How can we reconcile such opposite observations? How can such different effects arise from the same cause? How can movements that have abraded quartz, rock crystal, and the hardest stones into rounded pebbles, also have preserved light and fragile shells?
The simple answer to this specific and limited question may then lead to important generalities for the science of geology, and also to criteria for unraveling the particular history of the earth:
At first glance, this contrast of tranquillity and movement, of organization and disorder, of separation and mixture, seemed inexplicable to me; nevertheless, after seeing the same phenomena again and again, at different times and in different places, and by combining these facts and observations, it seemed to me that one could explain these striking observations in a simple and natural manner that could then reveal the principal laws followed by nature in the generation of horizontal strata.
Lavoisier then presents his idealized model of a two-stage cycle as an evident solution to this conundrum:
Two kinds of very distinct beds must exist in the mineral kingdom: one kind formed in the open sea … which I shall call pelagic beds, and the other formed at the coast, which I shall call littoral beds.
Pelagic beds arise by construction, as “shells and other marine bodies accumulate slowly and peacefully during an immense span of years and centuries.” But littoral beds, by contrast, arise by “destruction and tumult … as parasitic deposits formed at the expense of coastlines.”
In a brilliant ploy of rhetoric and argument, Lavoisier then builds his entire treatise as a set of consequences from this simple model of two types of alternating sediments, representing the cycle of a rising and falling sea. This single key, Lavoisier claims, unlocks the great conceptual problem of moving from one-dimensional observations of vertical sequences in several localities to a three-dimensional reconstruction of history. (I call the solution three-dimensional for a literal reason, emphasized earlier in this essay in my discussion of geological maps: the two horizontal dimensions record geographic variation over the earth’s surface, while the vertical dimension marks time in a sequence of strata):
This distinction between two kinds of beds … suddenly dispersed the chaos that I experienced when I first observed terranes made of horizontal beds. This same distinction then led me to a series of consequences that I shall try to convey, in sequence, to the reader.
The remainder of Lavoisier’s treatise presents a brilliant fusion of general methodology and specific conclusions, a combination that makes the work such a wonderful exemplar of scientific procedure at its best. The methodological passages emphasize two themes: the nature of proof in natural history, and the proper interaction of theory and observation. Lavoisier roots the first theme in a paradox presented on pages 98—99: the need to simplify at first in order to generalize later. Science demands repetition for proper testing of observations—for how else could we learn that the same circumstances reliably generate the same results? But the conventional geologies of Lavoisier’s time stymied such a goal—for one directional period of deposition from a single stationary sea offered no opportunity for testing by repetition. By contrast, Lavoisier’s model of alternating pelagic and littoral beds provided a natural experiment in replication at each cycle.
But complex nature defies the needs of laboratory science for simple and well-controlled situations, where events can be replicated under identical conditions set by few variables. Lavoisier argues that we must therefore try to impose similar constraints upon the outside world by seeking “natural experiments” where simple models of our own construction might work adequately in natural conditions chosen for their unusual clarity and minimal number of controlling factors.
Consider three different principles, each exploited by Lavoisier in this paper, for finding or imposing a requisite simplicity upon nature’s truly mind-boggling complexity.
1. Devise a straightforward and testable model. Lavoisier constructed the simplest possible scheme of seas moving in and out, and depositing only two basic (and strongly contrasting) types of sediment. He knew perfectly well that real strata do not arrange themselves in neat piles of exactly repeating pairs, and he emphasized two major reasons for nature’s much greater actual complexity: first, seas don’t rise and fall smoothly, but rather wiggle and jiggle in small oscillations superposed upon any general trend; second, the nature of any particular littoral deposit depends crucially upon the type of rock being eroded at any given coastline. But Lavoisier knew that he must first validate the possibility of a general enterprise—three-dimensional reconstruction of geological history—by devising a model that could be tested by replication. The pleasure of reveal ing unique details would have to wait for another time. He wrote:
Beds formed along the coast by a rising sea will have unique characteristics in every different circumstance. Only by examining each case separately, and by discussing and explaining them in comparison to each other, will it be possible to grasp the full range of phenomena…. I will therefore treat [these details] in a separate memoir.
2. Choose a simple and informative circumstance. Nature’s inherent complexity of irreducible uniqueness for each object must be kept within workable scientific bounds by intelligent choice of data with unusual and repeated simplicity. Here Lavoisier lucked out. He had noted the problem of confusing vari ation in littoral deposits based on erosion of differing rocks at varying coasts. Fortunately, in the areas he studied near Paris, the ancient cliffs that served as sources for littoral sediments might almost have been “made to order” for such a study. The cliffs had been formed in a widespread deposit of Cretaceous age called La Craie (or “the Chalk”—the same strata that build England’s White Cliffs of Dover). The Chalk consists primarily of fine white particles, swiftly washed out to sea as the clifís erode. But the Chalk also includes interspersed beds of hard flint nodules, varying in size from golf balls to baseballs in most cases. These nodules provide an almost perfect experimental material (in uniform composition and limited range of size) for testing the effects of shoreline erosion. Lavoisier noted in particular (I shall show his engravings later in this essay) that the size and rounding of nodules should indicate distance of deposition from the shoreline—for pebbles should be large and angular when buried at the coast (before suffering much wear and erosion), but should then become smaller and rounder as they tumble farther away from the coastline in extensive erosion before deposition.
3. Ask a simple and resolvable question. You needn’t (and can’t) discover the deep nature of all reality in every (or any!) particular study. Better to pose smaller, but clearly answerable, questions with implications that then cascade outward toward a larger goal. Lavoisier had devised a simple, and potentially highly fruitful, model of oscillating sea levels in order to solve a fundamental question about the inference of a region’s geological history from variation in vertical sections from place to place—the sections that he had placed in the margins of the maps he made with Guettard. But such a model could scarcely fail to raise, particularly for a man of Lavoisier’s curiosity and brilliance, the more fundamental question—a key, perhaps, to even larger issues in physics and astronomy—of why oceans should rise and fall in repeated cycles. Lavoisier noted the challenge and wisely declined, recognizing that he was busy frying some tasty and sizable fish already, and couldn’t, just at the moment, abandon such a bounty in pursuit of Moby Dick. So he praised his work in progress and then politely left the astronomical question to others (although he couldn’t resist the temptation to drop a little hint that might help his colleagues in their forthcoming labors):
It would be difficult, after such perfect agreement between theory and observation—an agreement supported at each step by proofs obtained from strata deposited by the sea—to claim that the rise and fall of the sea [through time] is only a hypothesis and not an established fact derived as a direct consequence of observation. It is up to the geometers, who have shown such wisdom and genius in diff
erent areas of physical astronomy, to enlighten us about the cause of these oscillations [of the sea], and to teach us if they are still occurring, or if it is possible that the earth has now reached a state of equilibrium after such a long sequence of centuries. Even a small change in the position of the earth’s axis of rotation, and a consequent shift in the position of the equator, would suffice to explain all these phenomena. But this great question belongs to the domain of physical astronomy, and is not my concern.
For the second methodological theme of interaction between observation and theory in science, Lavoisier remembered the negative lesson that he had learned from the failures of his mentor Guettard. A major, and harmful, myth of science—engendered by a false interpretation of the eminently worthy principle of objectivity—holds that a researcher should just gather facts in the first phase of study, and rigorously decline to speculate or theorize. Proper explanations will eventually emerge from the data in any case. In this way, the myth proclaims, we can avoid the pitfalls of succumbing to hope or expectation, and departing from the path of rigorous objectivity by “seeing” only what our cherished theory proclaims as righteous.
I do appreciate the sentiments behind such a recommendation, but the ideal of neutrally pure observation must be judged as not only impossible to accomplish, but actually harmful to science in at least two major ways. First, no one can make observations without questions in mind and suspicions about forthcoming results. Nature presents an infinity of potential observations; how can you possibly know what might be useful or important unless you are seeking an answer to a particular puzzle? You will surely waste a frightful amount of time when you don’t have the foggiest idea about the potential outcomes of your search.
Second, the mind’s curiosity cannot be suppressed. (Why would anyone ever want to approach a problem without this best and most distinctive tool of human uniqueness?) Therefore, you will have suspicions and preferences whether you acknowledge them or not. If you truly believe that you are making utterly objective observations, then you will easily tumble into trouble, for you will probably not recognize your own inevitable prejudices. But if you acknowledge a context by posing explicit questions to test (and, yes, by inevitably rooting for a favored outcome), then you will be able to specify—-and diligently seek, however much you may hope to fail—the observations that can refute your preferences. Objectivity cannot be equated with mental blankness; rather, objectivity resides in recognizing your preferences and then subjecting them to especially harsh scrutiny—and also in a willingness to revise or abandon your theories when the tests fail (as they usually do).
Lavoisier had spent years watching Guettard fritter away time by an inchoate gathering of disparate bits of information, without any cohesive theory to guide and coordinate his efforts. As a result, Lavoisier pledged to proceed in an opposite manner, while acknowledging that the myth of objectivity had made his procedure both suspect and unpopular. Nonetheless, he would devise a simple and definite model, and then gather field observations in a focused effort to test his scheme. (Of course, theory and observation interact in subtle and mutually supporting ways. Lavoisier used his preliminary observations to build his model, and then went back to the field for extensive and systematic testing.) In an incisive contrast between naive empiricism and hypothesis testing as modes of science, Lavoisier epitomized his preference for the second method:
There are two ways to present the objects and subject matter of science. The first consists in making observations and tracing them to the causes that have produced them. The second consists in hypothesizing a cause, and then seeing if the observed phenomena can validate the hypothesis. This second method is rarely used in the search for new truths, but it is often useful for teaching, for it spares students from difficulties and boredom. It is also the method that I have chosen to adopt for the sequence of geological memoirs that I shall present to the Academy of Sciences.
Lavoisier therefore approached the terranes of France with a definite model to test: seas move in and out over geographic regions in cycles of advancing and retreating waters. These oscillations produce two kinds of strata: pelagic deposits in deeper waters and littoral deposits fashioned from eroded coasts near the shoreline. Type of sediment should indicate both environment of deposition and geographic position with respect to the shoreline at any given time: pelagic deposits always imply a faraway shore. For the nearer littoral deposits, relative distance from shore can be inferred from the nature of any particular stratum. For littoral beds made mainly of flint nodules derived from Chalk, the bigger and more angular the nodules, the closer the shoreline.
From these simple patterns, all derived as consequences of an oscillating sea, we should be able to reconstruct the three-dimensional history of an entire region from variation in vertical sequences of sediments from place to place. (For example, if a continuous bed representing the same age contains large and angular flint nodules at point A, and smaller and more rounded nodules at point B, then A lay closer to the shoreline at the time of deposition.)
Lavoisier devotes most of his paper, including all of his seven beautifully drafted plates, to testing this model, but I can summarize this centerpiece of his treatise in three pictures and a few pages of text because the model makes such clear and definite predictions—and nature must either affirm or deny. Lavoisier’s first six plates—in many ways, the most strikingly innovative feature of his entire work—show the expected geographic distribution of sediments under his model.
Lavoisier’s first plate, for example, shows the predictable geographic variation in a littoral bed formed by a rising sea. The sea will mount from a beginning position (marked “ligne de niveau de la basse mer,” or line of low sea level, and indicated by the top of the illustrated waters) to a highstand marked “ligne de niveau de la haute mer,” or line of high sea level. The rising sea beats against a cliff, shown at the far left and marked “falaise de Craye avec cailloux,” or cliff of Chalk with pebbles. Note that, as discussed previously, this Chalk deposit contains several beds of flint nodules, drawn as thin horizontal bands of dark pebbles.
The rising sea erodes this cliff and deposits a layer of littoral beds underneath the waters, and on top of the eroded chalk. Lavoisier marks this bed with a sequence of letters BDFGHILMN, and shows how the character of the sediment varies systematically with distance from the shoreline. At B, D, and F, near the shore, large and angular pebbles (marked “cailloux roulés,” or rolled pebbles), formed from the eroded flint nodules, fill the stratum. The size of particles then decreases continually away from shore, as the pebbles break up and erode (changing from “sable grossier,” or coarse sand, to “sable fin,” or fine sand, to “sable très fin ou argille,” or very fine sand and clay). Meanwhile, far from shore (marked KK at the right of the figure), a pelagic bed begins to form (marked “commencement des bancs calcaires,” or beginning of calcareous beds).
From this model, Lavoisier must then predict that a vertical section at G, for example, would first show (as the uppermost stratum) a littoral bed made of large and angular pebbles, while a vertical section at M would feature a pelagic bed on top of a littoral bed, with the littoral bed now made of fine sand and clay. The two littoral beds at G and M would represent the same time, but their differences in composition would mark their varying distances from the shore. This simple principle of relating differences in beds of the same age to varying environments of deposition may seem straightforward, but geologists did not really develop a usable and consistent theory of such “facies” (as we call these variations) until this century. Lavoisier’s clear vision of 1789, grossly simplified though his example may be, seems all the more remarkable in this context.
Lavoisier’s first plate, showing spatial variation in sediments deposited in a rising sea.
Lavoisier then presents a series of similar diagrams of growing complexity, culminating in plate 6, also reproduced here. This final plate shows the results of a full cycle—the sea, having advance
d to its full height, has already retreated back to its starting point. The chalk cliff has been completely eroded away and now remains only as a bottom layer. (Note the distinctive bands of flint nodules for identification. I will discuss later the lowermost layer, marked “ancienne terre,” or ancient earth.) A lower littoral layer, formed by the rising sea, lies above the eroded chalk (marked HLMN as “bancs littoraux inférieurs formés à la mer montante,” or lower littoral beds formed by the rising sea). A pelagic bed, marked KKK (don’t blame Lavoisier for a later and fortuitous American anachronism!), lies just above (labeled “bancs pelagiens calcaires horizontaux supérieurs,” or upper calcareous horizontal pelagic beds). Note how the pelagic bed pinches out toward shore because sediments of this type can be deposited only in deep water. This pelagic bed forms when sea level reaches its highest point. Then, as the sea begins to fall, another littoral bed will be deposited in progressively shallower water atop the pelagic bed (marked HIGG and labeled “bancs littoraux supérieurs formés à la mer déscendante,” or upper littoral beds formed by the falling sea).
Again, Lavoisier’s insights are subtle and detailed—and several specific predictions can be derived from his model. For example, the upper and lower littoral beds will be confluent near the coast because the intervening pelagic bed didn’t reach this far inland. Thus, a vertical section taken near the coast should show a single thick littoral bed made of large and angular pebbles. But farther away from shore, a vertical section should include a full array of alternating beds, illustrating the entire cycle and moving (top to bottom, as shown by the vertical line, located just left of center and marked 12345) from the upper littoral bed of the falling sea (1), to the intervening pelagic bed (2), to the lower littoral bed of the rising sea (3), to the underlying chalk (4), and finally to the foundation of the ancienne terre (5).