I Have Landed

by Stephen Jay Gould

  Linnaeus’s taxonomic scheme designates a rigorously nested hierarchy of groups (starting with species as the smallest unit) embedded within successively larger groups (species within genera within families within orders, etc.). Such a nested hierarchy implies the organizing geometry of a single branching tree, with a common trunk that then ramifies into ever-finer divisions of boughs, limbs, branches, and twigs. This treelike form also happens to express the hypothesis that interrelationships among organisms record a genealogical hierarchy built by evolutionary branching. Linnaeus’s system thus embodies, quite apart from Linnaeus’s own intentions or theoretical connections, the causality of Darwin’s world.

  This correspondence between the Linnaean hierarchy and life’s evolutionary tree achieves its clearest expression in pictorial form. The accompanying drawings show a Linnaean ordering of box within box, as well as the alternate expression of this logical order in a branching diagram of sequential genealogical splitting—in this case, a successive carving of the kingdom of all animals into, first, chordates contrasted with all other animals; then vertebrates contrasted with all invertebrates; mammals contrasted with all other vertebrates; the order Carnivora contrasted with all other mammals; the family Canidae contrasted with all other carnivores; the genus Canis contrasted with all other doglike carnivores; and domestic dogs contrasted with all other members of the genus Canis. I stated that binomial nomenclature expresses the first step of this hierarchical ordering—and thus presents a microcosm of the entire scheme—because the two parts of a species’s name record the first act of embedding smaller units within more inclusive groups of relatives. The name Canis familiaris states that the smallest unit, the dog species, ranks as one member of the next most inclusive group, the genus Canis, which unites all other species (e.g., the wolf Canis lupus and the coyote Canis latrans) that originated from a common ancestor shared by no other species in any other group.

  Linnaeus thought that his chosen scheme of mapping biological relationships as smaller boxes within successively larger boxes, until all units nested within the most inclusive box of life itself, represented the best human device for expressing the eternal order that God had chosen when he populated the universe. I doubt that he ever explicitly said to himself (for I suspect that his mental mansion included no room for such a thought): “But if, quite to the contrary, life evolved by a process of ever-ramified branching from a single ancestor over a long period of time, then the hierarchical order of the binomial system will capture the topology of organic relationships just as well, because the logic of my system translates pictorially into a tree with a single trunk at the base, and subsequent division into branches that never coalesce thereafter. And such a topology might represent either God’s permanent order, preconceived from the first, or the happenstances of historical change and development on an evolutionary tree growing from a single starting point under the constraint of unbroken continuity (although branches may die and fall from the tree as lineages become extinct), and continuous bifurcation without subsequent joining of lineages.”

  Linnaean taxonomy in another of its geometric portrayals as a branching system with all divisions beginning from a common trunk and no fusion of separated branches allowed.

  I emphasize this property of irrevocable branching without subsequent amalgamation because the Linnaean logic of placing small boxes into larger boxes—which just happens to conform to the historical reality of Darwin’s system—establishes such a map of organic relationships as its primary and inevitable consequence. One can’t, after all, cram big boxes into smaller boxes. Therefore, for example, two species in the same genus can’t reside in different families, and two orders in the same class can’t be placed in different phyla. If lions and tigers rank as two species in the same genus (Panthera), they cannot then be allocated to different families of higher rank (lions to the Felidae and tigers to the Canidae, for example)—for the two larger family boxes would then have to fit within the smaller box of the genus Panthera, and both the rules of Linnaean logic, and the requirements of Darwinian evolutionary history, would then be fractured. I can only be a monkey’s uncle or a horse’s ass in a metaphorical sense—for my species fits into the small box of the genus Homo, which must nest within the larger box of the family Hominidae; and one member of my species can’t opt out of our box to join the Cercopithecidae or the Equidae, thus splitting a coherent lower group into two higher groups, and violating both Linnaean logic and Darwinian reality.

  So perhaps Linnaeus enjoyed a little bit of luck in choosing the one logic for a creationist system that would also fit without fuss into a new universe of historical evolution by branching. At least Linnaeus demonstrated exemplary survival skills in passing the test of time as taxonomy’s father. But I hesitate to ascribe his remarkable success to pure dumb fortune, and for a primary reason rooted in the key contention of this essay: that taxonomies transcend simple description and always embody particular theories about the causes of order, thus melding preferences of mind with perceptions of nature.

  I think that Linnaeus succeeded because, however unconsciously or preconsciously, he made some excellent decisions about both the mental and perceptual aspects of taxonomic systems. On the perceptual side, he must have seen better than any of his colleagues that under the logic of hierarchy and branching, organisms could be arranged into a consistent order that might win general assent without constant bickering among practitioners. Other contemporaries had proposed very different logics for classification, but had never found a way to push them through to an unambiguous and consistent system. In the most telling example, Linnaeus’s most famous contemporary and archrival, France’s celebrated naturalist the Baron Georges Leclerc Buffon (1707–1788), had struggled through more than forty volumes of his Histoire naturelle— in my judgment, the greatest encyclopedia of natural science ever written—to develop, without conspicuous success, a nonhierarchical system that joined each species to some others by physiology, to a different group by anatomy, and to a still different set by ecology.

  But I would, in addition, like to advance the unfamiliar argument that Linnaeus also succeeded because he made a very clever, and probably conscious, choice from the mental side of taxonomic requirements as well. In deciding to erect a hierarchical order based on continuous branching with no subsequent joining of branches, Linnaeus constructed his system according to the most familiar organizing device of Western logic since Aristotle (and, arguably, an expression of our innate and universal mental preferences as well): successive (and exceptionless) dichotomous branching as a system for making ever finer distinctions. In a logical tree of this form—often called a dichotomous key—one may move in either direction to place a particular basic object into ever larger groups by joining successive pairs, or to break down a large category into all component parts by successive twofold division.

  One may, for example, interpret the diagram that I presented earlier as a dichotomous key. We can reach dogs by starting with the largest category of all animals, dividing this totality into vertebrates and invertebrates, then splitting the vertebrates into mammals and non-mammals, the mammals into carnivores and non-carnivores, the carnivores into canids and non-canids, and finally the canids into dogs versus others. (We can also work outward in the opposite direction to learn how dogs fit into the hierarchy of all animals.)

  In fact, the idea for this essay came to me when I recently purchased an obscure late-sixteenth-century book on Aristotelian logic and noted that its numerous charts for working through the categories of reasoning, and the attributes of human form and behavior, had all been constructed as dichotomous maps bearing an uncanny resemblance to the taxonomic keying devices that I have seen and used in texts and guidebooks for naturalists throughout my career. Thus, Linnaeus gave himself quite a leg up by building his taxonomic system upon a familiar form of logic that scholars had applied to all subjects, scientific and otherwise, from the dawn of Western history—a style of reasoning, moreover, that
may track the basic operation of our brains.

  I took the accompanying chart from this 1586 treatise, published in Paris by the physician Nicolas Abraham, and entitled hogogethica ad rationis normam delineata (an introduction to ethics as delineated by the rule of reason). Abraham first divides the domain of ethical decisions into the dichotomous pair of mentis (by the mind) and moris (by custom). He then splits the lower domain of custom into the two categories of privatis (above) and publicis (below). Interestingly, and I suspect consciously, authors of dichotomous keys also seem—at least in my limited study of such devices from pre-Darwinian times—to order their pairings from the good and most valued on top to the least admirable on the bottom. In this case, reason beats custom, while, within custom, private decisions (presumably made for reasons of personal belief) trump public actions (that may be enforced by social pressure). I am particularly fond of the dichotomous key for birds of prey that the great English naturalist John Ray published in 1678—with a first division of day-flyers on top and night-flyers on the bottom; a second division of the preferred day-flyers into bigger species (above) and smaller (below); and a third division of the big species into “more generous” eagles above and “more cowardly and sluggish” vultures below.

  A dichotomous key, as presented by Nicolas Abraham in 1586 to classify ethical decisions.

  But let us follow Abraham’s key for the higher category of mental decisions, which then undergoes a further dichotomous split into sapientia (done by wisdom) above and prudentia (done for reasons of prudence) below. The third and final set of twofold divisions then separates judgments by sapientia into intelligent (achieved by pure reason) above in preference to scientia (achieved by knowledge about material things) below. The lower judgments of prudentia then divide into a preferred category of bona consultatio above (derived from our seeking a good advice from others) and the less worthy dichotomous alternative of sagacitas below (determined only by our own judgment).

  I do admire Linnaeus as an intellectually driven and brilliantly complicated, but arrogantly vainglorious, man. If I preferred the hagiographical mode of writing essays, I would stop here with a closing word of praise for Linnaeus’s perspicacity in harnessing both the observational and theoretical sides of his mental skills to construct a flexible and enduring taxonomic system that could survive intact in sailing right through the greatest theoretical transformation in the history of biology.

  But he who lives by the sword dies by the sword (as Jesus did not exactly say in a common misquotation that remains potent in truth and meaning despite a slight inaccuracy in citation). Linnaeus’s consistency and wisdom in developing and defending the binomial system of hierarchical classification carried him through to intellectual victory. But, like so many originators of grand and innovative systems, he reached too far (whether by arrogance or overexcitement) and became too committed to his procedure as the one true way for classifying any collection of related objects. (I cannot help recalling my experience with a customs official on a small West Indian island who classified my land snails as turtles because his forms only permitted a distinction between warm-blooded and cold-blooded “animals”—and the word “animal,” in his personal understanding, only designated vertebrates. Thus, snails became turtles because both are cold-blooded and move with legendary torpor.)

  Once Linnaeus had fully developed the binomial system and its supporting logic of a consistently nested hierarchy, he supposed that he had discovered the proper way to classify any group of natural objects, and he therefore began to apply binomial nomenclature to several classes of inappropriate phenomena, including rocks and even human diseases. Clearly, he had become overen-amored with his own device, and had lost sight of the key principle that hierarchical embedding by dichotomous branching only captures the causal order within certain kinds of systems—particularly those that develop historically by successive branching in unbroken genealogical continuity (with no later amalgamation of branches) from a common ancestor. Since Linnaeus tried to apply his binomial system to several groups of objects that, by their own rules of order or development, patently violate the required hierarchical logic, perhaps he never really did grasp the limitations (and therefore the essence) of his binomial system. So maybe Linnaeus did prevail partly by the luck of organic conformity to his general logic, rather than by his correct and conscious reasoning about distinctive causes of relationships among plants and animals.

  For example, the accompanying page from the seventh (1748) edition of Systema Naturae designates binomial species of the genus Quartzum from the classification of rocks that he presented as a third chapter, following his taxonomies for animals and plants. The first “species,” Quartzum aqueum (transparent quartz) includes ordinary glasslike quartz; the second, Quartzum album (white quartz) encompasses less valued, opaquely water-worn quartz pebbles; the third, Quartzum tinctum (colored quartz) gathers together the colored varieties that mimic more-valuable gemstones (Linnaeus calls them false topaz, ruby, and sapphire, for example); and the fourth, Quartzum opacum (opaque quartz) describes the even less useful and less transparent flintstones.

  But the nature of quartz, and the basis of relationships among minerals in general, defies the required logic of causality for any system legitimately described in Linnaean binomial terms. The members of Quartzum aqueum, for example, do not hang together as a set of mutually closest historical relatives, all physically derived in continuity from a common ancestor that generated no other offspring. Rather, the specimens of this false species look alike because simple rules of chemistry and physics dictate that transparent quartz will form whenever silicon and oxygen ions come together under certain conditions of temperature, pressure, and composition. The members of this “species” can claim no historical or genealogical coherence. One specimen might have originated half a billion years ago from a cooling magma in Africa, and another just fifty years ago in a bomb crater in Nevada. Minerals must be classified according to their own causes of order, a set of rules distinctly different from the evolutionary and genealogical principles that build the interrelationships among organisms.

  A page from the mineralogical chapter of the 1748 edition of Linnaeus’s Sy sterna Naturae, showing that he tried to apply his method of binomial nomenclature to rocks, as well as to organisms.

  Linnaeus clearly overreached in supposing that he had discovered the one true system for all natural objects. In the twelfth and final edition of Systema Naturae (1766), he included a section entitled Imperium Naturae, dedicated to extolling his hierarchical and binomial method as universally valid. God made all things, Linnaeus argues, and he must have used a single and universal method, now discovered by his most obedient (and successful) servant. Linnaeus writes: “Omnes res creatae sunt divinae sapientiae et potentiae testes” (all created things are witnesses of divine knowledge and power). Using a common classical metaphor (the thread of Ariadne that led Perseus out of the labyrinth after he had killed the Minotaur), Linnaeus praises himself as the code cracker of this universal order: “knowledge of nature begins with our understanding of her methods by means of a systematic nomenclature that works like Ariadne’s thread, permitting us to follow nature’s meanders with accuracy and confidence.”

  Ironically, however, Linnaeus had succeeded (in a truly ample, albeit not universal, domain of nature) precisely because he had constructed a logic that correctly followed the causes of order in the organic world, but could not, for the same reasons, be extended to cover inorganic objects not built and interrelated by ties of genealogical continuity and evolutionary transformation. The strength of any great system shines most brightly in the light of limits that give sharp and clear definition to the large domain of its non-universal action! By understanding why the Linnaean system works for organisms and not for rocks, we gain our best insight into the importance of his achievements in specifying the varied nature of disparate causes for nature’s order among her many realms.

  On the same theme of power in excep
tions, and to make a somewhat ironic point in closing, Linnaeus’s hope that he had discovered a fully universal basis (God’s own rules of creation) for classifying all natural objects has recently suffered another fascinating blow—but this time from inside (that is, from the world of organisms). Science had already denied Linnaeus’s universality more than two hundred years ago by accepting his procedures only for historically generated genealogical systems based on a geometry of branching (the evolution of organisms as a primary example), and rejecting his binomial schemes for rocks, diseases, and other systems based upon different theoretical foundations of order. (In fact, Linnaeus did try to establish a binomial classification of human diseases as well—in one of his least successful treatises, Genera morborum, published in 1736 and immediately forgotten!) But now, one of the most important biological discoveries of our age has also challenged the universal application of Linnaean taxonomy in the world of organisms as well.

  We need not fret for fat, furry, multicellular creatures—the plants, fungi, and animals of our three great multicellular and macroscopic kingdoms of life. For evolution, in this visible world of complex creatures, does follow the Linnaean topology nearly all the time. That is, the basic structural rule that validates the binomial system works quite satisfactorily at this level—for branches never join once they have separated, and each species therefore becomes a permanently independent lineage, forging no further combinations with others after its origin. Evolution cannot make a nifty new species of mammal by mixing half a dolphin with half a bat to generate an all-purpose flyer and swimmer.

  Until a few years ago, we thought that this rule of permanent separation also applied to the world of unicellular bacteria—the true dominators of earth and rulers of life in my opinion (see my book Full House, 1996). In other words, we assumed that the bacterial foundation of the tree of life grew in a fully Linnaean manner, just like the multicellular section. (Actually, and to emphasize the importance of the discovery described below, the bacterial domain occupies most of life’s tree because the three multicellular kingdoms sprout as three terminal twigs on just one of the tree’s three great limbs, the other two being entirely bacterial.)