The Linnaean system implies a definite geometry for the ordering of objects, and therefore only works when the causes of order produce results conforming with this geometry. Consider two essential properties: First, the Linnaean system is hierarchical. The basic units (species)
are grouped into genera, genera into families, families into orders, and so on. This scheme implies a treelike topology with the largest unit (say, the kingdom Animalia) as a single trunk; middle units as large boughs attached to the trunk (phyla like Arthropoda and Chordata); smaller units as branches emerging from the boughs (classes like Mammalia and Aves joined to the chordate bough);
and, finally, basic units as twigs growing from the branches (species like
Homo sapiens
and
Gorilla gorilla
attached to the primate branch of the mammalian bough of the animal trunk). This topology correctly represents a system of objects produced by branching evolution, with continuous divergence and no suturing together of separately formed branches. Since the history of life operates by this
geometry, the Linnaean system works splendidly for classifying organisms.
Second, basic units must be discrete and definable, not smoothly intergrading and constantly joining. Since organic species are independent and stable units (after the brief geological moment of their branching origin), the Linnaean system also functions ideally for complex, sexually reproducing organisms.
But the same
reasons that allow the Linnaean system to classify fossil organisms so well also guarantee an inapplicability in principle to the two categories of the mineral kingdom that Mendes da Costa’s age also called “fossils”—minerals and rocks. Minerals and their crystals have definite chemical formulae, and aggregate by simple physical rules. Their relative similarities are not genealogical, and their order
cannot therefore be expressed by a treelike geometry. Moreover, mineral “species” are not discrete entities composed of genealogically related individuals in historical continuity. Cambrian quartz, at half a billion years of age, does not differ from Pleistocene quartz separately made in a geological yesterday.
Rocks and soils, composed of mineral grains and the products of their erosion, fail
the Linnaean requirements for an even more fundamental reason. Rocks and soils form a broad continuum of fully intergrading compositions. We cannot identify discrete species of granites, marbles, or chalks. Granites, for example, are composed of quartz, two kinds of feldspar, and a dark mineral like biotite or hornblende—and all these constituents can be mixed and matched like house paint into any
desired composition.
Nonetheless, Mendes da Costa, as a devotee of the classical passion for order, struggled to do as the master Linnaeus had enjoined—to make every object of nature fit the binomial system, thereby bringing all phenomena into one grand style of order. Thus, in
The Natural History of Fossils
, Mendes da Costa presents a Linnaean classification of earths and stones into species,
genera, and other categories now used only for organisms. His great treatise has a wonderfully archaic ring today because he seems to treat objects of the mineral kingdom as a collection of organisms, and sets, as his highest goal for this branch of geology, a grouping of rocks to match a listing of beetles. I am particularly fond of his
Natural History of Fossils
because this treatise, more than
any other work written in English, records a short episode expressing one of the grand false starts in the history of natural science—and nothing can be quite so informative and instructive as a truly juicy mistake.
Consider Mendes da Costa’s classification of earths and stones. He does not use Linnaeus’s own names for categories, but he follows the same basic procedure. Linnaeus’s hierarchy
included four levels (we have since added several more): class, order, genus, species. Mendes da Costa uses six: series, chapter, genus, section, member, and species. At the highest level, he divides his geological objects into two series:
earths
and
stones.
Following the Linnaean principle (and a long history of Western thought traceable to Aristotle), he provides a definition of categories based
on a
fundamentum divisionis
, or basic criterion of difference. Earths are “fossils not inflammable, but divisible and diffusible, tho’ not soluble in water”—whereas stones have the same properties, but are not divisible and diffusible.
He separates the first series,
earths
, into seven genera within three chapters. Chapter 1, defined as “naturally moist, of a firm texture and which have a smoothness
like that of unctuous bodies,” includes three genera—boles (
Bolus
), clays (
Argilla
), and marles (
Marga
). Chapter 2 (“naturally dry or harsh, rough to the touch and of a looser texture”) encompasses two genera—chalks (
Creta
) and ochres (
Ochra
). Finally, Chapter 3 (“naturally and essentially compound, and never found in the state of pure earth”) also includes two genera—loams (
Terra miscella
) and
molds (
Humus
).
The second series,
stones
, includes nine genera in four chapters, based on interesting criteria that we would now regard as partly superficial and partly on the mark for wrong reasons. The four chapters comprise (1) stratified rocks made of grit (basically sandstones, divided by Mendes da Costa into two genera for finely stratified with many horizonal bedding planes versus massive
and blocky); (2) stratified rocks without grit and homogeneous (divided into two genera—basically limestones and slates in modern parlance—by the same criterion of massive versus thinly bedded); (3) marbles (separated as much for their importance to human arts as for any other reason); and (4) crystalline rocks, divided largely by the size of mineral grains into basalts and other finely crystalline
rocks, granites, and porphyries.
Mendes da Costa presents an interesting rationale—though ultimately flawed—for why a system that has worked so well for organisms should render equal service to inorganic geology. “It has been by pursuing such natural and simple methods as these, that botany has so eminently raised her head above her sister sciences,” he writes, acknowledging Linnaeus’s greatest
success.
Mendes da Costa recognizes the differences in formation between organic and inorganic objects, but he follows a common scientific conceit in thinking that a uniform system of classification will nonetheless suffice for all: he will engage only in humble and accurate description, not in fanciful theorizing. Differences in causality cease to matter when we cite only the pristine factuality
of objective appearances: “I have been very cautious not to indulge a speculative fancy in forming hypotheses or systems, the bodies being simply described, according to the appearances which they exhibit to the senses.”
Mendes da Costa then declares success because he has managed to balance, in a single system of compromises, all the competing schemes of his contemporaries. Such a “golden mean”
must yield optimality. Mendes da Costa argues that he has achieved two great balances in his system—first, by finding the “right” number of basic species as a compromise between “splitters” who love to make fine distinctions, and “lumpers” who search for essences and tend to unite objects in their quest for fundamental properties. (The terminology of splitting and lumping belongs to the twentieth
century, and the implied dichotomy cannot express all subtleties of this contrast, but the struggle between joiners and dividers has pervaded the history of taxonomy.) Mendes da Costa writes: “I have endeavored to reduce this study, hitherto deficient in respect of method, to a regular science, and in the attempt have been careful neither to multiply the species, nor lessen their number, unnecessarily.”
In a second balance, Mendes da Costa tries to unify the two disparate criteria then used to form systems for rocks and minerals—the efforts of his British compatriot John Woodward to base distinctions on observation of overt properties, both exterior and interior (“a method of arrangement founded on the growth, structure, and texture of fossils”); and continental systems based on “essential” properties
discovered by chemical experiment (for example, a threefold division according to various changes produced by fire into
calcarii
, for rocks calcined, or turned to lime [limestones and marbles, for example],
apyri
for those unaffected [asbestos and others], and
vitrificentes
for those vitrified to glass [quartz and other silicates]). Mendes da Costa tries to bring all systems together by making
primary divisions with observable properties (Woodward’s system), and then applying experimental and chemical results for refinements:
I have attentively examined the Woodwardian and Wallerian [continental] systems, and, finding them defective, have presumed to form a new one from the principles of both. I have endeavored to arrange fossils, not only according to their growth, texture, and structure,
but also their principles and qualities, as discovered by fire, and acid menstrua,
etc.
And in this way, I am confident that all the known fossils may be accurately distinguished; whereas, to attempt it by any one system hitherto followed, must occasion a strange confusion.
But Mendes da Costa’s efforts had to fail in principle because the causes and properties of rocks violate the requirements
of Linnaean geometry. Following the two central fallacies discussed earlier in this essay, Mendes da Costa could neither identify clear species, nor form distinct categories in a world of complete intergradation. Biological species are natural populations, distinct by historical continuity and current interaction, and unable to interbreed with others. Rock “species” are nondiscrete and intergrading.
Ultimately, Mendes da Costa just joined specimens that looked “enough alike”—a sure formula for endless bickering among specialists, for no two will ever agree. For example, he castigates his two great masters, Linnaeus and Wallerius, for designating too few species in the genus
Marmor
(marbles):
Wallerius, in his
Mineralogy
, and Linnaeus, in his
Systema Naturae
, are extremely confused in regard
to this genus of fossils; the former had divided all the marbles into only three species;
viz.
of uniform, variegated, and what he calls figured marbles . . . the latter has even made them all only varieties of one species; on which I cannot but make this observation, that it is a pity the learned should apply their studies rather to perplex science, than to elucidate it, and instruct mankind.
(I am no Freudian, but one certainly glimpses Oepidal feelings in Mendes da Costa’s complex attitude toward Linnaeus—basing his life’s work on the taxonomic system of his intellectual father, but then losing no opportunity to razz the master for moral turpitude on a range of issues from the terminology of clams to the number of species of marbles.)
But Mendes da Costa cannot claim final certainty
for his own divisions of
Marmor.
He designates eighty-one species, far more than for any other genus of rocks, and clearly to recognize the human utility of different colors and patterns, not because nature has fabricated discrete and discoverable “basic kinds.”
Equal difficulty and frustration dogged Mendes da Costa’s effort to establish the higher divisions of rocks and earths—for he encountered
complete intergradation between his genera as often as tolerable separation. For example, he frankly states his difficulty in dividing boles from clays, finally admitting that only convention dictates the separation:
Several authors do not make a distinct genus of the boles, but rank them among the clays; indeed very essential characteristics are wanting to make them different genera, for only
the extreme fineness of the particles of the boles is the cause of their being not so ductile or viscid as the clays, insomuch that speaking with propriety, they are only to be accounted very fine clays; I have, however, made them separate genera, as custom hath authorized it.
The human mind, with arrogance and fragility intermixed, loves to construct grand and overarching theories—a fault
perhaps encountered more often in our theological than in our scientific endeavors. But solutions often require the humbler, superficially less noble, and effectively opposite task of making proper divisions into different categories of meaning and causation. For only then can we build toward generality on a firmer substrate, and without feet sculpted of the genus
Argilla.
Mendes da Costa, following
Linnaeus, tried to bring all nature into one grand system of classification, but principles appropriate for the branching of organisms do not suit the continuities of rocks and earths. How ironic, ultimately, that a system doomed by too much togetherness should embody the best work of Emmanuel Mendes da Costa, the only Jewish naturalist of note in eighteenth-century Britain—a man from a culture
then defined by separation from a majority committed to the parochial notion that Anglican theology represented the one true and only way.
To heap irony upon irony, and to end with a return to the beginning, Mendes da Costa did understand the general principle that now leads us, with such confidence and justice, to classify rocks differently from organisms. I omitted one line in his critique
of Linnaeus’s sexual terminology for bivalves (as indicated by the ellipses on page 81), for these words cite a technical objection, rather than the moral argument then under discussion. In this line, now restored, Mendes da Costa rejects Linnaeus’s bivalve names “not only for their licentiousness, but also that they are in no ways the parts expressed.” How simple, and how correct! The top of a clam
is not the bottom of a person—and supposed visual similarities can only be misleading. Different terms should be used, lest people be lulled into false suppositions about meaningful or causal likeness. Similarly, rocks and organisms require different systems of classification to acknowledge their disparate modes of production—by timeless chemistry versus singular genealogy, by laws of nature versus
contingencies of history.