Science and Simplicity

1.

When asked to name, without reflection, the greatest scientific work that has ever been done, people who are themselves scientists will usually say “Newton’s Laws of Motion” or “Einstein’s Theory of Relativity.” Such answers are revealing of the image of ideal science with which we have been brought up, an image that has been of immense importance in the intellectual formation of working scientists. What we might call the “Newtonian Ideal” in science is the formulation of some principle of great generality, if not universality, a law or small set of laws that applies at all times and in all places.

This emphasis on the Newtonian Ideal differentiates the awarding of prestige within the community of scientists from the awarding of prizes and of popular recognition which, quite properly, often give considerable weight to the impact of science on the human condition. One can get a Nobel Prize for inventing a very useful gadget like the transistor¹ or for finding a treatment for prostate cancer,² and we are reminded daily at breakfast that Louis Pasteur invented a way to keep milk from growing bacteria before we can pour it on our cereal. Scientists, however, value most those assertions about nature that apply to the broadest possible domain of the material world, giving short notice, if any, to revelations about nature that apply only to a particular chemical or physical or biological object. There is a new crop of Nobel laureates every year, but there was only one Isaac Newton.

One consequence of the value placed on great generality is that there is necessarily a tenuous connection between what has actually been observed in the world of physical phenomena and the theoretical claim. Between the idea and the reality falls the shadow of abstraction. Newton’s First Law, that bodies at rest tend to stay at rest and bodies in motion tend to stay in motion in a straight line unless perturbed by an external force, could not possibly have been a generalization of the motions actually observed by him. Neither he nor any other seventeenth-century observer ever watched a body move in a perfect vacuum with no external forces operating on it.

The secret is in the word “tend.” Tendencies are not observable. They are an abstraction around which the observations of the actual movement of bodies in different circumstances can be organized, in an attempt to understand the “perturbing” forces. In a world in which real material objects have a diversity of sizes and composition and are being acted upon by a variety of forces, Newton’s First Law does not describe the motion of any particular object. The law that for every action there is an equal and opposite reaction may tell us what happens in the collision between two perfectly elastic bodies, but it is a poor prescription for winning at billiards.

The problem of the relation between the abstract structure of universal claims and the real world of particular events is especially acute in biology. Unlike planets, which are extremely large, or electrons, which are extremely small and internally homogeneous, living organisms are intermediate in size and are internally heterogeneous. They are composed of a number of parts with different properties that are in dynamic interaction with one another and the parts are, in turn, composed of yet smaller parts with their own interactions and properties. Moreover, they change their shapes and properties during their lifetimes, developing from a fertilized egg to a mature adult, ending finally sans teeth, sans hair, sans everything. Even the single-celled bacterium changes its internal properties from the moment that it is born from a division of its parent cell until the moment that it too divides. Organisms are a changing nexus of a large number of weakly determining interacting forces. As a consequence we have no universal laws in biology. The Biogenetic Law, “All life from life,” was only enacted a couple of billion years ago and could not always have been true or there would be no one to write for, publish, or read The New York Review of Books. Presumably once life arose from a handful of molecules it prevented more such events by gobbling up the rest of the soup.

Of Mendel’s famous three laws, inferred by him from studying a few characters in one species, two turn out to be untrue in a large fraction of cases and the third has a few very revealing exceptions. Is biology inevitably a story of different strokes for different folks, a collection of exquisitely detailed descriptions of the diverse forms and functions of organisms down to the molecular level, obtained from an unending history of experiment and observation? Or, from this booming, buzzing confusion can a biologist derive some general claims that are freed from the dirty particulars of each case, claims that, while not of the universality of Newton’s laws, at least characterize the properties of a very large part of the living world? Can there be a theoretical biology? Is there a possibility of Making Sense of Life?

One might have thought that Evelyn Fox Keller, by training a mathematician, would take an upbeat view of a program to formalize, mathematize, and generalize the observed diversity of living forms and processes. She is, however, by trade a philosopher and historian of science who, in analyzing the attempts to construct a theoretical biology, has come to a rather skeptical conclusion. Both history and epistemology seem to speak against it:

No one can deny the extraordinary advances that have been made over the course of this past century in our understanding of vital processes….

Yet I would argue that, despite such unquestionable success, biology is scarcely any closer to a unified understanding (or theory) of the nature of life today than it was a hundred years ago. The models, metaphors, and machines that have contributed so much to our understanding provide neither unity nor completeness. They work to answer some questions while avoiding (even obscuring) others.

It should not be supposed that Keller’s rather negative view flows from some particular and rigidly applied view of what counts as a successful explanation. On this matter she is more the sociologist than the philosopher, taking the position that each scientific community has its own “epistemological culture,” its own agreed-upon norms and standards of what counts as a sufficient understanding of a natural phenomenon. The epistemological culture that concerns her chiefly in Making Sense of Life is that of developmental biology, the science that is meant to explain how one fertilized egg cell, containing a variety of large molecules, including DNA and proteins, and intracellular structures, all organized in a particular spatial configuration, turns into a horse with a head at one end, a tail at the other, and legs at the four corners, while another fertilized egg cell, which looks pretty much like the first, turns into a clam.

There is an important distinction among theoretical structures in biology which Keller’s concept of epistemological cultures does not cover. Before even asking what counts as a sufficient explanation or understanding of some phenomenon, we need to decide what work a theoretical apparatus is supposed to do. Sometimes theoretical structures are nothing but calculating devices constructed from a complete and unproblematical knowledge of all the underlying mechanical details. The purpose of such a calculating device is to predict how differences in inputs into the system will be reflected in the output.

The classic example in biology is theoretical population and evolutionary genetics. All the relevant elementary processes are already known. These include all the mechanisms of inheritance, the phenomena of mutation, of migration, of the effect of limited population size, and of the operation of natural selection through differential survivorship and fertility. Theoretical evolutionary genetics assembles all these phenomena into a formal mathematical structure that predicts changes in the genetic composition of populations and species over time as a function of the numerical values of these elementary processes.

In contrast, sometimes theories are meant to help us “understand” a process whose outcome has been observed but whose dynamical details are not known from experiment or observation. The theory provides a formal structure into which, it is supposed, the actual mechanical details will fit if we ever get to know them. In the most extreme case the theory is without any direct reference to any underlying material phenomena. So there are theories of embryonic development that are nothing but networks of logical switches, formal components that act to increase or decrease the “output” (of unspecified physical nature) of other components. Even when some details of the phenomena are known from experiment, a model based on an incomplete knowledge leads to frustration. Keller describes a model of development that took into account interactions among five genes known to be among the critical elements in the normal formation of body segments in fruit flies. Even though only these five genes were considered, the mathematical model required 136 equations involving about 50 parameters that had never been measured experimentally.

In view of this absence of information, the investigators asked if there was any set of values of the parameters in a biologically reasonable range that would produce a stable development result resembling the observed pattern of body segments in flies. The answer was no. But Keller’s dissatisfaction transcends that failure. Even if the model had succeeded in mimicking the observed pattern of development, it was “cumbersome, messy, and, by itself effectively opaque to any kind of intuition (mathematical or otherwise).” That is, the work to be done by a model of development is not a computational one like those in evolutionary genetics or in the very cumbersome and messy models used for weather forecasting, but to help our “understanding” of the phenomenon. But this raises a very deep issue.

As Keller points out, models in physical science have traditionally aimed for an elegance and simplicity that would allow our limited mental powers to “understand” some aspect of nature. If models in biology are to advance our understanding, then they too must be simple and elegant. Keller devotes a chapter to two famous past attempts to construct a mathematical biology that is simple and elegant yet at the same time captures the essence of biological phenomena. The more ambitious was the founding of a school of mathematical biophysics by Nicolas Rashevsky in the late 1930s, with its own scientific journal and a formal graduate-degree program at the University of Chicago.

All aspects of biology were to be included in the intellectual program, but there was particular emphasis on cell biology, development, and the nervous system. The approach of the Rashevsky school was to make simplified physical models that were supposed to capture the essence of a biological phenomenon and then describe models in mathematical terms. What Rashevsky and his school failed to take into account was the conviction of biologists that real organisms were complex systems whose actual behavior would be lost in idealizations. The work of the school was regarded as irrelevant to biology and was effectively terminated in the late 1960s, leaving no lasting trace.³

The other well-known essay into mathematical simplification of a complex biological process was the work of Alan Turing, best known for having laid the logical foundations of computer programming. Turing suggested that early embryonic development could be understood as being induced by different concentrations of certain (unspecified) molecules, “morphogens,” at different places in the embryo. These spatial patterns would arise from chemical interactions of the morphogens that were at first spread throughout the embryo more or less uniformly or in simple increasing gradients from one end of the embryo to another.

The irony is that while Turing’s model turns out to be correct in its simple outline form, it did not play a significant part in the production of modern molecular developmental biology. Developmental geneticists now study how the cell’s reading of different genes produces spatial patterns of molecules in embryos, but that detailed and messy description, involving large numbers of genes and proteins, owes nothing to Turing’s model. It may be that the complexity of biological systems makes them more like the hydrodynamics that account for the weather rather than like the revolution of the earth around the sun. Then, given enough information, we might be able to compute an organism’s developmental trajectory, but we will never “understand” it. In the case of the weather, at least we can compute whether we need to carry an umbrella tomorrow. It is not at all clear why I would want to compute an organism.

What hope is there then for the ambitious biologist, before whose eyes the Newtonian Ideal has been held up as a model? If there is no abstract law of the living, there may be, at least, a mechanism that is common to all living beings and that distinguishes them from the inanimate world. If such a “mechanism of life” existed, then whoever elucidated it would have a claim on biological Newtonhood. The most striking feature of organisms is that they are reproduced generation after generation in such a way that they resemble their parents. Lions give birth to lions and lambs to lambs. But, more remarkable still, that resemblance is not the result of a mechanical copying like the reproduction made by an office copying machine. The organism develops its resemblance to its parent over a period of time, beginning with a rudimentary object that bears no outer hint of what it will become. What is passed between generations, then, is not form itself, but some kind of information about the construction of that form.

The material basis of that information is the organism’s set of genes. Whatever that genetic material may be, it must have two properties. First, it must be of such a nature that it can be copied over and over again from generation to generation with little or no change. Second, it must be capable of giving enough information to specify that a lion is to be built and not a lamb. It was long realized that organisms possessed and passed on to their offspring two sorts of molecules that filled the bill, proteins and nucleic acids. Both made up the material of the chromosomes, acknowledged to be the carriers of the genes. Proteins are made of long strings of smaller molecules, amino acids, of which there are twenty different kinds, so there would be an immense variety of genes that could be specified by even moderately long protein chains of different mixtures of amino acids. Moreover, proteins are manufactured by the cell machinery using other proteins as catalysts, so the mechanism of reproduction is in place. Nucleic acids are also made up of smaller subunits, nucleotides, which come in only four varieties (referred to by the cognoscenti by their initials, A, T, G, and C), but if genes consisted of long-enough chains of these, there would again be plenty of variety possible to account for all the different genes that organisms possessed.

By the early 1950s it had become apparent from critical experiments in bacteria and their parasitic bacteriophages that it was the nucleic acid, DNA, and not protein that was the material stuff of the gene. But the simple identification of DNA as the genetic material could have little consequence for further biological work unless a detailed narrative could be constructed of how that molecule was copied and how its information was read by the cell. It seemed obvious that the answer to these questions could come only if a picture of the three-dimensional structure of DNA were produced and that the revelation of that structure would have immense consequences for an understanding of heredity and development.

2.

The story of the brief struggle to solve the structure of DNA is part of the folklore of science. That folklore has been created by narratives told from the standpoint of the different participants, either by the actors themselves or by their sympathetic biographers. The first were James Watson’s macho The Double Helix⁴ and Anne Sayre’s alternative history, Rosalind Franklin and DNA⁵ (Francis Crick’s What Mad Pursuit⁶ seems to have been much less influential). Those original narratives are now succeeded by two more ample and very revealing biographies, Brenda Maddox’s Rosalind Franklin and Victor McElheny’s Watson and DNA. For Franklin, who died in 1958, her life prior to the work on DNA builds a consistent picture of a personality and style of science that, coupled with institutionalized sexism of the time, makes the final result seem inevitable. A member of a family that was as close to being upper class as English Jews can be (she was a descendant of Sir Moses Montefiore and a grand-niece of Lord Samuel), she had a self-confidence and hauteur that allowed her to resist pressures from her colleagues to take other scientific views into account. She knew when she was right and she would not be moved even by persons in authority.

For Watson, it is his life after the Big Announcement that reveals so much about the man and the institutions of science within which he operated and achieved success and notoriety. After 1953 Watson did essentially no science. He became an administrator, a politician, and a general guru. His considerable intelligence, his dominating personality, his unbounded self-confidence, and his double helical crown of laurel opened positions of authority to him. He became the director of one of the world capitals of molecular biology, the Cold Spring Harbor Laboratory, and, for a time, the very entrepreneurial head of the new Human Genome Project of the National Institutes of Health. He was finally forced out of the project by a new director of the NIH, Bernardine Healy, ostensibly because of his opposition to patenting genes, and, some claimed, conflicts of interest, which he denied. But I suspect that Healy simply found him overbearing.

The outline of the DNA story is uncontested. In 1951, Franklin, based on her reputation for X-ray diffraction studies on the structure of coal and other carbon forms, was invited to join the laboratory at King’s College, London, by the director, J.T. Randall. There she was to work on the structure of DNA, a project already being pursued at King’s by Maurice Wilkins, although neither she nor Wilkins were informed in advance by Randall of the other’s project. At about the same time, Watson arrived at the Cavendish Laboratory in Cambridge as a post-doctoral fellow where he fell in with Crick, who was a graduate student working on the structure of hemoglobin. Watson and Crick immediately got interested in DNA and decided to try to build models of its structure using data that were being produced by others. Their first model was hopeless and they were warned off DNA for a time by the head of the Cavendish Laboratory, Sir Lawrence Bragg, who had a gentlemen’s agreement with Randall that the DNA problem belonged to King’s College. But they quietly persisted. Meanwhile, Franklin was producing superb X-ray diffraction pictures of two forms of DNA, one of which was obviously helical, but the other seemed not to be. She was skeptical that native DNA was a helix and continued to try to resolve the problem by yet more empirical work. Watson and Crick continued to turn out one erroneous helical model after another, each error being corrected when someone else, including Franklin, pointed out a contradiction between the model and empirical data.

The contrast in views of how to do science could not be greater. For Franklin, whom Watson characterizes as “obsessively professional,” the evidence would finally speak for itself. “How will a model show which structure is right? The Patterson [a form of data representation] will tell us the structure.” For Watson and Crick (laid-back amateurs?) data were useless without a prior concept. The facts could serve only to suggest a range of models and as a check against errors. They garnered their facts where they could. DNA was already known to consist of long chains of alternating sugars and phosphates, along the length of which the nucleotides were attached. The chemical structure was already published, including the critical discovery by Erwin Chargaff that the four nucleo- tides were in paired amounts with the concentration of A equal to the concentration of T, and the G concentration equal to the C concentration. Franklin and others had already told them that the sugar-phosphate chains had to be on the outside surface of the molecule.

Early in 1953, Wilkins, an old friend of Crick’s, and now deputy director of the King’s College laboratory, showed Watson the now famous Photograph 51 that Franklin had made of DNA, showing unequivocally that DNA was a double helix. Soon after, they received from Max Perutz, a senior colleague in the Cambridge laboratory, a copy of a government review committee’s report on the work in London which convinced them that the two strands of the helix ran in opposite directions.

In making the final model Watson could not get the strands to fit together until another member of the laboratory, Jerry Donahue, noticed what was going on and told Watson he had the wrong chemical structure for the nucleotides. And then it worked. Watson and Crick now made a model of DNA that could carry a dazzling variety of information and, in its complementary double structure, provide one neat basis for the cell to replicate it. I emphasize “one” because the mechanism used by cells to read the information in DNA is already sufficient to replicate a single stranded form. Cells just don’t happen to do it that way.

The books about the “race” to find the structure of DNA capture some but not all of the dynamic. Quite properly they consider the effect of personal styles and visions of what it is to do science. In their emphasis on the effect of sexism and on the exploitation of lines of friendship and camaraderie to gain advantage in a competition for success they deal with forces of considerable power in determining the outcome. But these need to be embedded in a broader view of how academic science is organized. Science is carried out in research groups. Whether large or small, such groups are hierarchies consisting of a senior professor or laboratory director, perhaps with another fairly senior scientific colleague, a group of graduate students, postdoctoral fellows, visiting investigators, and a crew of laboratory assistants. All owe their positions to the senior laboratory head who determines entry into the group, controls most of the funds used to carry out research, and decides what research is to be done.

Flowing from this dependency relation, the senior members of the group have de facto intellectual property rights over the scientific product of the group members, property rights that are recognized, although resented, by the dependents. One of the most infamous manifestations of this feudal relationship is the almost universal practice of adding the director’s name as an author on every publication that results from the work of any member of the group even when no significant intellectual or experimental contribution has been made by the head. This de facto ownership means that the laboratory head can also dispose of the results of research in any way he or she feels fit, including sharing it with others in private or public. One of the rituals of science is the hour-long verbal report of research by a senior scientist at a national or international conference during which no credit is given to any person in the group except for a moment at the end with the display of a long list of names to whom thanks are due. It would not have occurred to the senior actors in the DNA story to have played any other roles but the ones to which custom had assigned them, and the fact that they were dealing with a “difficult woman” would have added further sources of self-justification.

One way to turn a correct guess about the structure of a molecule into a place among the immortals is to endow that chemical with mystical powers like the narcotic soma of Hindu ritual, to turn DNA, the sticky fiber, into DNA: The Secret of Life. The book, issued to mark the fiftieth anniversary of the publication of Watson and Crick’s original paper, is an odd work. It is the result of an active collaboration between Watson and Andrew Berry, a very accomplished, critical, and skeptical molecular population geneticist who has turned to writing about science. Despite the fact that the rhetoric is almost entirely Berry’s, DNA: The Secret of Life is cast as an autobiography with “I,” “we,” and “Nobel Prize” as the four most common words. Sometimes “we” is the standard rhetoric of scientists who speak for general scientific knowledge, as in “we now know,” but sometimes, as in “we labored” or “we found genes,” Watson seems to suggest his direct involvement in research which was, in fact, not his work at all. Only one chapter is devoted to yet another retelling of the Watson, Crick, Franklin, Wilkins story. The rest form a quite accessible explanation of the role that work on DNA has played in investigations of basic cell biology, experimental genetics, agriculture, drug and disease research, criminal investigation, human evolutionary history, and human diversity.

It is in these chapters that some of the contradictions between the views of the authors become apparent. In the chapter on “Who We Are,” for example, we see the tension between Watson, always anxious to emphasize the power of DNA, the secret of life, to determine all human personal and cultural properties, and Berry’s more critical view that takes account of the interaction between genetic and environmental causes. The systematically fraudulent data of Cyril Burt on the IQs of twins is treated with a light touch and with the suggestion that maybe it never happened or, if it did, involved only the invention of “a few twins from time to time if he needed to bolster sample sizes.”

While in some parts of the book emphasis is placed on the cell machinery that reads the DNA and manufactures the proteins, an emphasis consistent with Berry’s understanding, in the chapter on “Reading Genomes” we are told that “a gene producing the biological equivalent of a brick will, left to its own devices, produce a pile of bricks.” But this is Watsonian hype. Gene don’t have their “own devices.” Left to their own devices they will just sit there, dead molecules. One might as well say that a set of house plans, “left to its own devices,” will build a house. To carry out the synthesis of proteins and other components of a developing organism, the cell uses an elaborate machinery of proteins and a warehouse of small parts, both of which are already in place in the unfertilized egg. This machinery transcribes the information in the DNA into a related information-bearing molecule, RNA, which may be assembled from information from more than one gene. Using this RNA chain as a guide, the cell then assembles chains of amino acids using its stock of small parts. To make an active protein, a chain of amino acids must be folded into exactly the right three-dimensional form, a process that is partly determined by the sequence of amino acids, but is guided by yet other proteins and small molecules. Sometimes, before the folding can occur, pieces of the original amino acid chain are clipped out. This entire manufacturing process, from the original transcription of the DNA information to the finally assembled and folded protein, could not take place without the prior presence in the cell of a special set of protein catalysts, enzymes. So much for the gene’s “own devices.”

For the uninformed seeker after truth, reading DNA: The Secret of Life is like walking across a meadow inhabited by interesting plants and animals. For the most part the surface is firm and will bear weight, but occasionally one’s foot goes through to the muck below. When reading the book, don’t forget your rubbers.