The Great Fossil Enigma

by Simon J. Knell

  The first cracks appeared in 1954, when Rhodes answered Branson and Mehl's test. It will be recalled that these men had challenged Scott to find isolated conodont fossils in the same proportions as seen in natural assemblages. Branson and Mehl must have known this was a mischievous test as the assemblages were composed of both robust and delicate elements, making it almost impossible to imagine nature preserving their natural ratios so perfectly. Yet against all the odds, this is precisely what these durable and abundant little fossils revealed. It enabled Scott, Du Bois, and Rhodes to emphatically reaffirm the truth. But in 1954, there was no longer any need to answer this test, as Rhodes's arguments had already won the day and, anyway, Branson was dead.

  Now Rhodes realized that the test held other potential, for the search for ratios had forced him to think of fossil elements not in terms of species but as components. The animal was, to this way of looking, like an Airfix kit composed of the anatomical equivalents of wings, rudder, and fuselage. One only need find and recognize these components and construction could begin of assemblages that had never been found in any coherent form. So Rhodes divided the fossils into a handful of types, easily understood in the everyday language of bars, blades, platforms, and so on. Seeing these as different kinds of component, he looked again at the assemblages he had found and named, and discovered that each contained the same component parts. This suggested that the basic plan conformed to Scott's Lochriea, which Rhodes had rearranged so that it looked broadly similar to Schmidt's reconstruction. The only assemblage that did not adhere to this plan was Duboisella, which Rhodes used as the basis for another standard pattern. These now became the equivalents of box-lid illustrations, useful for guiding the construction of previously unseen assemblages from their components. By tracing the history of each component type so as to discover which coexisted, Rhodes could also say, with some certainty, that assemblages similar to Lochriea could be traced from the Silurian onward and those similar to Scottognathus – which shared the same broad body plan as Lochriea – from the Upper Devonian to the Lower Mississippian. The Duboisella type, in contrast, had existed from the Silurian to the Permian.1

  Rhodes knew, however, that this picture was incomplete, that there were other kinds of assemblage for which there was still insufficient data to begin to reconstruct them. No complete assemblages had been found in the Devonian, for example, and Müller was, in 1956, struggling to imagine what they might look like. His collections suggested that the preponderance of elongated blades seen in Carboniferous assemblages had been preceded by a prevalence of platforms in the Devonian animals. But puzzlingly, some rocks from the Middle Devonian produced Icriodus platforms and nothing else.2

  Hermann Schmidt was, in that year, in that same quarry that had first furnished him with assemblages. Here, with the help of three of his students, he spent eight days searching for yet more. In what Müller recalled as a difficult collaboration, he helped Schmidt to interpret what had been found.3 It was now that Müller realized that his problems imagining Devonian assemblages resulted from the incomplete survival of the different kinds of element. He warned others to beware – building assemblages, in the way Rhodes had began to do, held grave risks.

  Schmidt and Müller's paper did little to progress the study of assemblages, not least because these were types already well known. But the work also held other difficulties, for while the men could agree on the basic facts, they could not agree on what they meant. As a result, the paper, like a film with a choice of endings, supported two contrasting interpretations. But Müller did not mind too much; he had already decided that assemblages were relatively unimportant. Perhaps the most interesting outcome was a decision to return Schmidt's Westfalicus – a name invented to satisfy conventions being introduced with parataxa – back to its original name, Gnathodus. It will be recalled that this name had been chosen by applying the rules of zoological nomenclature. It was once again the only assemblage to be named in this way and a direct challenge to Moore's proposal for interpretive myopia.4

  Back in the 1950s, Müller had also wondered if assemblages were symmetrical. From the 1920s, it was believed that elements existed in mirror pairs, left and right. Thoughts of the fish made this as an expectation, but Müller believed that Chalmer Cooper had found an unpaired element in an assemblage in the 1940s. It seems probable, however, that Cooper was merely complaining that an assemblage was incomplete. Nevertheless, Müller used this new piece of information to suggest that an unpaired element was missing from Rhodes's Duboisella. Soon, and independently, Lindström, Adolf Voges, and Bergström and Sweet were also reporting unpaired elements. It was easy for this idea to take hold now that Walter Gross had demonstrated that the animal was neither fish nor worm, and that the elements were not teeth. Indeed, the thought of unpaired elements encouraged the ever-imaginative Lindström to wonder if the animals were always bilaterally symmetrical. He felt he had evidence to suggest that sometimes they were not. He consulted Carl Rexroad and Sam Ellison, who concurred; they too had platform elements that did not consist of mirror pairs. Lindström was then trying to imagine the animal for his book and speculated, “Some might have floated passively, perhaps even in colonies.”5 He was, however, alone in having any thoughts of the animal.

  While Lindström pondered the architecture of the assemblage and what it meant for the biology of the animal, others were trying to see assemblages in collections of discrete fossils. At Marburg, Reinhold Huckriede thought he could see the appearance and disappearance of whole groups of conodont fossils in his relatively sparse Triassic collections. He pulled these associations together, calling them “Satze” (sets) guided by the assemblages Schmidt and Rhodes had described. Similarly, in his 1964 paper on the Silurian of the Carnic Alps, Otto Walliser produced nine “apparatuses,” giving each an identifying letter, A to J, but no name. He knew that to name them according to rules of zoological nomenclature, as Eichenberg and Schmidt had done, would be to isolate his work. He also knew that the resulting names would reflect a random and curious history of discovery rather than the zoological logic of the apparatuses. “It would be difficult to convince colleagues,” he recalled. “I didn't dare to do this.”6 To advance science, Walliser realized, it was necessary to play ball, even if that meant ditching a few scientific ideals. For the moment his apparatuses remained convenient and practical associations.

  Walt Sweet and Stig Bergström also hinted in 1962 that a more natural taxonomy was desirable. Sweet had long been cultivating an interest in conodonts in his students. It was, however, with the arrival of some samples from Arthur Cooper at the U.S. National Museum that conodont science at Ohio State University took a new turn. Cooper had been using acids to extract fossil brachiopods from a thin Middle Ordovician limestone from near Pratt Ferry in Alabama and he sent the residues left behind to Sweet and Bergström for them to pick over in search of conodonts.7 As the two men examined the fossils, they detected a similar number of “right” and “left” elements and noticed that different kinds of element occurred in approximately the same numbers. With a large number of conodonts fossils at their disposal, these facts suggested that it would be possible to unite the elements in “natural species,” but they stopped short of doing so. Instead they held onto the hope that an Ordovician assemblage that would “ultimately indicate which of several possible combinations existed in fact” would be discovered. They had no thoughts of rocking the boat. Now converts of Cooper's “acidizing,” Bergström started to digest great volumes of rock.

  While these discoveries were being made, Lindström was quietly writing his book, but as he did so he became increasingly of the view that the science could only ignore the animal at its peril. Only a few years before, he had been a strong advocate of the utilitarian approach, but as he looked at his fossils he noticed that some could be arranged into gradational series, each element changing slightly as its lines of symmetry shifted and parts of the element were extended. These were fossils of precisely the same age; he
was not looking at evolutionary change over time. Lindström called these “symmetry transition series” and found that they only affected certain element types. They were not found among the platforms, for example. He recognized that these transition series reflected the positioning of each element in the assemblage and suggested that unpaired symmetrical elements might once have occupied the midline of the animal.8 Lindström's arguments were deep and complex, and a little hard to follow, but they provided a vital key for the reconstruction of apparatuses. They gave the apparatus an anatomical logic but made a mockery of naming individual elements as if they were species. When he told Ziegler this, Ziegler responded, “Yes, so what?” They had been firm friends since 1962, and Ziegler's unwillingness to make any concessions to the biological truths of the animal only encouraged Lindström to take the opposite view.

  The problem remained that only from the Carboniferous was there good evidence of the nature of assemblages. Then, in 1964, Carl Rexroad and Robert Nicoll of the Indiana Geological Survey published a paper describing two pairs of elements that had been fused together (figure 8.1). Each of these preserved a different element pairing and seemed to reflect the relative positions of the elements in life. These came from a drill core through the Silurian Kokomo Limestone near Logansport in the north of the state. As no natural assemblages had ever been found in rocks this old, the two men thought these new finds offered important clues to the animal's anatomy. The “fused clusters,” as they called them, had survived bulk processing because they were bound together with exactly the same material that made up the elements themselves. Rexroad and Nicoll had never seen anything like them, and they consulted Lindström, Collinson, and Klapper. Only Klapper recalled seeing something similar. Rexroad and Nicoll concluded that the animals must have befallen a “pathologic mishap” and had perhaps suffered a tetanus infection.9

  By the end of 1964, then, a number of authors were, in their own quiet ways, exerting pressure on the mountain Ziegler was scaling. They were producing small cracks and voids. Certainly, we can say that a number of conodont workers were beginning to see their fossils through the lens of the assemblage, and in complementary studies they were acting like catalysts one upon the other. Nevertheless, this remained simply a way of looking; it had not yet become a political movement for change. In that year, no one sensed the coming revolution, even if the germ of that revolution was already in the minds of Bergström and Sweet.

  8.1. Rexroad and Nicoll's fused conodont clusters. In one pair the left element overlaps the right; in the other the reverse is true. Reproduced with permission from C. B. Rexroad and R. S. Nicoll, Journal of Paleontology 26 (1964). SEPM (Society for Sedimentary Geology).

  In the summer of 1961, Bergström and Sweet had made a “grand traverse of the Ordovician exposures of the eastern Midcontinent.” This gave them a panoramic outlook that would affect their work together over the coming decade. With Sweet's other students, Bergström had been gathering and processing hundreds of samples from the Middle Ordovician Lexington Limestone in Kentucky, Ohio, and Indiana. This rock was extraordinarily rich in conodonts: “There were thousands and thousands and thousands of them.” Sweet insisted that Bergström tabulate the numbers and occurrences of those he found, sample-by-sample. This work began but was interrupted, in the fall of 1961, when Bergström had to return to Lund in the hope of securing an academic appointment there. Nevertheless, the two men continued to work on their report by mail. Then Sweet attended a lecture on pollen: “Sometime in 1964, I was struck by the similarity of our work to that of Aureal Cross, a palynologist from Michigan State and I went back to my office after this lecture and began to plot the frequencies of the most common Lexington conodonts on Cross-like relative abundance graphs. The results convinced me beyond the shadow of a doubt that 4 of the element types Stig and I (and my other students) had been treating as representatives of 4 separate species, were, in fact, components of a single ‘apparatus’ which should be regarded as that of just one conodont species.”10

  The paper was then already close to completion, but based on single element species. Bergström was busy in Lund preparing the illustrations. However, Sweet was now convinced they could and should do the job properly, that they should locate and name the natural species they had found. He wrote to Bergström, telling him that if they continued along their present course the paper would be out of date from the day it was published. When Bergström returned to Columbus from May to September 1964, they undertook a complete revision of the paper, piecing together natural species on the basis of size, color, secondary structures (like denticles), ornament, geographic and stratigraphic range, ratios of components, and so on. As no Ordovician assemblages had been found, they possessed no architectural plan and had to rely instead on the certainties that come with huge collections. In this regard, at least, they were extraordinarily fortunate, for they possessed about a quarter of a million conodont elements! They became connoisseurs: “We found that it was often possible to predict with uncanny accuracy the ultimate composition of a collection after just the first few specimens had been sorted from the residue.”11 The result of this colossal effort was the detection of what they considered to be twenty-three truly biological species. Twenty of these were composed of different kinds of elements and three of only one kind. Far from undermining the simple evolutionary model so effectively exploited in Germany, Bergström and Sweet found these associations gave new evolutionary insights. Now long-ranging elements became inextricably associated with short-lived and rapidly evolving ones. The critical step Bergström and Sweet took, however, was not merely to detect what they thought were natural species but to actually name them. No one had dared do this since Schmidt: “So we messed around with it. We didn't get it all right and there were some people who thought we didn't get any of it right!”12 The paper was completed in 1965 while Sweet was visiting professor at Lund. It would be published the following year.

  Sweet, Sweet's doctoral student, Tom Schopf, and Minneapolis student Gerald Webers gave the first indications of this approaching storm at the annual meeting of the Geological Society of America in 1965. All three had papers in press that in some degree promoted this new way. Between them, they had amassed three hundred thousand conodont fossils – a body of evidence vastly superior to any previous study. Inspired by the statistical predictions of Scott, Du Bois, and Müller, Webers had began his research in 1959 and had, without the slightest hesitation, adopted the rules on zoological nomenclature and used the earliest established element species to name his natural species.13 He possessed the smallest collection – thirty-five thousand fossils – but this was nevertheless a huge amount of data. He wasted little time reviewing the failed attempts of others to find alternate solutions and felt sure that this new way would find support and generate fewer problems than had been predicted. Webers and Sweet had met on occasion but Webers's project was totally independent. Schopf's project, in contrast, had been suggested to him by Sweet. He possessed some fifty-five thousand specimens in 1962, but while he was happy to discuss recurrent groups, he stopped short of naming natural species.14

  When Bergström and Sweet's Lexington paper appeared in 1966, it was, for Ziegler and others, as if a 100-megaton bomb had been dropped on the conodont community. The paper itself was about the stratigraphy of the Lexington Limestone; the new taxonomy was merely a necessary underpinning. To some degree Bergström and Sweet attempted to protect themselves by downplaying this innovative aspect: “The taxonomic philosophy involved is not new, nor is it novel in the interpretation of collections of discrete conodont elements.”15 Nevertheless, “we understood this might be Revolutionary,” Sweet later remarked.16 It shook the science out of its comfort zone and demanded that it think of the conodont as a biological entity. If this in itself was not challenge enough, it also demanded that the science learn a new language and transfer names that once spoke of individual elements to associations that were then to be understood as natural species. As we shall
see in the next chapter, this paper was, in all respects, conceptually grand and, in so many ways, startlingly original. It was one of the masterpieces of conodont science.

  8.2. Revolutionaries Walt Sweet (left) and, in a more recent picture, Stig Bergström. Photos: Dick Aldridge and Jeff Over, respectively.

  Bergström, Sweet, and Webers now found themselves “plunged headlong into the jungle so neatly avoided by Huckriede, Walliser, and Lindström.”17 As Bergström and Sweet noted, “The revised taxonomy was not greeted with enthusiasm, to say the least!” But Sweet could not have been surprised. He never seemed to shrink from what he considered the proper scientific course. Indeed, the two men continued their rebellion in the small print, too, pointing out that since the conodont fossils were no longer considered the teeth or jaws of vertebrates, terms like “tooth, jaw, oral, aboral, pulp cavity, fang, ramus” should also be ditched.18 They now reintroduced the concept of “form species” to describe individual elements; the only proper species names were those attached to element associations.