One hardly expects revolutionary work after such a humdrum title in the conventional form of a taxon from a time and a place. But Stensiö chose understatement as an antidote to Patten’s quest for ultimates. Stensiö had found some exceptionally well-preserved head shields of Cephalaspis and other ostracoderms on the island of Spitsbergen. He realized that the unusually heavy ossification of the shield suggested an exciting possibility for research—for bone permeated and tightly surrounded all soft anatomy of the head, including delicate blood vessels and cranial nerves, not to mention the more prominent brain and eyes. The soft parts had decayed after death, and had been replaced by matrix of a much lighter color than the surrounding bone. By distinguishing bone from matrix, Stensiö could reconstruct the soft anatomy of Cephalaspis in astounding detail.
Stensiö used two basic methods for resolving the anatomy of ostracoderms. First, he dissected head shields enlarged thirty to fifty times under a binocular microscope. He worked with fine needles on specimens immersed in alcohol or Canada balsam, for these liquids enhanced the contrast between bone and matrix. Each specimen required up to two months of work, but Stensiö managed to remove bone and leave the matrix behind as a perfect cast of soft anatomy. Second, Stensiö ground serial sections at intervals of one-fifteenth of a millimeter through the head. By lining up this long series of parallel cuts, and tracing the pathways of matrix and bone, Stensiö could reconstruct the soft parts. He then made wax models of this internal anatomy. By coordinating these methods, Stensiö was able to trace all the cranial nerves, identify all major arteries and veins, and provide a detailed reconstruction of the brain.
Stensiö’s detailed reconstruction of the soft anatomy of the brain and nervous system in Cephalaspis. This figure appeared in his 1927 monograph. Courtesy of Department of Library Services, American Museum of Natural History.
(If I may be excused one short tangent on the subject of narrow-mindedness in science, we taxonomists and comparative morphologists are often derided as second-class citizens, not quite true scientists, by colleagues who work with more familiar accouterments of the scientific method—numbers and experiments. A study without formulae or controls seems to lack the necessary rigor of the stereotype. Let those mired in such myopia try to duplicate the work of Erik Andersson Stensiö. Let them spend months with fine needles, separating matrix from bone, grain by grain. Let them try their hand at serial sections, not through the usual wax and tissue, but through bone and stone. And let them try to interpret the resulting mosaic of holes and connections. Stensiö’s work is the most elegant, the most beautiful example ever produced of care and rigor in another dimension. And his results are as firm as anything rooted in numbers and experiment. He was not right about everything; who can be? He misidentified as electric organs, for example, two areas that probably form part of the sensory system for responses to pressure. But the intricate details of his reconstructions for parts of the brain and cranial nerves have been upheld again and again in repeated studies.)
Stensiö’s work proved that Patten had been entirely wrong. Cephalaspis was all fish, and included not a whiff of arthropod. Moreover, the cranial anatomy of Cephalaspis showed detailed similarity, part after part, with the living lampreys—jawless fishes beyond a doubt. After 380 pages of text, Stensiö wrote as his last paragraph and final conclusion: “It is clear now that the Ostracodermi, though very lowly organized, are true craniate vertebrates which have nothing whatever to do either with the Arthropoda or with the Annelida.”
In establishing the position of ostracoderms, Stensiö had also resolved the order of early fishes. Lampreys and hagfishes had no jaws, and embryological evidence indicated the homology of gill-arch bones with later jaws. But before Stensiö’s treatise, ichthyologists didn’t know whether the lampreys and hags were curiously degenerate lines of jawed fishes or remnants of a primordial jawless group. By proving that the ostracoderms were a genealogically coherent group of jawless fishes, predating all jawed vertebrates by tens of millions of years—and by demonstrating the anatomical relationship between lampreys and Cephalaspis—Stensiö both established the pathway of early vertebrate evolution and proved that two lineages of the primordial group had survived.
What then can our realists and relativists make of this tale? The relativist correctly identifies three sequential and mutually incompatible worldviews behind the history of change—Agassiz’s creationism, Patten’s linear progressivism, and Stensiö’s branching tree. Yes indeed, each man read Cephalaspis in the light of his worldview. Yes again, Cephalaspis did not fashion the worldview, but found its inevitable slot in a preconceived structure. Yes once more, the worldviews were products of surrounding culture and personal psychology: Agassiz’s accident of birth in a pre-Darwinian world; Patten’s need for moral answers in nature.
But this history is not only a tale of social fashion—a story of varying dress lengths, tie widths, or degrees of abstraction in painting. Each worldview was a cultural product, but evolution is true and separate creation is not. Cephalaspis may have been buffeted from one social construction to another, but paleontologists also learned important facts about its anatomy at each step—and this accretion of genuine information about the external world must be identified as scientific progress. Agassiz proved that Cephalaspis was a fish, but knew nothing of its internal structure. Patten had resolved enough anatomy to know that ostracoderms were a coherent group of primordial fishes, not a hodgepodge of unplaceable oddballs. Stensiö mapped the brain, the cranial nerves, the blood vessels—while Agassiz could not even find the mouth.
Worldviews are social constructions, and they channel the search for facts. But facts are found and knowledge progresses, however fitfully. Fact and theory are intertwined, and all great scientists understand the interaction.
The debate of realists and relativists, when expressed as ends of a dichotomy vying for victory, is silly and tendentious. Science progresses by establishing facts about the world out there—and science is, and must be, socially embedded. The history of interaction between paleontologists and Cephalaspis is both a pageant of ideas and a growing compendium of information. I exult in the ideas, but I confess to a special love for the rock-hard primacy of Stensiö’s dissections and for Patten’s beautiful eurypterids in the corner of my office.
Isaac Newton mused on the interaction of fact and theory in his most famous passage:
I do not know what I may appear to the world; but to myself I seem to have been only like a boy playing on the seashore, and diverting myself in now and then finding a smoother pebble or a prettier shell than ordinary, whilst the great ocean of truth lay all undiscovered before me.
We would love to fathom that distant ocean, but it is no shabby thing to fondle those pretty pebbles on the shore.
30 | A Tale of Three Pictures
GOETHE, who coined the word morphology and therefore ought to know, once proclaimed that “we should talk less and draw more. I personally would like to renounce speech altogether and, like organic nature, communicate everything I have to say in sketches.” As a card-carrying member of the guild of essayists, I should resist this heresy tooth and nail. I might also argue that the world is a better place because Goethe did not take his own advice. We would all be a little poorer without Faust, while Goethe’s sketches, although no disgrace, are not depriving us of quintessential insight by their general oblivion.
Primates are visual animals. No other group of mammals relies so strongly on sight. Our attraction to images as a source of understanding is both primal and pervasive. Writing, with its linear sequencing of ideas, is a historical afterthought in the history of human cognition.
Yet traditional scholarship has lost this root to our past. Most research is reported by text alone, particularly in the humanities and social sciences. Pictures, if included at all, are poorly reproduced, gathered in a center section divorced from relevant text, and treated as little more than decoration. (Natural scientists, although not noted for insights
about communication, have better intuitions on this subject. Most scientific papers are illustrated, and slide projectors are automatically provided for scientific talks throughout the world. By contrast, I have, three or four times, suffered the acute embarrassment of arriving before a large audience in the humanities or social sciences, slides in hand, to deliver a talk that would be utterly senseless without pictures: no slide projector, no screen, not even a way to darken the room. My fault: I had forgotten to request the projector because, in my own scientific culture, slides are as automatic as words. And so, all you budding scientists who may read these essays, if I have taught you nothing in twenty years of these monthly efforts, at least remember this and thank me some day for a small boon of advice: If you are ever asked to talk before a department in the humanities, remember that you have to request the slide projector. Call this Gould’s law and let it be my immortality—long after everyone has forgotten those upside-down flamingos and pandas’ thumbs.)
Pictures are not peripheral or decorative; iconography offers precious insight into modes of thinking that words often mask or ignore—precisely because we tailor our words so carefully but reveal our secrets unconsciously in those “mere” illustrations. (My thanks to M. J. S. Rudwick, great historian of geology, who first taught me this lesson and who supplied the initial quote from Goethe.)
Pictures are revealing enough when they simply claim to represent an object “as it is.” Shading, emphasis, context, and surroundings all provide an artistic leeway for expressing (often unconsciously) a social or ideological framework. Have you ever seen a dodo pictured as anything other than alone and forlorn, although they once abounded on Mauritius? The classic dodo reconstruction shows a single bird dominating the foreground of a desolate terrain. For the dodo is both a large flightless pigeon and our conventional metaphor for extinction.
Iconography becomes even more revealing when processes or concepts, rather than objects, must be depicted—for the constraint of a definite “thing” cedes directly to the imagination. How can we draw “evolution” or “social organization,” not to mention the more mundane “digestion” or “self interest,” without portraying more of a mental structure than a physical reality? If we wish to trace the history of ideas, iconography becomes a candid camera trained upon the scholar’s mind.
This essay is a tale of three pictures. It tries to illustrate something crucial in the history of evolutionary thought by analyzing three sequential snapshots of “relationships among animals.” These pictures present two favorable features that may promote their expansion from anecdote to illumination. First, all three pictures tell the changing story of the same animal, thereby imparting coherence to a sequence that would otherwise have no anchor. Second, the pictures embody, in a visual epitome that I (at least) found stunning, what may be the most important general issue in our struggle to understand the distinctive character and history of scientific thought.
These pictures were all presented as simple sketches of “objective” relationships among animals; they are also (and primarily, I would argue) iconographies of three strikingly different and incommensurable worldviews. They were drawn by the three men discussed in the previous essay: by Louis Agassiz in the 1830s, by William Patten early in our century, and by Erik Andersson Stensiö in 1927. They all include, as a prominent feature, an attempt to fix the biological position of Cephalaspis, prototype of the jawless fishes that gave rise to all later vertebrates, ourselves included of course.
To summarize briefly the sequence of opinions about Cephalaspis (see previous essay for the details), Agassiz discovered that Cephalaspis was a fish, not a trilobite (the bony head shield looks like the external armor of an arthropod, but Agassiz found heads attached to indubitably fishy bodies). Agassiz denied evolution altogether, but Patten tried to interpret Cephalaspis as an intermediary form between arthropods and vertebrates, a key waystation on “the Great Highway of Animal Evolution.” Stensiö proved, by meticulous dissection of exceptionally preserved fossils from Spitsbergen, that Cephalaspis was “all fish,” without a whiff of arthropod. He also demonstrated that modern lampreys and hagfishes are close relatives of the great primordial group of jawless fishes represented by Cephalaspis (class Agnatha, meaning, quite appropriately, “jawless”).
The previous essay also set the story of Cephalaspis in the context of an old debate about progress in the history of scientific thought. I argued that this subject is often obscured by a false dichotomy drawn between equally untenable extremes: realists, who argue that science, with its timeless and universal methods, learns progressively more and more about an objective external reality; and relativists, who hold that the history of theories approximates the vagaries of fashion, a series of equally workable solutions altered by whim or social circumstances.
I think that each side of this controversy possesses a central insight, and that their marriage provides a workable solution sensitive to the fundamental concept of each camp. Notwithstanding a long history of arguments, ranging from the playful to the tendentious to the sophistic, there is a world out there full of stars, amoebas, and quartz crystals. (We must, in any case, behave as if this claim were true in order to negotiate life’s numerous difficulties with any success—and this behavior has brought consistent results, at least in the form of technological achievement.) Science does construct better and better maps of this outer reality, so we must assume that change in the history of scientific theories often records more adequate knowledge of the external world, and may therefore be called progress.
On the other hand, we must also admit that the history of scientific theories on any subject is no simple tale of good information driving out bad. Successive theories often display the interesting property of incommensurability. They do not speak the same language; they do not parse the world into the same categories; they embody fundamentally different views about the nature of causality. The new is not simply more and better information heaped upon the explanatory structure of the old. In this sense, the history of theories is a successive replacement of mutually incompatible worldviews, not a stroll up the pathway of objective knowledge.
The three pictures of relationships among early vertebrates demonstrate, with bold literality, this principle of scientific change as a series of incommensurable worldviews, each replacing rather than just building upon the last. Yet the sequence also records increasing objective knowledge about Cephalaspis; it is not a passive mirror of social change.
Louis Agassiz (1807–73), the great Swiss zoologist who became America’s premier naturalist, was the last great scientific creationist (I am writing this essay in the museum and laboratory that he opened in 1859). He built his career upon two fundamental achievements: the development of the theory of ice ages, and a monumental work on the classification and relationships of all fossil fishes. Agassiz summarized his fifteen-year project on fossil fishes with the first major example of an iconography that paleontologists have since adopted as canonical—the so-called spindle diagram (see figure). In these geological charts of relationships among organisms, the vertical axis represents time, as though the diagram portrayed a sequence of strata in the field. Each group of organisms is drawn as a spindle, with varying widths through time representing a history of fluctuating diversity, and the ends of the spindle marking origin and extinction. The ordering of spindles records degree of relationship, with physical closeness representing biological affinity. We have all seen so many of these diagrams that we read them automatically, rarely stopping to acknowledge that all these features are iconographic conventions, not necessary realities.
Agassiz’s creationist version of the evolution of fishes. Note Cephalaspis at the base of the ganoid radiation. From Louis Agassiz’s Les poissons fossiles, Vol. 1, p. 170, 1833.
These conventions leave great latitude for portraying a theoretical worldview in the guise of objective knowledge—and Agassiz’s famous chart is a striking example of concept as iconography. Note two features of Agassi
z’s fishes. First, of the various geometries that might be used to portray relationships among organisms—circles, chains, ladders, parallel lines like teeth on a comb—he chooses a topology of branching from a central stem in each of his four groups. This iconography embodies his biological theory of life’s history as a tale of differentiation through time from simple and highly generalized archetypes. Life diversifies on an embryological model. Just as all mammalian fetuses begin with a simple and similar form and differentiate later to bat, whale, and camel, so too does the geological history of a group generate more diverse and specialized creatures through time.
This view sounds so evolutionary that we wonder why Agassiz continued his lone holdout against Darwin to the death. But such a feeling only represents the chauvinism of later knowledge imposed upon a fundamentally different worldview. Differentiation from a common archetype need not imply a physical, evolutionary connection among successive forms. Suppose that differentiation is God’s grand design for all developmental processes in nature. Embryology proceeds in physical continuity, but geological succession may feature a series of independently fashioned forms, linked together as incarnations of an ordered pattern of thought in their creator’s mind.
Agassiz depicted his creationist interpretation in the second striking feature of his iconography. The separate spindles in each of his four groups may converge lovingly towards each other, and towards the central or archetypal line, but they never join! And Agassiz knew exactly what he was doing, and why: