and then in­terpolated this material into a structure already established. These interpola­tions often seem hasty or haphazard, and Goldschmidt's chapters on systemic mutation do not always cohere with the earlier material. Ironically, the pas­sages on systemic mutation in The Material Basis work much like an ordinary “hopeless monster” in the organic world. They do not mesh with the coher­ent outline or developmental program of a book planned and coordinated long before!

  In introducing the developmental theme to carry his ideas on macro­evolution, Goldschmidt (1933, p. 543) states that biologists have long recog­nized the need to understand the genetic basis and selective advantage of major evolutionary changes — but that a crucial third component has been missing: “But there is a third point, often neglected, which lies, I think, at the basis of the whole problem, namely, the nature of the developmental system of the organism which is to undergo evolutionary change.” Goldschmidt then argues that his macroevolutionary ideas arose “as a logical consequence of my views on gene controlled development” (p. 544), with a key in the con­cept of alterations in rate: “The most probable mutational change with a chance to lead to a normal organism is a change in the typical rate of certain developmental processes” (p. 544). He then praises D'Arcy Thompson for lo­cating the phyletic meaning of these ideas in small mutational changes in rates, operating early in development to yield a saltational origin of new adult [Page 464] phenotypes: “Translated into phylogenetic language, this would mean that immense evolutionary effects could be brought about by changing the differ­ential growth rates of the whole body or organ at an early point in develop­ment, with all the necessary secondary effects of such a change” (p. 545). In rare cases, such ontogenetic cascades will produce a viable organism (by working within developmental channels) lucky enough to find a favorable en­vironment — in other words, a hopeful monster: “We certainly know of many cases of mutational shifts of the rate of certain developmental processes lead­ing to non-viable results, for example, caterpillars with pupal antennae, lar­vae of beetles with wings . . . But I cannot see any objection to the belief that occasionally, though extremely rarely, such a mutation may act on one of the few open avenues of differentiation and actually start a new evolutionary line” (p. 544).

  My pleasure in locating this resolution (in a 1933 article) for the textual difficulties in Goldschmidt's 1940 book then became enhanced when I noted another theme, by no means absent from the later book, but stressed in 1933 to a far greater extent, and with clearer purpose. I have repeatedly empha­sized, as the central notion of this chapter, that the full formalist (or internal­ist) critique of Darwinian functionalism embraces two themes, both illus­trated by Galton's incisive metaphor of the herkyjerky polyhedron — facet flipping (saltationism) and channeling (constraint in the positive sense of pre­ferred directions for change).

  One might expect that the chief apostate and whipping boy of ortho­doxy would embrace the full range of a coherent opposing philosophy. We usually view Goldschmidt as a pure saltationist, and the vehemence of ortho­dox reaction to only half a loaf might seem puzzling. But in the 1933 article, Goldschmidt gives equal weight to both internalist arguments — as he repeat­edly, and explicitly, ties the theme of channeling to its strongest version of orthogenesis. Saltation and channeling march in tandem throughout his argu­ment, and the entirety builds a satisfying version of the full formalist critique.

  I have already cited one Goldschmidtian invocation of orthogenesis linked to saltationism, from the concluding paragraph: “I tried to show that a di­rected orthogenetic evolution is a necessary consequence of the embryonic system which allows only certain avenues for transformation” (p. 547). But the two themes remain indissolubly connected throughout the article, and channeling receives as much attention as saltation — whereas Goldschmidt did emphasize saltation and downplay channeling in his later writings. In fact, in the 1933 article, Goldschmidt invokes channeling at the very begin­ning of his macroevolutionary discussion, just after citing the importance of development and even before he introduces the argument for saltation:

  A considerable number of developmental processes between egg and adult have to be changed, in order to lead to a different organization. Development, however, within a species is, we know, considerably one tracked. The individual developmental processes are so carefully interwoven and arranged so orderly in time and space that the typical result is [Page 465] only possible if the whole process of development is in any single case set in motion and carried out upon the same material basis.* Changes in this developmental system leading to new stable forms are only possible as far as they do not destroy or interfere with the orderly progress of developmental processes (p. 543).

  The explicit invocation of orthogenesis then follows (p. 544): “If there are only a few avenues free for the action of mutational changes without knock­ing out of order the whole properly balanced system of reactions, the proba­bility is exceedingly high that repeated mutations will go in the same direc­tion, will be orthogenetic . . . We have pointed out a long time ago and still hold that orthogenesis is not the result of the action of selection or of a mysti­cal trend, but a necessary consequence of the way in which the genes control orderly development — a way which makes only a few directions available to mutational changes.”

  Only now, after explicating the theme of channeling, does Goldschmidt introduce the rationale for saltation in its context: “But how about the possibil­ity of occasional successful mutational changes acting upon earlier develop­mental processes? Would such a change, if possible at all without breaking up the whole system of the orderly sequence of development, not at once have the consequence of changing the whole organization and bridging with one step the gap between taxonomically widely different forms?” (p. 544).

  Thereafter, as in the summary statement cited previously, Goldschmidt combines the two themes. He conjectures, for example, how saltation and orthogenesis might jointly explain phyletic sequences of limb rudimentation:

  Let us assume a mutational change in rate of differentiation of the limb-bud of a vertebrate . . . The consequent rudimentation of the organ would probably not interfere with orderly development of the organ­ism. Here, then, an avenue would be open to
considerable evolutionary change with a single basic step, provided that the new form could stand the test of selection, and that a proper environmental niche could be found to which the newly formed monstrosity would be preadapted and where, once occupied, other mutations might improve the new type. And in addition, the possibility for an orthogenetic line of limb-rudimentation would be a further consequence.

  In the extensive reading required to compose a chapter like this, one acquires great respect for rare scientists with the mental power, and basic thoughtfulness, to explore and integrate the full set of implications and rami­fications within great themes — and formalist vs. functionalist thinking must rank among the greatest of all biological themes (if only because this contrast expresses an attitude towards nature so deep and basic that the most impor­tant watershed in the history of biology — the development and acceptance of [Page 466] evolutionary theory — did not disrupt the discussion, but only reclothed an old antithesis in new language and causality).

  In the post-Darwinian debates, I feel that only four evolutionists in my study fully plumbed the depth of this dichotomy (Galton specified all the themes, and even developed the canonical metaphor of the polyhedron, but he was only dipping): C. O. Whitman, Hugo de Vries, William Bateson, and Richard Goldschmidt. The views of these four men embody important dif­ferences, and Goldschmidt emerges as the best standard bearer for the full version of formalism. Two of the four embraced one aspect of Galton's poly­hedron, but rejected the other for interesting reasons: Whitman as an orthogeneticist and gradualist; and de Vries as a saltationist who accepted the isotropy of species-level variation (and therefore constructed a higher-level Darwinism for trends among species). Bateson understood the connection and brought the themes together, but his generation hadn't gained enough knowledge about potential mechanisms to suggest more than an abstract and speculative synthesis. (Interestingly, Goldschmidt begins his 1933 article, his best presentation of the full critique, with a reference to Bateson's famous 1914 address to the British Association.)