Hen’s Teeth and Horse’s Toes

by Stephen Jay Gould

  I got into Cuvier’s sanctum sanctorum yesterday, and it is truly characteristic of the man. In every part it displays that extraordinary power of methodising which is the grand secret of the prodigious feats which he performs annually without appearing to give himself the least trouble…. There is first the museum of natural history opposite his house, and admirably arranged by himself, then the anatomy museum connected with his dwelling. In the latter is a library disposed in a suite of rooms, each containing works on one subject. There is one where there are all the works on ornithology, in another room all on ichthyology, in another osteology, in another law books! etc., etc…. The ordinary studio contains no bookshelves. It is a longish room, comfortably furnished, lighted from above, with eleven desks to stand to, and two low tables, like a public office for so many clerks. But all is for the one man, who multiplies himself as author, and admitting no one into this room, moves as he finds necessary, or as fancy inclines him, from one occupation to another. Each desk is furnished with a complete establishment of inkstand, pens, etc…. There is a separate bell to several desks. The low tables are to sit to when he is tired. The collaborators are not numerous, but always chosen well. They save him every mechanical labour, find references, etc., are rarely admitted to the study, receive orders and speak not.

  Cuvier has suffered primarily because posterity has deemed incorrect the two cardinal conclusions that motivated his work in biology and geology—his belief in the fixity of species and his catastrophism. Since being wrong is a primary intellectual sin when we judge the past by its approach to current wisdom, dubious motives must be ascribed to Cuvier. How else can one explain why such a brilliant man went so far astray? Cuvier then becomes an object lesson for aspiring scientists. Cuvier must have failed because he allowed prejudice to cloud objective truth. Conventional theology must have dictated both his creationism and the geological catastrophism that supposedly squeezed our earth into the Mosaic chronology. Consider this assessment of Cuvier presented by a leading modern textbook in geology:

  Cuvier believed that Noah’s flood was universal and had prepared the earth for its present inhabitants. The Church was happy to have the support of such an eminent scientist, and there is no doubt that Cuvier’s great reputation delayed the acceptance of the more reasonable views that ultimately prevailed.

  I devote this essay to defending Cuvier (who ranks, in my judgment, with Darwin and Karl Ernst von Baer as the greatest of nineteenth-century natural historians). But I do not choose to do so in the usual manner of historians—by showing that Cuvier’s beliefs were not rooted in irrational prejudice, but both arose from and advanced beyond the social and scientific context of his own time. Nor (obviously) will I defend Cuvier’s creationism or more than a sliver of his catastrophism. Instead, I want to argue that Cuvier used the very doctrines for which he stands condemned—creationism and catastrophism—as specific and highly fruitful research strategies for establishing the basis of modern geology—the stratigraphic record of fossils and its attendant long chronology for earth history. Some types of truth may require pursuit on the straight and narrow, but the pathways to scientific insight are as winding and complex as the human mind.

  Cuvier is often portrayed as an armchair speculator because his conclusions are now regarded as incorrect and error supposedly arises from aversion to hard data. In fact, he was a committed empiricist. He railed against the prevalent tradition in geology for constructing comprehensive “theories of the earth” with minimal attention to actual rocks and fossils. “Naturalists,” he wrote, “seem to have scarcely any idea of the propriety of investigating facts before they construct their systems.” (Cuvier correctly includes Hutton, subject of essay 6, among the system builders, although he confesses more sympathy for his Scottish colleague than for most of his ilk.)

  Instead, Cuvier argues, we must seek some empirical criterion for unraveling the earth’s history. But what shall it be? What has changed with sufficient regularity and magnitude to serve as a marker of time? Cuvier recognized that the lithology of rocks would not do, since limestones and shales look pretty much alike whether they occur at the tops or bottoms of stratigraphic sequences. What about the fossils entombed in rocks?

  The idea that fossils reflect history is now so commonplace, we tend to regard it as an ancient truth. It was, however, a contentious issue in Cuvier’s day, when debate centered on whether or not species could become extinct—for without extinction, all creatures are coeval and fossils cannot measure time (unless new forms keep accumulating and we can date rocks by first appearances. But a finite earth would seem to preclude continual addition with no subtraction).

  Many of Cuvier’s illustrious contemporaries (including Thomas Jefferson who, when not preoccupied with other matters, devoted a paper to the subject) argued strongly that extinction could not occur. Cuvier decided that the a priori (and often explicitly biblical) defenses of nonextinction were worthless and that the issue would have to be decided empirically. But previous studies of fossil vertebrates (his specialty) had been undertaken in the mindless manner of mere collection. Fossils had been gathered primarily as curiosities—but scientists must ask questions and collect systematically in their light.

  Other naturalists, it is true, have studied the fossil remains of organized bodies; they have collected and represented them by thousands, and their works will certainly serve as a valuable storehouse of materials. But, considering these fossil plants and animals merely in themselves, instead of viewing them in their connection with the theory of the earth; or regarding their petrifactions…as mere curiosities, rather than historical documents…they have almost always neglected to investigate the general laws affecting their position, or the relation of the extraneous fossils with the strata in which they are found.

  Cuvier then provides a two-page compendium of questions, an empiricist’s vade mecum to combat the older speculative tradition.

  Are there certain animals and plants peculiar to certain strata and not found in others? What are the species that appear first in order, and those which succeed? Do these two kinds of species ever accompany one another? Are there alternations in their appearance; or, in other words, does the first species appear a second time, and does the second species then disappear?

  But this research program for establishing a geological record cannot work unless extinction is a common fact of nature—and ancient creatures are therefore confined to rocks of definite and restricted ages. Cuvier’s great four-volume work (Recherches sur les ossemens fossiles, “studies on fossil bones”) is a long demonstration that fossil bones belong to lost worlds of extinct species.

  Cuvier used the comparative anatomy of living vertebrates to assign his fossils to extinct species. Since fossils come in bits and pieces, a tooth here or a femur there, some method must be devised to reconstruct a whole from scrappy parts and to ascertain whether that whole still walks among the living. But what principles shall govern the reconstruction of wholes from parts? Can it be done at all? Cuvier recognized that he must study the anatomy of modern organisms—where we have unambiguous wholes—to learn how to interpret fragments of the past. The second paragraph of his essay presents this program for research:

  As an antiquary of a new order, I have been obliged to learn the art of deciphering and restoring these remains, of discovering and bringing together, in their primitive arrangement, the scattered and mutilated fragments of which they are composed…. I had…to prepare myself for these enquiries by others of a far more extensive kind, respecting the animals which still exist. Nothing, except an almost complete review of creation in its present state, could give a character of demonstration to the results of my investigations into its ancient state; but that review has afforded me, at the same time, a great body of rules and affinities which are no less satisfactorily demonstrated; and the whole animal kingdom has been subjected to new laws in consequence of this Essay on a small part of the theory of the earth.

  As his cardinal
rule for reconstruction, Cuvier devised a principle that he called “correlation of parts.” Animals are exquisitely designed and integrated structures—perfect Newtonian machines of a sort. Each part implies the next, and a whole lies embodied in the implications of any fragment—a grand version of that immortal commentary on Ezekiel’s vision, “the foot bone’s connected to the ankle bone….”

  Cuvier presents the law of correlation as if it could be applied by reason alone, using the principles of animal mechanics:

  Every organized individual forms an entire system of its own, all the parts of which mutually correspond and concur…. Hence none of these separate parts can change their forms without a corresponding change in the other parts of the same animal, and consequently each of these parts, taken separately, indicates all the other parts to which it has belonged…. If the viscera of an animal are so organized as only to be fitted for the digestion of recent flesh, it is also requisite that the jaws should be constructed as to fit them for devouring prey; the claws must be constructed for seizing and tearing it to pieces; the teeth for cutting and dividing its flesh; the entire system of the limbs, or organs of motion, for pursuing and overtaking it; and the organs of sense, for discovering it at a distance…. Thus, commencing our investigation by a careful survey of any one bone by itself, a person who is sufficiently master of the laws of organic structure, may, as it were, reconstruct the whole animal to which that bone had belonged.

  Cuvier’s principle of correlation lies behind the popular myth that paleontologists can see an entire dinosaur in a single neck bone. (I believed this legend as a child and once despaired of entering my chosen profession because I could not imagine how I could ever obtain such arcane and wondrous knowledge.) Cuvier’s principle may well apply in the most general sense: if I find a jaw with weak peglike teeth, I do not expect to find the sharp claws of a carnivore on the accompanying legs. But a single tooth will not tell me how long the legs were, how sharp the claws, or even how many other teeth the jaw held. Animals are bundles of historical accidents, not perfect and predictable machines.

  When a paleontologist does look at a single tooth and says, “Aha, a rhinoceros,” he is not calculating through laws of physics, but simply making an empirical association: teeth of this peculiar form (and rhino teeth are distinctive) have never been found in any animal but a rhino. The single tooth implies a horn and a thick hide only because all rhinos share these characters, not because the deductive laws of organic structure declare their necessary connection. Cuvier, in fact, knew perfectly well that he operated by empirical association (and not by logical inference), although he regarded his observational method as an imperfect way station to a future rational morphology:

  As all these relative conformations are constant and regular, we may be assured that they depend upon some sufficient cause; and, since we are not acquainted with that cause, we must here supply the defect of theory by observation, and in this way lay down empirical rules on the subject, which are almost as certain as those deduced from rational principles, especially if established upon careful and repeated observation. Hence, any one who observes merely the print of a cloven hoof, may conclude that it has been left by a ruminant animal, and regard the conclusion as equally certain with any other in physics or in morals.

  Since Cuvier didn’t know the laws of rational morphology (we now suspect that they do not exist in the form he anticipated), he proceeded by his favorite method of empirical cataloging. He amassed an enormous collection of vertebrate skeletons, and noted an invariant association of parts by repeated observation. He could then use his catalog of recent skeletons to decide whether fossils belong to extinct species. The earth, he argued, has been explored with sufficient care (for large terrestrial mammals at least) that fossil bones outside the range of modern skeletons must represent vanished species.

  The four volumes of the 1812 treatise form a single long argument for the fact of extinction, the resultant utility of fossil vertebrates for ascertaining the relative ages of rocks, and the consequent antiquity of the earth. The introductory Discours préliminaire sets out basic principles. In the first technical monograph, on mummified remains of the Egyptian ibis, Cuvier finds no difference between modern birds and fossils from the beginning of recorded history as then construed. The present creation therefore has considerable antiquity; if extinct species inhabited still earlier worlds, then the earth must be truly ancient. The next set of monographs discusses the detailed anatomy of large mammals found in the uppermost geological strata—Irish elks, woolly rhinos, and a variety of fossil elephants (mammoths and mastodons). They are similar to modern relatives, but the sizes and shapes of their fragmentary bones lie outside modern ranges and will not correlate with the normal skeletons of living forms (no modern deer could hold up the antlers of an Irish elk). Hence, extinction has occurred and life on earth has a history. The final monographs demonstrate that still older bones belonged to creatures even more unlike modern species. Life’s history has a direction—and great antiquity if it has passed through so many cycles of creation and destruction.

  Mummified skeleton of an Egyptian ibis, from Cuvier’s Ossemens fossiles of 1812. Cuvier showed that this bird is identical with modern ibises and that no organic changes had occurred during the long period from ancient Egypt to today. Since so many changes had occurred in earlier periods, the earth must be ancient.

  Cuvier did not give an evolutionary interpretation to the direction that he discerned, for the very principle that he used to establish extinction—the correlation of parts—precluded evolution in his mind. If an animal’s parts are so interdependent that each one implies the exact form of all others, then any change would require a total remodeling of an entire body, and what process can accomplish such a complete and harmonious change all at once? The direction of life’s history must reflect a sequence of creations (and subsequent extinctions), each more modern in character. (We would not deny Cuvier’s inference today, but only his initial premise of tight and ubiquitous correlation. Evolution is mosaic in character, proceeding at different rates in different structures. An animal’s parts are largely dissociable, thus permitting historical change to proceed.)

  Thus, ironically, the incorrect premise that has sealed Cuvier’s poor reputation today—his belief in the fixity of species—was the basis for his greatest contribution to human thought and hard-nosed empirical science: a proof that extinction grants life a rich history and the earth a great antiquity. (I note the further irony that Cuvier’s creationism—good science in his time—disproved, more than 150 years ago, the linchpin of modern fundamentalist creationism: an age of but a few thousand years for the earth—see essays of section 5).

  Cuvier’s reputation took a second strike from his adherence to (and partial invention of) the geological theory of catastrophism, a complex doctrine of many parts, but focusing on the claim that geological change is concentrated in rare episodes of paroxysm on a global or nearly global scale: floods, fires, the rise of mountains, the cracking and foundering of continents—in short, all the components of traditional fire and brimstone. Cuvier, of course, linked his catastrophism to his theory of successive creations and extinctions by identifying geological paroxysms as the agent of faunal debacles.

  A perverse reading of history had led to the usual claim—as in the textbook assessment of Cuvier cited earlier—that catastrophism was an antiscientific feint by a theological rear guard laboring to place Noah’s flood under the aegis of science, and to justify a compression of earth’s history into the Mosaic chronology. Of course, if the earth is but a few thousand years old, then we can only account for its vast panoply of observed changes by telescoping them into a few episodes of worldwide destruction. But the converse does not hold: a claim that paroxysms sometimes engulf the earth dictates no conclusion about its age. The earth might be billions of years old, and its changes might still be concentrated in rare episodes of destruction.

  Cuvier’s eclipse is awash in ir
ony, but no element of his denigration is more curiously unfair than the charge that his catastrophism reflects a theological compromise with his scientific ideals. In the great debates of early-nineteenth-century geology, catastrophists followed the stereotypical method of objective science—empirical literalism. They believed what they saw, interpolated nothing, and read the record of the rocks directly. This record, read literally, is one of discontinuity and abrupt transition: faunas disappear; terrestrial rocks lie under marine rocks with no recorded transitional environments between; horizontal sediments overlie twisted and fractured strata of an earlier age. Uniformitarians, the traditional opponents of catastrophism, did not triumph because they read the record more objectively. Rather, uniformitarians, like Lyell and Darwin, advocated a more subtle and less empirical method: use reason and inference to supply the missing information that imperfect evidence cannot record. The literal record is discontinuous, but gradual change lies in the missing transitions. To cite Lyell’s thought experiment: if Vesuvius erupted again and buried a modern Italian town directly atop Pompeii, would the abrupt transition from Latin to Italian, or clay tablets to television, record a true historical jump or two thousand years of missing data? I am no partisan of gradual change, but I do support the historical method of Lyell and Darwin. Raw empirical literalism will not adequately map a complex and imperfect world. Still, it seems unjust that catastrophists, who almost followed a caricature of objectivity and fidelity to nature, should be saddled with a charge that they abandoned the real world for their Bibles.