The Structure of Evolutionary Theory

by Stephen Jay Gould

  WEISMANN'S ARGUMENT ON LAMARCK AND THE ALLMACHT OF SELECTION

  I first learned about August Weismann in high school biology as the man who “disproved” Lamarckism by cutting off mouse tails for numerous genera­tions and noting the fully retained tails of all offspring (a good example of terrible teaching based upon the myth of crucial experiments as the source of all insight in science). Weismann did perform these experiments (1888, in 1891, pp. 431-461), but they (by his own admission) did little to combat Lamarckism, which is, as supporters parried, a theory about the inheritance of functional adaptations, not of sudden and accidental mutilations.

  Weismann's strong anti-Lamarckian argument does not rest upon an experiment, or an empirical observation at all. The rejection of soft inheritance arises as a logical deduction from Weismann's most distinctive contribution — his theory of inheritance and the continuity of germ-plasm (1885, in 1891, pp. 163-256). If germ-plasm is “immortal” (by passage across generations) and soma-plasm limited in existence by the death of each multicellular organ­ism; and if germ-plasm is sequestered early in ontogeny (“locked away” as the guardian of posterity, and protected from all somatic influence); then Lamarckian inheritance becomes structurally impossible because acquired somatic adaptations cannot affect the protected germ plasm. Weismann wrote in his Allmacht paper (1893, p. 608): “Nature has carefully enclosed the germ-plasm of all germ-cells in a capsule, and it is only yielded up for the formation of daughter-cells, under most complicated precautionary con­ditions.”

  Once Lamarckian inheritance becomes impossible, Weismann's argument for the Allmacht of selection proceeds in four logical steps. This fourfold de­velopment will strike most modern scientists as curious and unsatisfactory, for the sequence not only requires no empirical contribution, but actively de­nies the possibility of effective input from this conventional source of scien­tific affirmation. The argument breaks no rules of logic, but several of its premises are (to say the least) not self-evidently true.

  1. Adaptation is ubiquitous in nature; explaining adaptation therefore be­comes the chief goal of evolutionary theory. As “the greatest riddle that living Nature presents to us” (1909, p. 18), Weismann identified “the purposiveness of every living form relative to the conditions of its life, and its marvelously exact adaptation to these” (loc. cit.).

  I believe it can be clearly proved that the wing of a butterfly is a tablet on which Nature has inscribed everything she has deemed advantageous to the preservation and welfare of her creatures, and nothing else (1896, p. 5). [Page 202]

  Everything we see in animals is adaptation, whether of today, or of yesterday . . . Every kind of cell... is adapted to absolutely definite and specific functions, and every organ which is composed of these different kinds of cells contains them in the proper proportions, and in a particu­lar arrangement which best serves the function of the organ . . . The or­ganism as a whole is adapted to the conditions of its life, and it is so at every stage of its evolution” (1909, pp. 64-65. This statement comes from Weismann's contribution to the “official” centennial celebration of Darwin's birth. Thus, Weismann chose to honor Darwin by stressing panselection.)

  2. Adaptation must be attributed either to some materialistic cause, or to teleology (in the classic sense of spiritually directed purpose). The validity of science depends upon our ability to supply explanations in the former mode.

  3. Among materialistic proposals, only Lamarckism and natural selection can explain adaptation — for adaptation is ubiquitous and clearly too com­plex to ascribe to chance or to render as a side consequence of any process serving unrelated ends.

  4. Since Lamarckism is logically impossible (under the doctrine of continu­ity of germ-plasm), selection must be correct. To assert the Allmacht of selec­tion, we need no evidence beyond the disproof of Lamarckism. In fact, given the complexities of nature, and our inability to reconstruct past conditions in detail, we probably could not supply adequate direct evidence in any sin­gle case.

  We accept it, not because we are able to demonstrate the process in de­tail, not even because we can with more or less ease imagine it, but sim­ply because we must, because it is the only possible explanation that we can conceive. For there are only two possible a priori explanations of adaptations for the naturalist — namely, the transmission of functional adaptations [i.e. Lamarckism] and natural selection; but as the first of these can be excluded, only the second remains . . . We are thus able to prove by exclusion the reality of natural selection, and once that is done, the general objections which are based on our inability to demonstrate selec­tion-value in individual cases, must collapse, as being of no weight... It does not matter whether I am able to do so or not, or whether I could do it well or ill; once it is established that natural selection is the only princi­ple which has to be considered, it necessarily follows that the facts can be correctly explained by natural selection (1893, pp. 336-337).

  In 1893, when he made this bold assertion to counter Spencer's claim for the “inadequacy of natural selection,” Weismann advocated a kind of double exclusivity — for natural selection over Lamarckism, and for selection upon organisms as the only mode of Darwinian action. As a terminological matter, Weismann equated the general phrase “natural selection” with selection upon organisms alone (“personal selection” in his words. For example, he wrote (1903, vol. 2, p. 126): “It is upon this that the operation of natural selection, [Page 203] that is, personal selection, must depend”). However, spurred by Spencer's cri­tique, he soon expanded the boundaries of selection to include other levels of nature's hierarchy.

  THE PROBLEM OF DEGENERATION AND WEISMANN'S IMPETUS

  FOR GERMINAL SELECTION

  As discussed in the last chapter, the primary and standard refutation of Dar­winism by late 19th century evolutionists held that natural selection could eliminate, but not create — and that some other factor must therefore be iden­tified to explain the origin of adaptations and species. For example, T. H. Morgan wrote in 1905, before he became a Darwinian: “It appears that new species are born; they are not made by Darwinian methods, and the theory of natural selection has nothing to do with the origin of species, but with the survival of already formed species” (in Kellogg, 1907, p. 95).

  Darwinians, of course, understood this challenge, and responded with the argument that differential survivals, long cumulated, produce gradual and substantial changes meriting the designation “creative.” Weismann himself, for example (1896, p. 1), spoke of “the opposition of our own day, which contends that selection cannot create but only reject, and which fails to see that precisely through this rejection its creative efficiency is asserted.”

  On this contentious question of creativity, several standard anti-Darwinian arguments invoked the earliest stages of features easily recognized as adaptive in their perfected form, for selection can preserve and accentuate a feature fully in place, but how can an organism move from an initial “there” to a fully functional “here”? Two claims predominated (see Mivart, 1871, for the classic statement that provoked Darwin's own response in later editions of the Origin): first, that initial steps are too small to provide any conceivable benefit in selection; and second, that earliest stages cannot initiate the final function in any sense (a bird cannot fly with 5 percent of a wing).

  Darwinians developed satisfactory responses to both arguments about incipient stages of useful structures — the palpable value of tiny benefits for the first, and the principle of functional shift (preadaptation) for the second (see extensive discussion in Chapter 11, pp. 1218–1246). But the same problem seemed far more acute for the opposite dilemma of degeneration. Incipient stages of useful structures posed enough difficulties, although ultimate adaptiveness did suggest a Darwinian solution. But what conceivable pressure of natural selection could account for gradual stages in the disappearance of a functionless organ — for loss of function should remove a structure from the domain of selection entirely, and knowledge a
bout an eventually adaptive state could not be invoked to guide an explanation for intermediary stages along such a functionless path.

  (We might designate this problem by its classic example — the complete disappearance of eyes in some cave fishes. Despite a century of adequate Darwinian explanation, this issue continues to provide a rallying point for vernacular Lamarckism. I can testify to this in a personal way. As a result of [Page 204] writing more than 300 popular monthly essays on evolutionary topics during the past 25 years, I have become a statistically adequate sampling point, through thousands of letters received from lay readers, for both the fre­quency and intensity of standard confusions about our profession. I can tes­tify that three items top the list of puzzlement: (1) evolution seen as ana­genesis rather than branching (“if humans evolved from apes, why are apes still around”); (2) panselectionism (“what is the adaptive significance of male nipples”); and (3) Lamarckism and the failure of natural selection (“doesn't the blindness of cave fishes imply a necessary space for Lamarckian evolution by disuse”).)

  The problem of incipiency in degeneration poses more difficulty than the opposite issue of construction — for what can mediate the sequence if selec­tion does not regulate the final outcome? Weismann struggled to encompass this issue with his favored apparatus of Allmacht for selection — and he failed. Degeneration acted as the lever that pried Weismann from his panselec­tionism, and led him through a chronological series of honorable changes that must be read, in one sense, as retreats from a former pugnacious insis­tence on Allmacht, but that also represents a complexification and strength­ening of his original views.

  Consider the example that Spencer raised with such effectiveness against Weismann, and that eventually prompted the theory of germinal selection — reduction of hind limbs in some whales to tiny vestiges with no exterior ex­pression at all. Two classical explanations had been invoked by panselectionists: (1) the limbs became so reduced by ordinary negative selection, as a consequence of the hindrances they imposed upon efficient, streamlined swimming; (2) the limbs are not, in themselves, harmful, but energy invested in any useless structure must handicap a creature relative to conspecifics with fewer vestiges and neutral organs.

  Weismann invoked these standard arguments, but he became convinced (long before his debate with Spencer) that only part of the puzzle could be re­solved thereby. Selection would reduce the limbs to some degree (perhaps considerably), but surely the increments of further reduction soon become too small for granting a continuing, believable role to selection. Consider the figures that Spencer presents (1893b, p. 25), based on the efforts of a Dr. Struthers of Aberdeen, who had “kindly taken much trouble in furnishing the needful data, based upon direct weighing and measuring and estimation of specific gravity.” Spencer cites a Greenland Right Whale, weight 44,800 pounds, femur weight, 3-1/2 ounces; and a Razorback at 56,000 pounds, with a femur weight of 1 ounce — “so that these vanishing remnants of hind limbs weighed but 1/896,000th part of the animal.” Could one possibly be­lieve that a profound relative, but inconsequential absolute, reduction — from a two-ounce to a one-ounce femur, for example — might materially aid stream­lining (especially since external expression had disappeared long before) or conserve meaningful energy? Weismann accepted the implausibility of such a claim and recognized that he would have to seek an explanation beyond [Page 205] organismal selection for such late stages in the reduction of degenerate or­gans. “To use Herbert Spencer's striking illustration, how could the balance between life and death, in the case of a colossus like the Greenland whale, be turned one way or another by the difference of a few inches in the length of the hind-leg, as compared with his fellows, in whom the reduction of the hind-limb may not have gone quite so far? . . . Further reduction to their modern state of great degeneration and absolute concealment within the flesh of the animal cannot be referred even to negative selection” (Weismann, 1903, vol. 2, p. 114).

  This example, and the general phenomenon of degeneration, deeply trou­bled Weismann because common sense seemed to demand that his Lamarckian bugbear and bogeyman — so recently and, as he thought, finally and effec­tively buried — be disinterred to explain reduction as inheritance of features shriveled by disuse. Spencer himself raised this example in order to defend a Lamarckian explanation prima facie:

  Thus, the only reasonable interpretation is the inheritance of acquired characters. If the effects of use and disuse, which are known causes of change in each individual, influence succeeding individuals . . . then this reduction of the whale's hind limbs to minute rudiments is accounted for. The cause has been unceasingly operative on all individuals of the species ever since the transformation began. In one case see all. If this cause has thus operated on the limbs of the whale, it has thus operated in all crea­tures on all parts having active functions (Spencer, 1893b, p. 26).

  Weismann first attempted to resolve the difficulties posed by degeneration with his hypothesis of panmixia (not the later Fisherian definition now famil­iar to evolutionists). By panmixia, Weismann referred to the effect of recom­bination in sexual reproduction (amphimixis in his vocabulary) upon organs no longer subject to selection. When selection operates, Weismann argued, organs will be actively maintained, with constant vigilance and no relaxation, at the peak of their potential size and complexity by elimination of individu­als bearing substandard parts. But as soon as selection ceases to act, formerly “substandard” attributes will no longer be eliminated; they now mix freely with “good” parts, and the organ slides, by continuous dilution, down an in­clined plane towards total elimination. In a poignant example (since poor eyesight plagued his own career), Weismann wrote (1903, pp. 114-115). “If this conservative action of natural selection secures the maintenance of the parts and organs of a species at their maximum of perfection, it follows that these will fall below this maximum as soon as the selection ceases to operate ... Those with inferior organs of vision will, ceteris paribus, produce as good offspring as those with better eyes, and the consequence of this must be that there will be a general deterioration of eyes, because the bad ones can be transmitted as well as the good, and thus the selection of good eyes is made impossible.”

  By his own admission and explicit defense (see p. 201), Weismann's argument [Page 206] for Allmacht, and against Lamarckian inheritance, rested upon a log­ical structure of inferences from premises, not upon observation — for an em­pirical approach, Weismann held, could not achieve resolution, given the impossibility of “seeing,” at their minute sizes, the material bearers of hered­ity. Panmixia did compromise Allmacht in a sense, for this process yielded evolutionary change without selection. But following Kellogg's key distinc­tion of auxiliary from contradictory hypotheses (see pp. 163–169), panmixia worked as an adjunct and aid — a mopping-up operation for organs fallen below the purview of selection, and, more importantly, a moat to prevent the incursion of a true enemy, the antiselectionist forces of Neo-Lamarckism. (Lamarck battled against Darwin for the common ground of universal adap­tation, while panmixia only worked to finish what selection had started, and only in the limited domain of degeneration.)

  But Weismann's panmixia, having no support beyond the internal logic of the argument itself, could not survive the detection and exposure of crucial flaws. Spencer was not the first writer to illustrate the weaknesses of pan­mixia, but the debate of 1893 does mark Weismann's last attempt to explain degeneration by panmixia alone, and therefore contains the seeds for his next and final attempt — the theory of germinal selection.

  Spencer, referring to “the vexed question of panmixia” (1893b, p. 22), of­fered three major rebuttals. “When from the abstract statement of it we pass to a concrete test, in the case of the whale, we find that it necessitates an un­proved and improbable assumption respecting plus and minus variation; that it ignores the unceasing tendency to reversion; and that it implies an effect out of all proportion to the cause” (1893b, pp. 28-29). The second point, based on Gal
ton's principle of regression to the mean, denies that “minus” varia­tions can continue to accumulate differentially; the third brands panmixia as too weak a force to secure the total elimination of a useless organ. The first argument, however, proved to be not only decisive in itself, but unusual in sci­entific discourse by accusing Weismann (correctly) of conflating linguistic us­age with biological reality.

  Weismann continually argued that selection maintained an organ “at its highest level.” Relaxation of selection might then impel an accumulation of previously eliminated variation in the minus direction only. But, as Spencer and others protested, why should selective optimization hold an organ at the summit of its potential size and complexity. Shouldn't optimality lie some­where in the middle of a possible range, with selective elimination of both plus and minus variations? “Take the case of the tongue,” Spencer argued (1893b, pages 23-24). “Certainly there are tongues inconveniently large, and probably tongues inconveniently small. What reason have we for assuming that the inconveniently small tongues occur more frequently than the incon­veniently large ones?” Without the invalid metaphor of selective summits, panmixia cannot reduce an organ to oblivion, for release from selection does not impart an inexorably downward trend to preserved variation.