. . . every work of science great enough to be remembered for a few generations affords some exemplification of the defective state of the art of reasoning of the time when it was written; and each chief step in science has been a lesson in logic.

― Charles Sanders Peirce

15.1 The values of science

Science may be said to have great value both as a practical resource, and as an intellectual satisfaction.

As a practical matter, modern science has profoundly changed almost every aspect of our lives. Technological developments ― in farming and manufacturing, in communication and transportation, in public and private health ― have resulted in greatly improved standards of living over much of the world. On the other hand, some practical results of science have not been cheerful. We have harnessed nuclear power, but the destructive power of nuclear weapons has become a menace to civilization. Computers enhance our powers, yet threaten our privacy. Scientific advances in the understanding of the genetic makeup of human beings promise new cures for terrible diseases, but they also confront us with ethical dilemmas. Industrial improvements make it possible to support the increasing world population, while the industrial pollution of the planetary environment that has accompanied this growth has threatened the very habitability of the globe.

Most will agree that the development of science and its applications have, on balance, benefited humanity. Terrible as recent wars have been, their toll of human life has been much smaller than that of the great plagues that formerly swept over crowded continents, decimating the population. Those plagues have been almost completely wiped out by modern medical science. More recently, terrible famine has threatened overpopulated lands, for which the best hope lies in the scientific management of the environment. The practical value of science lies in the easier and more abundant life made possible by technological advances based on scientific knowledge. This great benefit is cogently defended in the following passage:

It's sometimes argued that there's no real progress; that a civilization that kills multitudes in mass warfare, that pollutes the land and oceans with ever larger quantities of debris, that destroys the dignity of individuals by subjecting them to a forced mechanized existence can hardly be called an advance over the simpler hunting and gathering and agricultural existence of prehistoric times. But this argument, though romantically appealing, doesn'tfar less moral justification than modern ones. A technology that produces debris can find, and is finding, ways of disposing of it without ecological upset. And the schoolbook pictures of primitive man sometimes omit some of the detractions of his primitive life ― the pain, the disease, famine, the hard labor needed just to stay alive. From that agony of bare existence to modern life can be soberly described only as upward progress, and the sole agent for this progress is quite clearly reason itself.[1]

But the laws and principles discovered in scientific investigation have a value apart from any practical utility they may possess. This intrinsic value is the satisfaction of curiosity, the fulfillment of the desire to know. Aristotle wrote long ago that "to be learning something is the greatest of pleasures not only to the philosopher but also to the rest of mankind, however small their capacity for it."[2] The usefulness of science is widely understood, but we need not reduce all human wants to practical wants, and the great contributors to scientific progress have not themselves emphasized this pragmatic aspect of what they do. When asked about their own motives for research they commonly answer as did Albert Einstein, perhaps the most distinguished scientist of the twentieth century:

What, then impels us to devise theory after theory? Why do we devise theories at all? The answer to the latter question is simply: because we enjoy "comprehending," that is, reducing phenomena by the process of logic to something already known or (apparently) evident).[3]

This penetrating response suggests a very fruitful conception of the nature of science.

The task of science, we all know, is to discover facts; science is built up with facts, as a house may be built up with stones. But a mere collection of facts no more constitutes a science (as the philosopher Jules Poincaré observed) than a heap of stones constitutes a house.[4] Some parts of science do focus on this or that particular fact. A geographer, for example, may be interested in the exact configuration of a particular coastline, or a botanist in the characteristics of a particular plant. But in the more advanced activities of science, bare descriptive knowledge of this or that particular fact is of little importance. General truths are what the scientist is after; it is to illustrate these truths and to provide evidence for them that particular facts are chiefly sought.

Isolated particular facts may be known ― in a sense ― by direct observation. That a particular object falls when released, that this ball moves more slowly down an inclined plane than it did when dropped directly downward, that the tides ebb and flow ― all these are matters of fact open to direct inspection. But scientists seek more than a mere record of such phenomena; they strive to understand them. To this end they seek to formulate general laws that state the patterns of all such occurrences and the systematic relationships between them. The scientist searches for natural laws that govern particular events, and for the fundamental principles that underlie them.

This preliminary exposition of the theoretical aims of science can be made clearer by means of an example. By careful observation and the application of geometrical reasoning to the data thus collected, the Italian physicist and astronomer Galileo Galilei (1564 1642) succeeded in formulating the laws of falling bodies, which gave a very general description of the behavior of bodies near the surface of the earth. At about the same time the German astronomer Johannes Kepler (1571 1630), basing his reasonings very largely on the astronomical data collected by Denmark's Tycho Brahe (1546 1601), formulated the laws of planetary motion describing the elliptical orbits traveled by the planets around the sun. Each of these two great scientists succeeded in unifying the various phenomena in his own field of investigation by formulating the interrelations between them: Kepler in celestial mechanics, Galileo in terrestrial mechanics. Their discoveries were great achievements, but they were, after all, separate and isolated. Just as separate particular facts challenge the scientist to unify and explain them by discovering their lawful connections, so a plurality of general laws challenges the scientist to unify and explain them by discovering a still more general principle that subsumes the several laws as special cases. In the case of Kepler's and Galileo's laws, this challenge was met by one of the greatest scientific geniuses of all time, Sir Isaac Newton (1642 1727). By his theory of gravitation and his three laws of motion, Newton unified and explained celestial and terrestrial mechanics, showing them both to be deducible within the framework of a single more fundamental theory.[5] Scientists seek not merely to know what the facts are, but to explain them, and to this end they devise "theories." To understand exactly what is involved here, we must consider the general nature of explanation itself.

15.2 Explanations: scientific and unscientific

In everyday life it is the unusual or startling for which we demand explanations. An office boy may arrive at work on time every morning and no curiosity will be aroused. But let him come an hour late one day, and his employer will demand an explanation. What is it that is wanted when an explanation for something is requested? An example will help to answer this question. The office boy might reply that he had taken the seven thirty bus to work as usual, but the bus had been involved in a traffic accident which entailed considerable delay. In the absence of any other transportation, the boy had had to wait (he reports) a full hour for the bus to be repaired. This account would probably be accepted as a satisfactory explanation. It can be so regarded because from the statements that constitute the explanation the fact to be explained follows logically and no longer appears puzzling. An explanation is a group of statements or a story from which the thing to be explained can logically be inferred and whose acceptance removes or diminishes its problematic or puzzling character. Of course, the inference of the fact as conclusion from the explanation as premiss might have to be enthymematic, where the "understood" additional premisses may be generally accepted causal laws,[6] or the conclusion may follow with probability rather than deductively. It thus appears that explanation and inference are very closely related. They are, in fact, the same process regarded from opposite points of view. Given certain premisses, any conclusion that can logically be inferred from them can be regarded as being explained by them. And given a fact to be explained, we say that we have found an explanation for it when we have found a set of premisses from which it can logically be inferred. As was indicated in section 1.8, "Q because P" can express either an argument or an explanation.

Of course, some proposed explanations are better than others. The chief criterion for evaluating explanations is relevance. lf the tardy office boy had offered as explanation for his late arrival the fact that there is a war in progress in Afghanistan that would have been a very unsatisfactory explanation, or "no explanation at all." Such a story would have had nothing to do with the case; it would have been irrelevant, because from it the fact to be explained cannot be inferred. The relevance of a proposed explanation, then, corresponds exactly to the cogency of the argument by which the fact to be explained is inferred from the proposed explanation.

Any acceptable explanation must be relevant, but not all stories that are relevant in this sense are acceptable explanations. There are additional criteria for deciding the worth or acceptability of proposed explanations. The most obvious requirement to propose is that the explanation be true.

In the example of the office boy's lateness, the truth of the explanation would not be difficult to ascertain, since the crucial part of his explanation was a particular fact, the traffic accident, of which he claimed to be an eyewitness. But the explanations of science are for the most part general rather than particular. The keystone of Newtonian mechanics is the law of universal gravitation, whose statement is

Every particle of matter in the universe attracts every other particle with a force which is directly proportional to the product of the masses of the particles and inversely proportional to the square of the distance between them.

Newton's law is not directly verifiable in the same way as a bus accident. There is simply no way in which we can inspect all particles of matter in the universe and observe that they do attract each other in precisely the way that Newton's law asserts. Few propositions of science are directly verifiable as true. In fact, none of the important ones is. For the most part these concern unobservable entities, such as molecules and atoms, electrons and protons, chromosomes and genes. Hence the proposed requirement of truth is not directly applicable to most scientific explanations. Before considering more useful criteria for evaluating scientific theories, it will be helpful to compare scientific with unscientific explanations.

Science is supposed to be concerned with facts, and yet in its further reaches we find it apparently committed to highly speculative notions far removed from the possibility of direct experience. How then are scientific explanations to be distinguished from those that are frankly mythological or superstitious? An unscientific "explanation" of the regular motions of the planets was the ancient doctrine that each heavenly body was the abode of a separate "intelligence" that controlled its movement. Engine failures may be unscientifically explained as the work of mischievous "gremlins," and diseases may be unscientifically explained as the result of evil spirits that have invaded the body. We sense the unsatisfactoriness of these explanations, but as far as direct verifiability is concerned, a modern scientific theory about subatomic particles is not greatly different from an unscientific myth. One can no more see or touch a Newtonian "particle" or an electron than see or touch an "intelligence" or an evil spirit. What, then, are the differences between scientific and unscientific explanations?

There are two important and closely related differences. The first significant difference lies in the attitudes taken toward the explanations in question. The typical attitude of one who really accepts an unscientific explanation is dogmatic. The unscientific explanation is regarded as being absolutely true and beyond all possibility of improvement or correction. During the Middle Ages, the word of Aristotle was the ultimate authority to which scholars commonly appealed when attempting to decide questions of fact. However empirically and open mindedly Aristotle himself may have arrived at his views, they were accepted by some medieval scholars in a completely different and unscientific spirit. One of those scholars to whom Galileo offered his telescope to view the newly discovered moons of Jupiter declined to look, being convinced that none could possibly be seen because no mention of them could be found in Aristotle's treatise on astronomy! Because unscientific beliefs are absolute and final, within the framework of any such doctrine or dogma there can be no rational method of ever considering the question of its truth. Scientists' attitude toward their explanations is altogether different. Every explanation in science is put forward tentatively and provisionally. Any proposed explanation is regarded as a mere hypothesis, more or less probable on the basis of the available facts or relevant evidence.

The vocabulary of scientists is sometimes misleading on this point. When what was first suggested as a "hypothesis" becomes well confirmed, it is frequently elevated to the position of a "theory." And when, on the basis of a great mass of evidence, it achieves well nigh universal acceptance, it is promoted to the lofty status of a "law." This terminology is not always strictly adhered to: Newton's discovery is still called the "law of gravitation," whereas Einstein's contribution, which supersedes or at least improves on Newton's, is referred to as the "theory of relativity." The vocabulary of "hypothesis," "theory," and "law" is unfortunate, since it obscures the important fact that all of the general propositions of science are regarded as hypotheses, never as dogmas.

The second and more fundamental difference between scientific and unscientific explanations lies in the basis for accepting or rejecting the view in question. Many unscientific views are mere prejudices that their adherents could scarcely give any reason for holding. Since they are regarded as "certain," however, any challenge or question is likely to be regarded as an affront and met with abuse. If those who accept an unscientific explanation can be persuaded to discuss the basis for its acceptance, there are only a few grounds on which they will attempt to "defend" it. It is said to be true because "we've always believed it" or because "everyone knows it." These all too familiar phrases express appeals to tradition or popularity rather than evidence. Or a questioned dogma may be defended on the grounds of revelation or authority. The absolute truth of their religious creeds and the falsehood of all others have been revealed from on high, at various times, to Moses, to Paul, to Mohammed, to Joseph Smith, and to many others. That there are rival traditions, conflicting authorities, and revelations that contradict one another does not disturb those who Nave embraced an absolute creed. In general, unscientific beliefs are held independently of anything we should regard as evidence in their favor. Because they are absolute, questions of evidence for them are regarded as having little or no importance.

The case is quite different in the realm of science. Since every scientific explanation is regarded as a hypothesis, it is thought worthy of acceptance only to the extent that there is evidence for it. As a hypothesis, the question of its truth or falsehood is open, and there is continual search for more and more evidence to decide that question. The term "evidence" as used here refers ultimately to experience; sensible evidence is the ultimate court of appeal in verifying scientific propositions. Science is empirical in holding that sense experience is the test of truth for all its pronouncements. Consequently, it is of the essence of a scientific proposition that it be capable of being tested by observation.

Not all propositions can be tested directly, but some can. To decide the truth or falsehood of the proposition that it is now raining outside, we need only glance out the window. To tell whether a traffic light shows green or red, all we have to do is to look at it. But the propositions offered by scientists as explanatory hypotheses are not of this type. Such general propositions as Newton's laws or Einstein's theory are not directly testable in this fashion. They can, however, be tested indirectly. The indirect method of testing the truth of a proposition is familiar to all of us, though we may not be familiar with this name for it. For example, if his employer had been suspicious of the office boy's explanation of his tardiness, she might have checked up on it by telephoning the bus company to find out whether an accident had really happened to the seven thirty bus. lf the bus company's report confirmed the boy's story, this would serve to dispel the employer's suspicions; whereas lf the bus company denied that an accident had occurred, it probably would convince the employer that her office boy's story was false. This inquiry would constitute an indirect test of the office boy's explanation.

Indirect testing or indirect verification consists of two parts. First, one deduces from the proposition to be tested one or more other propositions capable of being tested directly. Then these conclusions are tested, and are found to be either true or false. lf the conclusions are false, any proposition that implies them must be false also. On the other hand lf the conclusions are true, that provides some evidence for the truth of the proposition being tested, which is thus confirmed indirectly.

But note that indirect testing is never demonstrative or certain. In order to deduce directly testable conclusions from a proposition, one usually requires additional premisses. The conclusion that the bus company will confirm that its seven thirty bus had an accident does not follow validly from the proposition that the seven thirty bus did have an accident. Additional premisses are needed; for example, that all accidents get reported to the company's office, that the reports are not mislaid or forgotten, and that the company does not make a policy of "covering up" its accidents. So the bus company's denying that an accident occurred would not prove the office boy's story to be false, because the discrepancy might be due to the falsehood of one of the other premisses mentioned. In this case, for example, the accident report may indeed have been mislaid. Those other premisses, however, ordinarily have such a high degree of probability that a negative reply on the part of the bus company would render the office boy's story very doubtful indeed.

Similarly, establishing the truth of a conclusion does not demonstrate the truth of the premisses from which it was deduced. We know very well that a valid deductive argument may Nave a true conclusion even though its premisses are not all true. In the present example, the bus company might confirm that an accident happened to the seven thirty bus because of some mistake in their records, even though no accident had occurred. So the inferred conclusion might be true even though the premisses from which it was deduced were not. In the usual case, though, that is highly unlikely; so a successful or affirmative testing of a conclusion serves to corroborate the premisses from which it was deduced.

Every proposition, scientific or unscientific, that is a relevant explanation for any observable fact has some evidence in its favor; namely, the fact to which it is relevant. Thus the regular motions of the planets must be conceded to constitute evidence for the (unscientific) theory that the planets are inhabited by "intelligences" that cause them to move in just the orbits that are observed. The motions themselves are as much evidence for that myth as they are for Newton's or Einstein's theories. The difference lies in the fact that that is the only evidence for the unscientific hypothesis. Absolutely no other directly testable propositions can be deduced from the myth. On the other hand, a very large number of directly testable propositions can be deduced from the scientific explanations mentioned. Here, then, is the difference between scientific and unscientific explanations. A scientific explanation for a given fact will have directly testable propositions deducible from it, other than the one stating the fact to be explained. But an unscientific explanation will have no other directly testable propositions deducible from it. It is of the essence of a scientific proposition to be empirically verifiable.

We have been using the term "scientific explanation" in a quite general sense. As here defined, an explanation may be scientific even though it is not a part of one of the various special sciences like physics or psychology. Thus the office boy's explanation of his tardiness would be classified as a scientific one, for it is testable, even if only indirectly. But had he offered as explanation the proposition that "God willed him to be late that morning, and God is omnipotent," the explanation would have been unscientific. For although his being late that morning is deducible from the proffered explanation, no other directly testable proposition is, and so the explanation is not even indirectly testable and hence is unscientific.

15.3 Evaluating scientific explanations

How are scientific explanations to be evaluated? That is, how are they to be judged as good or bad, or at least as better or worse? This question is important, because there is often more than a single scientific explanation for one and the same fact. A person's abrupt behavior may be explained either by the hypothesis that the person is shy or by the hypothesis that the person is unfriendly. In a criminal investigation, two different and incompatible hypotheses about the identity of the criminal may equally well account for the known facts. In the realm of science proper, that an object expands when heated is explained both by the caloric and the kinetic theory of heat. The caloric theory regards heat as an invisible weightless fluid, called "caloric," with the power of penetrating, expanding, and dissolving bodies, or dissipating them in vapor. The kinetic theory, on the other hand, regards the heat of a body as consisting of random motions of the molecules of which the body is composed. These are alternative scientific explanations that serve equally well to explain some of the phenomena of thermal expansion. They cannot both be true, however, and the problem is to evaluate or choose between them.

What is wanted here is a list of conditions that a good hypothesis may be expected to fulfill. Such a list of conditions will not provide a recipe with which anyone can construct good hypotheses. There has never been a set of rules laid down for the invention or discovery of hypotheses, and probably there never will be such a set of rides, because devising hypotheses is the creative side of the scientific enterprise. Creativity is a function of talent and imagination, and cannot be reduced to a mechanical process. A great scientific hypothesis, with wide explanatory powers such as those of Newton or of Einstein, is a product of creative genius, just as is a great work of art. But although there is no formula for discovering new hypotheses, there are certain rules to which acceptable hypotheses may be expected to conform. These can be regarded as the criteria for evaluating hypotheses.

Five criteria are commonly used in judging the worth or acceptability of hypotheses:

1. Relevance

2. Testability

3. Compatibility with previously well established hypotheses

4. Predictive or explanatory power

5. Simplicity

Although the first two of these have already been introduced, we will review each of these criteria here.

1. Relevance

An hypothesis is never proposed for its own sake, but is always intended as an explanation of some fact or other. Therefore it must be relevant to the fact it is intended to explain. That is, the fact in question must be deducible from the proposed hypothesis ― either from the hypothesis alone, or from it together with certain causal laws that may be presumed to have already been established as highly probable, or from these together with certain assumptions about particular initial conditions. A hypothesis that is not relevant to the fact it is intended to explain simply does not explain it, and can only be regarded as having failed to fulfill its intended function. A good hypothesis must be relevant.

2. Testability

The chief distinguishing characteristic of scientific hypotheses (as contrasted with unscientific ones) is that they are testable. That is, there must be the possibility of making observations that tend to confirm or disprove any scientific hypothesis. It need not be directly testable, of course. As has already been observed, most of the really important scientific hypotheses are formulated in terms of such unobservable entities as electrons or electromagnetic waves. But there must be some way of getting from statements about such unobservables to statements about directly observable entities such as tables and chairs, or pointer readings, or lines on a photographic plate. Every scientific hypothesis, in other words, must be connected in some way with empirical data or facts of experience.

3. Compatibility with Previously Well established Hypotheses

The requirement that an acceptable hypothesis must be compatible or consistent with other hypotheses that have already been well confirmed is eminently reasonable. Science, in seeking to encompass more and more facts, aims at achieving a system of explanatory hypotheses. Of course such a system must be self consistent, for no self contradictory set of propositions could possibly be true ― or even intelligible. Ideally, the way in which scientists hope to make progress is by gradually expanding their hypotheses to comprehend more and more facts. For such progress to be made, each new hypothesis must be consistent with those already confirmed. Thus Leverrier's hypothesis that there was an additional but not yet charted planet beyond the orbit of Uranus was perfectly consistent with the main body of accepted astronomical theory. The discovery of the planet Neptune, in 1846, was a product of that hypothesis as recounted in section 14.2(4). If there is to be orderly progress in scientific inquiry, a new theory must it with older theories.

The importance of this third criterion is sometimes overestimated. Although the ideal of science may be the gradual growth of theoretical knowledge by the addition of one new hypothesis after another, the actual history of scientific progress has not always followed that pattern. Many of the most important new hypotheses have been inconsistent with older theories and have in fact replaced them rather than fitted in with them. Einstein's relativity theory was of that sort, shattering many of the preconceptions of the older Newtonian theory. The phenomenon of radioactivity, first observed during the last decade of the nineteenth century, led to the overthrow ― or at least the modification ― of many cherished theories that had almost achieved the status of absolutes. One of these was the principle of the conservation of matter, which asserted that matter could be neither created nor destroyed. The hypothesis that radium atoms undergo spontaneous disintegration was inconsistent with that old, established principle ― but the older principle was eventually relinquished in favor of the newer hypothesis.

Theories in science are not abandoned quickly, or without resistance, in favor of newer and shinier ones. Indeed, older theories are not so much abandoned as corrected. Einstein himself always insisted that his own work was a modification rather than a rejection of Newton's. The principle of the conservation of matter was modified by being absorbed into the more comprehensive principle of the conservation of mass energy. Every established theory has been established through having proved adequate to explain a considerable mass of data, of observed facts. And it cannot be dethroned or discredited by any new hypothesis unless that new hypothesis can account for the same facts as well or even better. There is nothing capricious about the development of science. Every change represents an improvement, a more comprehensive and thus more adequate explanation of the way in which the world manifests itself in our experience. Where inconsistencies occur between hypotheses, the greater age of one does not automatically prove it to be correct and the newer one wrong. The presumption is in favor of the older one if it has already been extensively confirmed. But if the new one in conflict with it also receives extensive confirmation, considerations of age or priority are irrelevant. Where there is a conflict between two hypotheses, we must turn to the observable facts to decide between them. The ultimate court of appeal in deciding between rival hypotheses is experience.

Our third criterion, compatibility with previous well established hypotheses, comes to this: The totality of hypotheses accepted at any one time should be consistent with each other,[7] and ― other things being equal ― of two new hypotheses, the one that fits in better with the accepted body of scientific theory is to be preferred. The question of what is involved in "other things being equal" takes us directly to our fourth criterion.

4. Predictive or Explanatory Power

By the predictive or explanatory power of a hypothesis is meant the range of observable facts that can be deduced from it. This criterion is related to, but different from, that of testability. A hypothesis is testable if some observable fact is deducible from it. If one of two testable hypotheses has a greater number of observable facts deducible from it than from the other, then it is said to have greater predictive or explanatory power. Thus, Newton's hypothesis of universal gravitation, when joined together with his three laws of motion, had greater predictive power than did either Kepler's or Galileo's hypotheses, because all observable consequences of the last two were also consequences of the former, and the former had many more besides. An observable fact that can be deduced from a given hypothesis is said to be explained by it and also can be said to be predicted by it. The greater the predictive power of a hypothesis the more it explains, and the better it contributes to our understanding of the phenomena with which it is concerned.

Our fourth criterion has a negative side that is of crucial importance. If a hypothesis is inconsistent with any well attested fact of observation, the hypothesis is false and must be rejected. Where two different hypotheses are both relevant to explaining some set of facts and both are testable, and both are compatible with the whole body of already established scientific theory, it may be possible to choose between them by deducing from them incompatible propositions that are directly testable. If H₁ and H₂, two different hypotheses, entail incompatible consequences, it may be possible to set up a crucial experiment to decide between them. Thus if H₁ entails that under circumstance C phenomenon P will occur, while H₂ entails that under circumstance C phenomenon P will not occur, then all we need do to decide between H₁ and H₂ is to produce circumstance C and observe the presence or absence of phenomenon P. If P occurs, this is evidence for H₁ and against H₂, while if P does not occur, this is evidence against H₁ and for H₂.

This kind of crucial experiment to decide between rival hypotheses may not be easy to carry out, for the required circumstance C may be difficult or impossible to produce. Thus the decision between Newtonian theory and Einstein's general theory of relativity had to await a total eclipse of the sun ― a situation or circumstance clearly beyond our power to produce. In other cases, the crucial experiment may have to await the development of new instruments, either for the production of the required circumstances, or for the observation or measurement of the predicted phenomenon. Thus proponents of rival astronomical hypotheses must often bide their time while they await the construction of new and more powerful telescopes. The topic of crucial experiments will be discussed further in section 15.6.

5. Simplicity

It sometimes happens that two rival hypotheses satisfy the first four criteria equally well. Historically, the most important pair of such hypotheses were those of Ptolemy (ca. 127 151) and Copernicus (1473 1543); both intended to explain all of the then known data of astronomy. According to the Ptolemaic theory, the earth is the center of the universe, and the heavenly bodies move about it in paths that require a very complicated geometry to describe: the orbits were supposed to be perfect circles on whose circumference many smaller circles, or epicycles, accounted for the actually observed, apparently erratic movements of the planets. Ptolemy's theory was relevant, testable, and compatible with previously well established hypotheses, satisfying the first three criteria perfectly. According to the Copernican theory, the sun rather than the earth is at the center, and the earth itself moves around the sun along with the other planets. Copernicus's theory too satisfied the first three criteria perfectly. With respect to the fourth criterion, that of predictive power, there was not a great deal of difference between the two theories. But with respect to the fifth criterion there was a very significant difference between the two rival hypotheses. Although both required the clumsy device of epicycles to account for some of the observed positions of the various heavenly bodies, fewer such epicycles were required within the Copernican theory. The Copernican system was therefore much simpler, and this simplicity contributed greatly to its acceptance by later astronomers.

There are some today who believe that extraterrestrial beings, ETs, are lurking right now in our solar system, hidden by cloaking technology that renders them undetectable while they wait patiently for our society to mature. In response to this position one contemporary astronomer writes:

When scientists are confronted by multiple explanations for a phenomenon, they generally apply Occam's razor: Accept the simplest explanation with the fewest assumptions and reject those that are more fantastic and convoluted. Perhaps the galaxy is bustling with life and civilizations. But the simplest explanation [of the lack of success in all previous searches for extraterrestrials] given the evidence in hand, points in the direction that we share the galaxy with few others, or none at all.[8]

The criterion of simplicity is a "natural" one to invoke. In ordinary life as well as in science, we are inclined to accept the simplest theory that fits all the available facts. In a criminal trial, for example, the prosecution seeks to develop a plausible hypothesis that includes the guilt of the accused and that fits in with all the available evidence, while the defense seeks to develop an alternative plausible hypothesis that also fits all of the available evidence but that includes the innocence of the accused. Both sides may succeed. The case may then be decided ― perhaps ought to be decided ― in favor of the hypothesis that is simpler, more natural.

But "simplicity" is a very difficult term to define. Not all controversies are as straightforward as the Ptolemaic Copernican one, in which greater simplicity consisted merely in requiring a smaller number of epicycles. Of two competing theories, one may rely on mathematical equations that are simpler, while the other may be simpler in relying on a smaller number of supposed entities. Simplicity may be of different kinds. Ir is tempting to fall back on "naturalness," but that term also may prove highly deceptive. Many persons would say, for example, that it seems more "natural" to believe that the apparently unmoving earth is still and that the apparently moving sun really does move around us. Simplicity is an important criterion, often a decisive one, but it is difficult to formulate and not always easy to apply.

15.4 Seven stages of scientific investigation

We are now in a position to describe the general pattern of scientific research. This pattern may be broken down into seven steps, or stages. These can be readily distinguished in the abstract but they are by no means always sharply distinct in practice, as they interpenetrate and blend in many contexts. To help make the general pattern clear, the seven stages will first be briefly set forth in general terms. Then, with this pattern of inquiry in mind, we will examine an extended illustration of scientific investigation that exemplifies the several stages.

1. Identifying the Problem

Scientific investigation begins when the investigator is confronted with something that appears to need explanation. A problem may be characterized as a fact or a group of facts for which we have no acceptable explanation. The detective, for example, confronts a crime; his problem is to solve it; that is, to identify and prove the guilt of the perpetrator. Sometimes ― as in Conan Doyle's stories of the great detective Sherlock Holmes, the problem may be (at least at the outset) associated with some peculiar event or circumstance that is not yet a crime. Scientists may on occasion begin an investigation with their problem sharply identified; or they may come gradually to discover the inconsistencies or peculiarities that evolve into a specifiable problem.

No one ― not even Sherlock Holmes or Albert Einstein ― can engage in profound thought unless there is something to think about. Even a genius must have been presented with a problem before he, or she, can solve it. All reflective thinking ― and this term includes a wide range of activities from criminal investigation to abstract thinking in physics and mathematics ― is problem solving activity, as John Dewey and many other modern philosophers have rightly insisted. The problem must be recognized, at least in some vague form, before the scientist can go to work.

2. Selecting Preliminary Hypotheses

Any systematic reflection about the problem at hand, even the most tentative consideration of alternative explanations, requires some preliminary theorizing. A final judgment normally will not be reached in the first attempt at solution, but some theorizing is required in order to know what sort of evidence needs to be collected, and where or how it might best be sought. The detective examines the scene of the crime, interviews suspects, and collects clues ― but bare facts are not clues. Clues become meaningful only if they can be fitted into some pattern that is coherent, even one that is rough and tentative.

So too the scientist begins the collection of evidence with some preliminary hypothesis about the nature of the explanation sought. Some previous knowledge must be relied upon, of course; science does not begin from absolutely nothing, or go on in a vacuum ― and indeed there must have been some prior beliefs if the facts to be explained appear problematic.

In the case of any serious problem, there are too many relevant facts, too much data in the world for anyone to collect it all. Some matters will be noticed and attended to, others not. The most patient and thorough investigator must pick carefully among all the facts revealed, choosing which to study and which to set aside. This requires some working hypothesis for which, or against which, relevant data may be collected. That hypothesis need not be a complete theory, of course ― but at least the outline of a theory must be there. If it were not, the investigator could not determine which facts, from the totality of facts, to sift and select. However incomplete and tentative, a preliminary hypothesis is needed before any serious inquiry can begin.

3. Collecting Additional Facts

The fact or facts that initially seemed puzzling are generally too meager to suggest a wholly satisfactory explanation for themselves; if that were not the case, those facts are unlikely to have appeared problematic. But, especially to a scientist who is familiar with facts or circumstances of that general kind (say celestial, or sociological, or historical phenomena), the original problem will suggest a preliminary hypothesis that can lead to the search for additional relevant facts. This additional evidence, it is hoped, will serve as clues, leads, suggestions pointing to a fuller and more nearly adequate solution. This task of collecting evidence is often arduous and time consuming, and very frequently it is disappointing and frustrating. Good science is hard work. But the labor of this collection process constitutes the substance of much scientific work.

Of course, steps 2 and 3 are not fully separable in real life science; rather, they are intimately connected and interdependent. We require some preliminary hypothesis In order to begin the collection of evidence, true; but the process of gathering evidence by using that working hypothesis becomes, at the same time, the process of adjusting and refining the hypothesis itself, which then guides the further search . . . leading perhaps to new findings . . . which suggest yet more refined hypotheses . . . and so on and on.

4. Formulating the Explanatory Hypothesis

In any successful investigation that point sooner or later will be reached at which the investigator the scientist, the detective, perhaps some ordinary person ― will come to believe that all the facts needed for solving the original problem are in hand. The pieces of the puzzle ― more likely the chunks, each consisting of smaller pieces ― are before him or her, and the task becomes that of assembling them in such a way as to make sense of the whole. The endproduct of such thinking, lf it is successful, is some hypothesis that accounts for all the data, both the original set of facts that created the problem, and the additional facts to which the preliminary hypotheses pointed.

There is no mechanical way of arriving at this overarching theory. The actual discovery, or invention, of a truly explanatory hypothesis is a process of creation, one in which imagination as well as knowledge is involved. Some investigators, such as Sherlock Holmes and Albert Einstein, exhibit genius in this process of "reasoning backward" to the explanation of existing phenomena.[9] But every successful scientist must undertake this challenging task of intellectual integration: constructing and formulating the final hypothesis that explains the problematic facts by which the investigation was provoked.

5. Deducing Further Consequences

We have seen that predictive power is one of the criteria with which explanations may be appraised. A really fruitful hypothesis will explain not only the facts that originally inspired it, but many others as well. The good hypothesis points beyond the initial facts to new and different facts whose very existence may not earlier have been suspected. lf those facts are verified, that verification tends to confirm (but of course does not prove with certainty) the hypothesis that led to them.

A splendid illustration of such prediction is exhibited by the cosmological theory known as "the Big Bang." If, as this theory holds, the present universe began with one extraordinary explosive event, the initial fireball would have been smooth and homogenous, lacking all structure. By contrast, the present-day universe has a great deal of structure, is very lumpy, its visible matter plainly clumped into galaxies, clusters of galaxies, and so forth. Such structure was, we know, essential for the origin and evolution of life. But when and how did this structure arise? By observing the very most distant objects in an expanding universe, astronomers can "look back" in time. They must eventually find, through those observations, evidence of the seeds of present structure. lf such early structure is not detectable by the most sensitive instruments, the Big Bang theory would appear to be indefensible. lf such structure is detected, the Big Bang theory is confirmed, though of course not proved.

6. Testing the Consequences

The predictions made on the basis of the explanatory hypothesis must be tested, and may require various means for their testing. Some predictions, such as those made by Sherlock Holmes in many of Arthur Conan Doyle's great detective stories, may be tested directly. Will the bank robbers break into the vault, as predicted by Holmes on the basis of his hypothesis in "The Adventure of the Red Headed League"― We wait for them, and they do. Will Dr. Roylott slip a venomous snake through the dummy ventilator, as Holmes predicts in "The Adventure of the Speckled Band"? We watch from hiding, and he does. Holmes's explanatory theories were directly tested and confirmed.

Scientific theories, as we have seen, normally cannot be tested by such straightforward observation. The structure of the early universe cannot be directly observed. But if there were some such early structure, there would have to be irregularity, unevenness in the background radiation currently encountered that stems from that early time. It is possible, in principle, to measure that background microwave radiation, and in this way to determine, indirectly, whether there were such irregularities very shortly after the supposed Big Bang. A very few years ago, a satellite was designed that would detect those irregularities if they were present. The observations to be made through that instrument ― the Cosmic Background Explorer [COBE] satellite ― would be critical for the Big Bang theory. If the long sought evidence of early structure in the universe were not eventually detected, the Big Bang account of the expanding universe would have to be considered seriously doubtful. But in the spring of 1992 the predicted irregularities, surviving from the earliest time to which astronomers can look back, were indeed detected and measured by COBE. This successful test, although of course it did not prove that theory correct, did confirm the Big Bang theory impressively.

7. Applying the Theory

Through science we aim to explain the phenomena we encounter, but we aim also to control those phenomena to our advantage. The abstract theories of Newton and Einstein have played a central role in the modern exploration of our solar system. But suppose, to take an example of a very different kind, that the problem confronted is some disease, and the explanatory hypothesis devised is that the disease is caused by certain specified microbial agents ― say, some previously undetected bacteria. Suppose that this theory has been tested by infecting mice or other rodents with those bacteria, and that such tests strongly confirm the explanatory hypothesis by producing, in the animal subjects, the very same disease. We will seek to apply that theory in clinical medicine, of course, and that would be done (first in experimental human groups, later as a matter of routine medical care) by eliminating those bacteria from patients suffering from that disease, thereby curing the disease itself. In just this way we have learned how to combat, and in some cases even to eliminate entirely, many terrible human diseases. We seek to understand our world through science. But through science we want also to exert some measure of control over the hazards the world presents.

Exercises

1. Take some detective story, and analyze its structure in terms of the seven steps discussed in the preceding sections.

2. Find an account in a popular or semipopular book on science of some specific line of research, and analyze its structure in terms of the seven steps discussed in this preceding section.

15.5 Scientists in action: the pattern of scientific investigation

The pattern that pervades all scientific inquiry is expressible in terms of the seven steps explained in the preceding section. Of course, the methods of science are not confined to professional scientists; anyone may be said to be proceeding scientifically who follows the general pattern of reasoning from observable facts and other evidence to conclusions that can be tested by experience. The skilled detective is a scientist in this sense, as are most of us at times. We are now in a position to examine an extended illustration of that pattern of rational inquiry; we follow contemporary scientists in their recent quest for the solution of the structure of deoxyribonucleic acid ― DNA.[10]

1. The Problem. All living things begin from a single cell, and all living things reproduce their kind; therefore the characteristics that plants and animals inherit must somehow be buried in their first cell. But where? How is the genetic message conveyed from generation to generation, and why do the parts of each developing organism take the complicated forms that they do? This deep and puzzling problem ― solving "the secret of life" ― became an obsession of many scientists, cooperating and competing, during the middle decades of the 20th century. The quest for the gene is one of the most exciting chapters in the recent history of science.

The solution must lie in one of the four categories of substances that make up living cells: (1) fats (lipids); (2) sugars and starches (polysaccharides); (3) proteins; and (4) nucleic acids. The first two had been eliminated definitively long before the present inquiry began. The fourth, nucleic acids, whose chemical elements were known, were also known to be quite simply constructed, their parts appearing in fixed and repeated order. One of these parts is a sugar called ribose; one of the omnipresent nucleic acids contained sugars with one oxygen atom missing ― and was therefore called deoxy ribo nucleic acid, or DNA. Ir was widely believed at that time that DNA was a "stupid" substance, no more than a structural stiffener in cells ― like the cardboard that preserves the shape of a new shirt ― and thus not a candidate to be the stuff of which genes are made.

If DNA does not carry the hereditary message, that message must be conveyed through some protein not yet identified. But there was good evidence on hand, by 1944, that whatever carried the genetic message was not likely to be a protein. Yet when the alpha helix, a key structural element in proteins, was discovered by Linus Pauling in 1949, using a technique involving detection with X rays, proteins again became exciting targets. Moreover, the extraordinary complexity of genetic messages ― the enormous detail and specificity that had to be conveyed from generation to generation ― persuaded many scientists that the secret of the gene could lie only in some large and very complicated protein molecules. In that direction, therefore, the hunt for the gene was widely pressed. Even were that direction correct, there are bewilderingly many proteins; but in any case, the search for the gene among them met with no success.

2. Preliminary hypotheses. When James Watson and Francis Crick began their pursuit of the gene in 1951, at the Cavendish Laboratory in Cambridge, England, their data were confused and incomplete. Their puzzlement was magnified by inconsistencies among beliefs and theories widely accepted; if the observations and eliminative reasoning of Oswald Avery in 1944 were reliable, the search for the gene among the proteins was destined to fail. In that case, as Watson later wrote, "DNA would have to provide the key."[11] This was the preliminary hypothesis with which Watson and Crick began their research: that the genetic message was somehow carried in the structure of DNA. Two other preliminary hypotheses guided their quest: First, that its structure, as suggested by X ray diffraction pictures taken by Rosalind Franklin, and by Maurice Wilkins (who later shared the Nobel Prize with Watson and Crick) was regular. Watson wrote:

Suddenly I was excited about chemistry . . . . I had worried about the possibility that the gene might be fantastically irregular. Now, however, I knew that genes could crystallize hence they must have a regular structure that could be solved in a straightforward fashion.[12]

Second, they hypothesized that the DNA filaments, in view of their great length and the diffraction pictures of them, probably took the form of a spiral or helix, perhaps similar to the alpha helix that Pauling had earlier found in some proteins.

3. Collecting additional facts. To reconcile these preliminary hypotheses with the known but confusing facts about the constituents of DNA, much more would have to be learned ― some of it buried in the scientific literature, some just being discovered.

Nucleic acids were known to have a long "backbone" consisting of a sugar (ribose) alternating with a phosphate (a grouping of phosphorus with 4 oxygen atoms). At each "knuckle" of this backbone a third molecular unit, called a base, was somehow stuck onto the chain. Each base was one of four known kinds: adenine, guanine, cytosine, and thymine, called by their initials A, G, C, and T. The order of the appearance of these four substances on the backbone was a puzzle, and even how the bases were attached to the backbone was not known. As more data were collected, and preliminary hypotheses were refined, the problem became that of fitting the pieces of DNA together. Each three piece unit in the chain (sugar, plus phosphate, plus one base) was called a nucleotide. How could the nucleotides fit together to form the acid known as DNA? The more general problem ("What is a gene?") that obsessed them had been refined, by Watson and Crick, into this more tractable problem of structure.

Progress at first was very slow. Using super sized models made of cardboard and wire, they tried every chain like configuration they could devise. Some specifiable conditions they knew had to be met: certain water content, certain angle of pitch, certain methods of chemical bonding. Everything had to be consistent with previously discovered facts, recent X ray pictures, and established theories. The four bases (A, G, C, and T) were known to be flat. Watson and Crick tried models in which they were stuck like plates to the inside of the helix backbone, or to the outside, or to each other. The angle of the spiral was adjusted; the theory of bonding to sugar molecules was reexamined. Nothing worked to yield a plausible structure.

4. Formulating the refined explanatory hypothesis. The solution to a large problem commonly relies upon contributions from many different quarters. lt is a cooperative enterprise in great part, but becomes highly competitive at times. Other scientists also were racing toward the solution of the structure of DNA. Wilkins and Franklin were getting better X ray diffraction pictures in their London laboratory. Pauling described what he thought to be the DNA structure as a three chain helix, but Watson and Crick had enough information to realize (with a mixture of disappointment and glee) that Pauling's account contained a fatal error. Once Pauling's manuscript was published, Watson wrote:

it would be only a matter of days before the error would be discovered. We had anywhere up to six weeks before Linus again was in full time pursuit of DNA . . . . I let Francis [Crick] buy me a whiskey. Linus had not yet won his Nobel.[13]

The refined hypothesis that would solve the problem had to account for two different powers of the gene. (1) how is the enormous detail in living structures conveyed in the genetic message?; and (2) how does the genetic message duplicate itself in successive generations? Needed was a threedimensional structure, consistent with known facts and theories, that could provide the coding for all the detail of life and that could replicate itself in generation after generation.

The research of Erwin Chargaff, at Columbia University, helped to put them on the right track. Chargaff had made a striking discovery: In all tested samples of DNA, the relative quantities of the four bases ― adenine, guanine, cytosine, and thymine ― were fixed. Two of these, A and G, are called purines; the other two, C and T, are called pyrimidines. Chargaff had proved that the number of A molecules always is equal to the number of T molecules, and the number of G molecules always equal to the number of C molecules. The quantity of purines (A and G) always is identical to the quantity of pyrimidines (T and C). But no one could explain why this was so.

Through calculation and the manipulation of models, Crick determined that the structures of A and T were such that they would naturally stick to one another; and that the forces which naturally would attract G and C to one another could also be specified. Then, if the DNA chain were so constructed that for every A there was a fitting T, and for every G a fitting C, the chain ― if it were split down the middle ― would provide an elegant system of self replication: each side of the chain could be viewed as a lock to which the other side was a key; each would be a template for the construction of a new, matching key. And if the chain of matching pairs of bases was very long, their order and number might explain the required genetic coding of detail. The solution, they now hypothesized, would be some sort of double helix. They tried models with the backbone in the center and the bases sticking out; they tried models with the backbone on the outside and the bases projecting inward. Still no success. Yet they believed they were close to elucidation of the structure of DNA.

Could the remaining difficulty lie in the accepted theory of how the bases (A, G, T, and C) could bond to one another? If the accepted theory were flawed, and could be replaced by a new account of the chemical bonding of the bases to one another, a workable model of the double helix might be devised. That possibility was explored. The pieces of the puzzle began to come together in Watson's mind:

When I got to our still empty office the following morning, I quickly cleared away the papers from my desk top so that I would have a large flat surface on which to form pairs of bases held together by hydrogen bonds.[14]

Still no success. With an A on one side matching an A on the other, and a C matching a C, and so on, the bases pointed inward and tied to one another across the hollow center of the chain, they simply could not be fitted into a double spiral.

And then at last, struggling to revise this structure so as to achieve compatibility among all the elements of the theory, Watson was able to formulate a fully refined hypothesis that proved to be correct: the DNA molecule was indeed a double helix, in which the bases did indeed extend inward ― but the matching of the pairs of bases was complementary: every A fitted to a T, every G fitted to a C.

I . . . began shifting the bases in and out of other various airing possibilities. Suddenly I became aware that an adenine thymine pair held together by two hydrogen bonds was identical in shape to a guanine cytosine pair held together by at least two hydrogen bonds. All the hydrogen bonds seem to form naturally; no fudging was required to make the two types of base pairs identical in shape . . . .

[W]e now had the answer to the riddle of why the number of purine residues (A and G) exactly equaled the number of pyrimidine residues (C and T). Two irregular sequences of bases could be regularly packed in the center of a [double] helix . . . . Adenine would always pair with thymine, while guanine could pair only with cytosine . . . . the base sequences of the two intertwined chains were complementary to each other. Given the base sequence of one chain, that of its partner was automatically determined.

Conceptually, it was thus very easy to visualize how a single chain could be the template for the synthesis of a chain with the complementary sequence.[15] [See Figure 15 1.]

When Francis Crick, at lunch that day at the Eagle Pub in Cambridge, told everyone within hearing that "we had found the secret of life," Watson wrote, "I felt a little queasy.[16]

Figure 15-1. A schematic illustration of the complementary double helix.
The two sugar-phosphate backbones twist about on the outside. The flat bases, in pairs ― A always bonding with T, C always bonding with G ― make up the core. The structure resembles a spiral staircase, with the pairs of bases forming the steps. (From J. D. Watson,The Double Helix, p. 130.) Courtesy of the Copyright holder, Gunther S. Stent.

5 and 6. Deducing and testing consequences. The hypothesis now having been formulated, it had next to be confirmed. The first deduction was straightforward: If the double helix proposed by Watson and Crick were indeed a correct account of the structure of DNA, it should be possible to construct a three dimensional model of that double helix in which the bases would fit together internally, and the angles of the spiral as well as all other features of the chain would satisfy the requirements established by earlier X ray pictures and other experiments. That construction quickly was achieved.

Many additional theoretical deductions were made; every test of them proved successful. One such analysis now made understandable certain data that had long puzzled and frustrated molecular biologists; the quantity of DNA in the reproductive cells coming from each parent was known to be only half that to be found in an ordinary cell. The reason for this was now clear: If the double helix splits in preparation for reproduction, those split cells from each parent would of course contain just one half the normal quantity of DNA. Evidence for the correctness of the Watson Crick solution to the structure of DNA mounted quickly; soon their hypothesis was very fully confirmed.

7. Applications. A 128-line report by Watson and Crick on the structure of DNA[17] made scientific history; the course of biological science was dramatically and permanently altered. Extensive and powerful applications of this knowledge culminated their achievement. The codes used in DNA sequences became known in subsequent decades; a complete map of the entire human genome, now in the making, will be available to scientists before long. Techniques for cutting and recombining DNA chains have been developed and are now widely used in manufacturing new drugs, vaccines, and synthetic hormones. The applications of recombinant DNA technology ― possible only because the problem of the structure of DNA has at last been solved ― have revolutionized biology and medicine, and are constantly expanding.

15.6 Crucial experiments and ad hoc hypotheses

A. Crucial Experiments

Progress in science is rarely straightforward or easy. Ir would be foolish to suppose that simply by applying the several steps of the hypothetico deductive method to any problem, the solution will be obtained. Solutions ― correct explanatory hypotheses ― often are obscure, may require very elaborate theoretical machinery, and devising the finally correct hypothesis may be exceedingly difficult. The process, far from being mechanical, often requires, in addition to laborious observation and experimentation, deep insight and great creativity.

Once the new hypothesis has been formulated, if it is inconsistent with some previously accepted theory it may be difficult to determine which of the alternative accounts is correct. In some cases two competing hypotheses may be tested by means of what is called a "crucial experiment," an experiment deliberately constructed so as to reveal that one but not the other of the explanations offered is in fact correct. Such crucial experiments, when they can be devised, may prove exciting and highly productive.

For example: The American physicist Albert Michelson and chemist Edward Morley collaborated in 1887 on an experiment to measure the speed of light, and by doing so to put a widely held theory (which they believed to be correct) to a crucial test. It had been long believed that space was filled with a hypothetical substance called "ether" which (supposedly) permitted waves of light to travel in much the same way as air permits sound waves to travel. Either the ether exists, or it does not exist. If it does exist, then the measured speed of light in the direction of Earth's motion should be different from the speed of light determined at right angles to Earth's motion. The experiment had what may be called a "negative" result, and yet because it was a crucial experiment testing a widely accepted theory of that time, it became one of the most famous in the history of physics. No difference could be found in the speed of light moving in the two different directions. This result effectively killed off the long held concept of the ether.[18]

Unfortunately, such powerfully crucial experiments are not always feasible. Alternative observable consequences may not be presently deducible from the alternative hypotheses, or they may be deducible but we may presently lack the ability to arrange circumstances so as to test which of those consequences are manifesting themselves.

There is yet another difficulty involved in the devising of so called crucial experiments which, when understood, brings to light a feature of scientific inquiry deserving emphasis: The consequences of some proposed explanatory hypothesis, which we may wish to test by conducting some crucial experiment, can never be deduced from that hypothesis by itself. We deduce the consequence to be tested by applying the hypothesis in mind, together with other theories which, for this purpose, are assumed to be completely reliable. Those other theories may indeed be completely reliable. But they may not be so, and if they are not, the hypothesis in question may turn out to be correct even though the crucial experiment appeared to disconfirm it. Advancement in science depends upon sets of hypotheses, any one of which may be flawed.

Where hypotheses of a fairly high level of abstractness are involved, no directly testable prediction can be deduced from just a single one of them. A unified group of hypotheses must be used as premisses for the deduction, and if the observed facts are other than those predicted we may conclude that at least one of the hypotheses in the group is shown to be false. But such a conclusion will not Nave established which one is in error. In the preceding account of the discovery of the structure of DNA, for example, there was a point at which Watson and Crick tested the hypothesis that the filaments of nucleic acid were in the form of a double helix, its bases pointed inward ― and found that such an arrangement simply could not be made consistent with all other known facts and accepted theories. Those "known facts and accepted theories" ― the water content, the pitch of the helix, the way in which the bases (adenine, guanine, cystine, and thymine) would bond ― were assumed to be correct as they went about testing their hypothesis. If all of those suppositions had indeed been correct, their double helix could not Nave been the structure of the filament. In the actual case, however, Watson and Crick, having confidence in their hypothesis, came to suspect that the accepted theory describing the ways in which the bases (A, G, C, and T) bond to each other was not entirely correct. By relinquishing that element in the set, and replacing it with a different account that supposed hydrogen bonds instead, the newly hypothesized double helix (with its allied theories) could be confirmed.

Thus, an experiment may be "crucial" in showing the untenability of a group of hypotheses. Such a group usually will contain a considerable number of separate hypotheses, the truth of any one of which can be maintained in the teeth of any experimental result, however apparently unfavorable, by the expedient of rejecting some other hypothesis in the group. This consideration has led some to conclude that no individual hypothesis can ever be subjected to a crucial experiment.

B. Ad Hoc Hypotheses

To this critique it may be objected that an experiment can indeed be crucial in disconfirming a single new hypothesis, because the effort to "save" that hypothesis by conveniently rejecting some other element in the group (suggested as possible just above) is purely ad hoc, a Latin expression meaning literally "for this [special purpose]." There is one sense of the term ad hoc in which all hypotheses are ad hoc, since it makes no sense to speak of a hypothesis that was not devised to account for some antecedently established fact or other. But when used as a term of abuse, ad hoc suggests that the adjustment in the set of hypotheses was made only for the purpose of saving the hypothesis being tested, and that it has no other explanatory power or testable consequences.

No scientific hypothesis is ad hoc in this second sense. That "gremlins are the cause of the breakdown" would be an obviously unscientific explanation if introduced to account for the malfunction of a complicated machine; such hypotheses we rightly ridicule as ad hoc in this negative sense. But in any actual scientific investigation, when a new hypothesis is proposed and an older theory adjusted, it remains to determine whether the adjustment made in the set of hypotheses is indeed ad hoc in that negative sense.

Another illustration from the history of science will help to make this clear. In the 19th century, with the theory of celestial mechanics quite well understood, it became evident to astronomers that the orbits of two of the planets, Uranus and Mercury, were not what accepted theory had predicted they should be. The theory of planetary motion might then have been altered, but in fact it was retained. It was proposed that, consistent with that theory, there existed some as yet undiscovered planets whose gravity was causing the observed anomalies. The resultant prediction by Leverrier in 1845 ― of the orbit of the new planet that might account for the apparent discrepancies in the orbit of Uranus ― was very soon verified by the discovery of the planet Neptune, precisely where it would have had to be to account for those discrepancies.[19] The hypothesis that there was such a planet was certainly not ad hoc in the negative sense, since many consequences could be deduced from that hypothesis which rendered it independently testable.

But in the case of Mercury, the hypothesis that there was another planet (prematurely named "Vulcan") perturbing its orbit could not be confirmed. If a theory supposing some imagined "mercurial forces" had then been introduced to account for the aberrations in Mercury's orbit, forces that accounted for nothing else and could be identified in no other way, such an invention certainly would have been ad hoc. In the actual case, the matter long remained problematic; it was not until the development of the general theory of relativity in 1915 that the observed irregularities in Mercury's orbit could be fully reconciled with other well-established astronomical accounts. The fact that the anomaly in Mercury's orbit could be predicted using the general theory of relativity became one of the most compelling confirmations of that theory. Einstein called it "the most splendid work of my life."[20] Only then had an adequate ― that is, a genuinely theoretical ― explanation of the data been given.

This problem in the history of astronomy points to a third sense, also derogatory, in which the expression ad hoc sometimes is used: to denote a mere descriptive generalization. A descriptive hypothesis that is ad hoc in this third sense will assert only that all facts of a particular sort occur in just some particular kinds of circumstances; but the hypothesis will (like those in the preceding sense) have no explanatory power or scope. The classic example of such an hypothesis is the "Fitzgerald contraction effect," introduced to account for the results of the Michelson Morley experiment on the velocity of light. By affirming that bodies moving at extremely high velocities contract, Fitzgerald did account for the given data, and his hypothesis could be tested by repetitions of that same experiment ― but his "contraction effect" explained nothing else. It was at the time generally held to be ad hoc rather than explanatory, and (as in the matter of the apparent discrepancies in the behavior of Mercury) it was not until relativity theory ― in this case, Einstein's Special Theory of Relativity ― that an adequate theoretical explanation of the experimental results of the Michelson Morley experiment could be given.

We may conclude that it is not only because hypotheses often are ad hoc in the derogatory senses of that term that experiments are never crucial for a single hypothesis. More fundamentally, experiments are never crucial for individual hypotheses because hypotheses are testable only in groups.[21] This limitation serves to illuminate the systematic character of science. Scientific progress consists in building ever more adequate theories so as to account for the enlarging body of observations made and facts experienced. lsolated, particular facts can be of great value too, for the ultimate basis of science is factual. But the structure of science grows not chiefly through the piecemeal collection of bits, but more organically, within the framework of a generally accepted body of theory. The notion that scientific hypotheses or laws are wholly discreet and independent is a naive and outdated view.

Working within such a framework of theory, one that we are not at the time concerned to question, the notion of conducting a "crucial experiment" in order to confirm or disconfirm some hypothesis can still make sense. lf a negative result is obtained ― that is, if some phenomenon fails to occur that had been predicted on the basis of some single dubious hypothesis, taken in tandem with accepted parts of scientific theory ― then the experiment is crucial and that dubious hypothesis may be rejected. But, as we have seen, there is nothing absolute about such a procedure, for even well accepted scientific theories come to be changed in the face of new and contradictory evidence. Science is not monolithic, either in its practices or in its aims.

The lesson to be learned from the preceding discussion is the importance to scientific progress of dragging "hidden assumptions" into the open, so that what had been tacitly assumed may, on occasion, be reconsidered. When a critical assumption is hidden there is no apparent need, and therefore no good occasion, to examine it, and to decide intelligently whether it really is true or false. Progress often is achieved by formulating explicitly an assumption that previously had been hidden, and then scrutinizing and (perhaps) rejecting it.

For example, it seems entirely unproblematic in ordinary life to refer to the occurrence of two events as taking place "at the same time." We commonly assume that events often occur simultaneously. But an important and dramatic advance in science was initiated when Einstein brought this assumption into the open, asking how an observer could determine whether or not two distant events truly occurred at the same time. Ultimately he was led to the conclusion that two events can be simultaneous for some observers but not for others, depending upon their locations and their velocities relative to the events in question. It was this rejection of the assumption of simultaneity which led Einstein to his Special Theory of Relativity, which in turn constituted a tremendous step forward in explaining such phenomena as those revealed by the Michelson Morley experiment. But of course, an assumption must be recognized before it can be challenged. Hence it is enormously important in science to formulate explicitly all of the relevant assumptions at work within any theory, allowing none of them to remain hidden.

There may be no better way of summarizing this account of the methods of science, and to highlight the importance of its systematic advance, than by describing and discussing one of the most extraordinary chapters in the history of science: the observational confirmation, by Galileo Galilei, of the Copernican account of the solar system.

By the early 1600s, the movements of the planets against the backdrop of the fixed stars had been so carefully studied that their apparent movements were predictable. The moon, also much studied, was believed by theologians to be a perfect sphere. The heavenly bodies, deemed flawless in shape and movement, were widely believed to travel in perfect circles around the Earth, which was the center of the world God had created. Galileo had devised, by 1609, a telescope with 20 power magnification, its chief uses being thought at first to be maritime, or as a spyglass that could gain military advantage. With this instrument he observed the heavens, almost by accident, in January of 1610. On the 7th of that month he began a long letter, reporting in detail his observations of the moon and other bodies. He wrote:

I have observed with one of my telescopes . . . the face of the moon, which I have been able to see very near . . . [W]hat is there can be discerned with great distinctness, and in fact it is seen that the moon is most evidently not at all of an even, smooth and regular surface, as a great many people believe of it and of the other heavenly bodies, but on the contrary it is rough and unequal. In short, it is shown to be such that sane reasoning cannot conclude otherwise than that it is full of prominences and cavities similar, but much larger, to the mountains and valleys spread over the Earth's surface . . .[22]

To save the hypothesis that the moon was indeed a perfect sphere, and thus to retain the coherence of the theological account of the heavenly bodies of which that perfection was one element, some of Galileo's critics later proposed the hypothesis ― outrageously ad hoc ― that the apparent cavities and irregularities on the surface of the moon were, in fact, filled in by a celestial substance that was flawless and crystalline, and thus invisible through Galileo's telescope!

More than the moon was examined by Galileo. His letter continued:

And besides the observations of the Moon . . . many fixed stars are seen with the telescope that are not [otherwise] discerned; and only this evening I have seen Jupiter accompanied by three fixed stars, totally invisible [to the naked eye] by their smallness, and the configuration was in this form:[23]

Figure 15-2. A photograph of the letter begun by Galileo on 7 January 1610, on which isa recorded his first monumental observations of the four major satalites of Jupiter, thus confirming the Copernician account of the movement of the celestial bodies. The letter itself was to be sent to the Doge in Venice, with a telescope that Galileo intended to present him. On a draft of that letter which he happened to have in hand, Galileo made the critical notes of his observations, which appear on the bottom half of the sheet. The translation of the bottom half into English appears bellow. (Courtesy of the Special Collections Library, University of Michigan.)

At that point Galileo inserts a sketch that appears here as Figure 15-2, showing the three stars in a straight line, two to the east and one to the west of Jupiter; he reports that they did not extend more than one degree of longitude, but since at that time he supposed them to be fixed stars, their distances from Jupiter and from one another were indicated only very roughly.

On the following day, 8 January 1610, "led by I know not what," Galileo happened to observe Jupiter once again; the earlier positions of those "fixed stars" had fortunately been written down. His letter remained unsent; at the bottom of the sheet he wrote the following note:

On the 8th thus: [he inserts a sketch showing Jupiter and three stars now closer to one another and nearly equidistant from one another, and all three to the west of Jupiter!]

This created a serious theoretical problem for Galileo, since at this time the assumption that the newly discovered stars were fixed had not been seriously doubted. Therefore their appearance on the other side of Jupiter had to be accounted for by Jupiter's movements. On the 8th he adds the note:

It [Jupiter's movement] was therefore direct and not retrograde.

If on the 8th, Jupiter was to the east of all three stars, while the day before, Jupiter had been to the west of two of them, Jupiter must have moved, and moved in a way that was contrary to reliable astronomical calculations! One can imagine Galileo's agitation as he waited for the observations of the following night; could his direct observations and his calculations remain so sharply inconsistent? But on the 9th it was too cloudy to observe. On the 10th, Jupiter apparently had moved back to the west, now apparently obscuring the third star, and the other two were again observed to the east of the planet! On 11 January a similar pattern was observed, but on this night Galileo later wrote,

The star nearer Jupiter was half the size of the other, and very close to the other, whereas the other evenings all three of the said stars appeared of equal size and equally far apart . . .

Clearly, something had to give. From the accepted theories and beliefs a prediction confidently could be drawn, a deduction concerning the movements of Jupiter, which ― if those three new stars were fixed, and Galileo's observations were accurate ― did not take place. One could save the belief that those new stars were fixed by somehow revamping the entire set of astronomical calculations, but these were not in serious doubt; or, one could challenge the accuracy of Galileo's observations ― which is what some of his critics later sought to do, calling his telescope an instrument of the devil. Galileo himself had no doubt about what he had seen, and he grasped quickly which element in the set of accepted hypotheses had to be relinquished, to the great distress of his dogmatic opponents. His note on the observation of the 11th continued:

. . . from which it appears that around Jupiter there are three moving stars invisible to everyone to this time.

And these three moving stars, he later wrote,

revolved round Jupiter in the same manner as Venus and Mercury revolved round the sun.

The observations of the following nights confirmed this revolutionary conclusion, which, together with his earlier observations of the moon, cast serious doubt upon the account of celestial bodies that had been widely and dogmatically affirmed for very many centuries.

On 13 January 1610, Galileo observed a fourth "star," and the four major satellites of Jupiter had been discovered. These observations provided very strong confirmation of the Copernican hypothesis ― an account of the celestial bodies difficult to reconcile with the established theological doctrine of Galileo's time. These four moons of Jupiter (many more have been discovered since) ― Ganymede, lo, Europa, and Callisto ― are appropriately called "the Galilean satellites." Their revolutions about that planet can be readily reconfirmed on a clear night when Jupiter is visible, by anyone with an ordinary pair of binoculars.

15.7 Classification as hypothesis

It could be objected that hypotheses play important roles only in the more advanced sciences, not in those that are relatively less advanced. It might be urged that although explanatory hypotheses are central to such sciences as physics and chemistry, they play no such role ― at least not yet ― in the biological or social sciences. The latter are still in their descriptive phases, and it may be felt that the method of hypothesis is not relevant to the so called descriptive sciences, such as botany or history. This objection is easily answered. An examination of the nature of description will show that description itself is based on, or embodies, hypotheses. Hypotheses are as basic to the various systems of taxonomy or classification in biology as they are in history or any of the other social sciences.

The importance of hypothesis in the science of history may be easily shown, and will be discussed first. Some historians believe that the study of history can reveal the existence of a single cosmic purpose or pattern, either religious or naturalistic, which accounts for or explains the entire course of recorded history. Others deny the existence of any such cosmic design, but insist that the study of history will reveal certain historical laws that explain the actual sequence of past events and can be used to predict the future. On either of these views, the historian seeks explanations that must account for and be confirmed by the recorded events of the past. On either of these views, therefore, history is a theoretical rather than a merely descriptive science, and the role of hypothesis must be admitted as being central to the historian's enterprise.

There is, however, a third group of historians who set themselves what is apparently a more modest goal. According to them, the task of historians is simply to chronicle the past, to set forth a bare description of past events in their chronological order. On this view, it might seem, "scientific" historians have no need of hypotheses, since their concern is with the facts themselves, not with any theories about them.

But past events are not so easily chronicled as this view would have us believe. The past itself simply is not available for this kind of description. What is available are present records and traces of the past. These range all the way from official government archives of the recent past, to epic poems celebrating the exploits of half legendary heroes; from the writings of older historians, to artifacts of bygone eras unearthed in the excavations of archaeologists. These are the only facts available to historians, and from these they must infer the nature of those past events it is their purpose to describe. Not all hypotheses are general; some are particular. The historian's description of the past is a particular hypothesis that is intended to account for present data, and for which the present data constitute evidence.

Historians are detectives on a grand scale.[24] Their methods are the same, and their difficulties too. The evidence is scanty, and much of it has been destroyed ― if not by the bungling local constabulary, then by intervening wars and natural disasters. And just as the criminal may have left false or misleading clues to throw pursuers off the scent, so many present "records" are falsifications of the past they purport to describe; either intentional, as in the case of such forged historical documents as the "Donation of Constantine," or unintentional, as in the writings of early uncritical historians. Just as the detective must use the method of science in formulating and testing hypotheses, so too must the historian. Even those historians who seek to limit themselves to bare descriptions of past events must work with hypotheses: They are theorists in spite of themselves.

Biologists are in a somewhat more favorable position. The facts with which they deal are present, and available for inspection. To describe the flora and fauna of a given region, they need not make elaborate inferences of the sort to which historians are condemned. The data can be perceived directly. Their descriptions of these items are not casual, of course, but systematic. They usually are said to classify plants and animals, rather than merely to describe them. But classification and description are really the same process. To describe a given animal as carnivorous is to classify it as a carnivore; to classify it as a reptile is to describe it as reptilian. To describe any object as having a certain attribute is to classify it as a member of the class of objects having that attribute.

Classification, as generally understood, involves not merely a single division of objects into separate groups, but further subdivision of each group into subgroups or subclasses, and so on. This pattern is familiar to most of us, if not from our studies in school, then probably from playing the old game called "Animal, Vegetable, or Mineral?" more commonly called "Twenty Questions." Classification is a universal need. Primitive peoples were obliged to classify roots and berries as being edible or poisonous, animals as dangerous or harmless, and other tribes as friends or enemies. All people tend to draw distinctions that are of practical importance to them, and to neglect those that play a less immediate role in their affairs. A farmer will classify grains and vegetables carefully and in detail, but may call all flowers "posies," whereas florists will classify their merchandise with the greatest of care but may lump all the farmer's crops together as "produce."

Two basic motives may lead us to classify things. One is practical, the other theoretical. Having only three or four books, one could know them all well and could easily take them all in at a glance, so that there would be no need to classify them. But in a library containing many thousands of volumes, the situation is different. If the books there were not classified, the librarian could not find the ones that were wanted, and from a practical standpoint the collection would be useless. The larger the number of objects, the greater is the need to classify them. A practical purpose of classification is to make large collections accessible. This is especially apparent in the case of libraries, museums, and public records halls of one sort or another.

When we consider the theoretical purpose of classification, we must realize that the adoption of this or that alternative classification scheme is not a matter of truth or falsehood. Objects may be described in different ways, from different points of view. The scheme of classification adopted will depend upon the purpose or interest of the classifier. Books, for example, would be classified differently by a librarian, a bookbinder, and a bibliophile. The librarian would classify them according to their content or subject matter, the bookbinder according to their bindings, and a bibliophile according to their date of printing and perhaps their relative rarity. The possibilities are not thereby exhausted, of course: A book packer would divide books according to their shapes and sizes, and persons with still other interests would classify them differently in the light of those different interests.

Now, what special interest or purpose do scientists have, that can lead them to prefer one scheme of classification to another? The scientist's aim is knowledge, not merely of this or that particular fact for its own sake, but of the general laws to which the facts conform, and of their causal interrelations. One classification scheme is better than another, from the scientist's point of view, to the extent that it is more fruitful in suggesting scientific laws and more helpful in the formulation of explanatory hypotheses.

The theoretical or scientific motivation for classifying objects is the desire to increase our knowledge of them. Increased knowledge of things provides us with further insight into their attributes, their similarities and differences, and their interrelations. A classification scheme made for narrowly practical purposes may tend to obscure important similarities and differences. Thus, a division of animals into "dangerous" and "harmless" will assign the wild boar and the rattlesnake to one class and the domestic pig and the grass snake to the other, calling attention away from what we should today regard as more profound similarities in order to emphasize superficial resemblances. Any scientifically fruitful classification of objects will require a considerable knowledge about them. A slight acquaintance with their more obvious characteristics might lead one to classify the bar with the birds, as flying creatures, and the whale with the fishes, as creatures that live in the sea. More extensive knowledge leads us to classify both bats and whales as mammals, because their being warm blooded, bearing their young alive, and suckling them, are more important characteristics on which to base a classificatory scheme.

A characteristic is important when it serves as a clue to the presence of other characteristics. An important characteristic, from the point of view of science, is one that is causally connected with many other characteristics, and hence relevant to the framing of a maximum number of causal laws and the formulation of very general explanatory hypotheses. That classification scheme is best, then, which is based on the most important characteristics of the objects to be classified. But we do not know in advance which causal laws obtain, and causal laws themselves partake of the nature of hypotheses, as we have emphasized. Therefore any decision as to which classification scheme is best will itself constitute a hypothesis, one that subsequent investigations may lead us to reject. If later investigations reveal other characteristics to be more important ― that is, involved in a greater number of causal laws and explanatory hypotheses ― we can reasonably expect the earlier classification scheme to be rejected in favor of a newer one based upon the more important characteristics.

This view of classification schemes as hypotheses is borne out by the actual role such schemes play in the sciences. Taxonomy is a legitimate, important, and still growing branch of biology, in which some classification schemes, such as that of Linnaeus, have been adopted, used, and subsequently abandoned in favor of better ones, which are themselves in turn subject to modification in light of new data. Classification generally is most important in the early or less developed stages of a science. It need not always diminish in importance as the science develops, however. For example, the standard classification scheme for the elements, as set forth in Mendeleeff's table, is still an important tool for the chemist.

The foregoing account of the uses of classification in the natural sciences suggests a further point of some importance regarding its use in the study of history. That the historian's descriptions of past events are themselves hypotheses based upon present data has already been noted. Yet there is an additional, equally significant role that hypotheses play in the descriptive historian's enterprise. No historical event of any magnitude can be described in complete detail. Even if all of its details could be known, historians could not possibly include them all in their narratives. Life is too short to permit an exhaustive description of anything. Historians must, therefore, describe the past selectively, recording only some of its features. Upon what basis shall they make their selection? Clearly, historians want to include what is significant or important in their descriptions, and to ignore what is insignificant or trivial. The subjective bias of this or that historian may lead him or her to place undue stress on the religious, the economic, the personal, or some other aspect of the historic process. But to the extent that they can make an objective or scientific appraisal, historians will regard those aspects as important which enter into the formulation of causal laws and general explanatory hypotheses. Such appraisals are, of course, subject to correction in the light of further research.

The first Western historian, Herodotus, described a great many aspects of the events he chronicled, personal and cultural as well as political and military. The so called "first scientific historian," Thucydides, restricted himself much more to the political and the military. For a long period of time most historians followed Thucydides, but now the pendulum is swinging in another direction and the economic and cultural aspects of the past are being given greatly increased emphasis. Just as biologists' classification schemes embody their hypotheses as to which characteristics of living things are implicated in a maximum number of causal laws, so historians' decisions to describe past events in terms of one rather than another set of characteristics embody their hypotheses as to which characteristics are causally related to a maximum number of others. Some such hypotheses are required, before historians can even begin to do any systematic describing of the past. It is this hypothetical character of classification and description, whether biological or historical, that leads us to regard hypothesis as the all pervasive method of scientific inquiry.

Exercises

In each of the following passages,

a. What data are to be explained?

b. What hypotheses are proposed to explain them?

c. Evaluate the hypotheses in terms of the criteria presented in Section 15.3.

1. In a stunning discovery of new worlds, two California as, Dr. Geoffrey Marcy and Dr. Paul Butler, today reported the detection of two planets orbiting sun like stars. The temperatures of the planets appear to be warm enough for water to exist in liquid form, a condition conducive to chemical processes that could, just possibly, be producing extraterrestrial life.

   The two newly discovered extra solar planets accompany the stars 70 Virginis, in the constellation Virgo, and 47 Ursa Majoris, in the constellation Ursa Major. They are 35 light years away, relatively close by cosmic standards. They are too small and too dim to be seen against the glare of their parent stars, but their gravitational presence has been definitely established.

   Both stars are analogues of the Sun, astronomers said, with no appreciable differences in size, temperature, or age. The search for other planets has so far been focused on about 120 such Sun like stars within 100 light years of Earth.

   When Dr. Butler saw that the wobble in the motion of the star 70 Virginis fit precisely the orbit of a large planet, he reported, "I was sure of what we had in three minutes, and I almost literally fell out of my chair."

   As far as scientists know, other, smaller planets may also orbit these stars. But current detection techniques may not find them. In an interview, Dr. Marcy said:

   Theorists now have a greater challenge in front of them. They have more than the solar system to explain. Other systems may be more diverse and may contain objects that defy normal classification.

   ― "Hints of Life in Space," The New York Times,
   18 January 1996




2. In the United States, regardless of the way health is measured (mortality, morbidity, symptoms, or subjective evaluation), and regardless of the unit of observation (individuals, city or state averages), years of schooling usually emerges as the most powerful correlate of good health. Michael Grossman, an economist who has done extensive research on this question, has tended to interpret this relationship as evidence that schooling increases the individual's efficiency in producing health, although he recognizes that some causality may run from better health to more schooling. The way schooling contributes to efficiency in producing health has never been made explicit, but Grossman has speculated that persons with more education might choose healthier diets, be more aware of health risks, choose healthier occupations, and use medical care more wisely.

   ― VICTOR R. FUCHS, "The Economics of Health in a Post Industrial
   Society," The Public Interest, Summer 1979




3. The central geographical and climatic characteristic of North Africa and the Mideast is its aridity. A current hypothesis is that there exists a feed back relationship between the plant growth of a marginally arid area and its rainfall. If for some reason ― overgrazing, for example ― the area is partially denuded of growth, its albedo, or reflectivity, will increase. A greater percentage of sunlight is returned to space, the corresponding heat loss is compensated by sinking air motions; and mean cloudiness, and hence mean rainfall, decreases. Then plant growth decreases still further, and a feed back, or vicious circle, mechanism is set in motion.

   ― MORTON G. WURTELE and JEHUDA NEUMANN, "Some Areas for
   International Cooperation in the Geophysical Sciences,"
   Middle East Review, Spring 1978




4. One of the most challenging problems in all of social science has been untangling the environmental and genetic influences of the family on children's intellectual, occupational, and economic attainments. The educational level of parents correlates fairly well with both school achievement and mental ability test scores of their children. This correlation is usually assumed to indicate the strength of the influence of environment on school success, since parents with more years of schooling tend to expect their children to do well in school and create a richer educational environment in the home than do poorly educated parents. If the causal connection runs from the rich family environment to the academic ability level of the child, then it makes sense to try to induce all parents to provide more educative environments, as a way of improving school performance of educationally disadvantaged children.

   If mental ability is to some extent inherited, however, a different set of causal linkages may be involved: Parents possessing high levels of mental ability will tend to spend more years in school than others do, will pass on some of their ability to their children, and will create more educative home environments. In this view, correlation between home environment and the child's academic performance may mask a more important genetic relation between parents' abilities and children's abilities.

   ― HARRY L. MILLER, "Hard Realities and Soft Social Science,"
   The Public Interest, Spring 1980




5. The mechanism of stimulus and response in geotropism has often been studied. If very young seedlings in which the root and stem are just appearing are fixed in any position whatever, the young root will invariably grow downward and the young stem upward. The English horticulturalist Knight, more than a century ago, suggested that this behavior was due to gravity. He reasoned that if this were so, it should be possible to substitute a stronger force for gravity and thus to change the direction of growth. Knight fastened young plants in various positions to the rim of a wheel, which he revolved rapidly in a horizontal plane, thus subjecting the plants to a "centrifugal force" greater than gravity. Under these conditions the roots grew outward, in the direction of the centrifugal pull, and the stems grew inward, toward the hub, in an exactly opposite direction. Knight thus proved that plant structures orient themselves to this force in just the same way that they do to gravity.

   ― EDMUND W. SINNOT and KATHERINE S. WILSON,
   Botany: Principles and Problems




6. A team of researchers recently explored the relationship between shift work and heart disease among 79,109 women enrolled in the Nurses Health Study. In 1988 the women were asked how many years they had worked rotating night shifts. At the time none of these women had a history of coronary heart disease. Most of the women had done some shift work, 7% of them for 15 or more years. Compared with the women who had never worked shifts, those who had done so were slightly heavier and more likely to smoke cigarettes. Longer durations of shift work were associated with high blood pressure and diabetes. During the next four years of follow up, 292 of the women developed evidence of coronary artery disease. The women who had done shift work were 40% more likely to develop heart disease, and longer periods of shift work were associated with higher overall risk. Women who had performed more than 6 years of shift work had a 51% increase in heart disease risk, and a 29% increase in the risk of dying during the follow up period. Even when the researchers accounted for weight, smoking, and as many other cardiac risk factors as they could, the influence of shift work was still present.

   But is shift work itself the culprit? Or are women who do work shifts different from women who do not, in ways this research could not detect or take into account? These questions cannot be resolved without an experiment in which large number of women are randomly assigned or a prolonged period, either to shift work or to a regular schedule. That experiment is not likely to be conducted any time soon.

   ― I. KAWACHI, et al., "Prospective Study of Shift Work and Risk of
   Coronary Heart Disease in Women," Circulation, 1 December 1995




7. Again however solid things are thought to be, you may yet learn from this that they are of rare body: in rocks and caverns the moisture of water oozes through and all things weep with abundant drops; food distributes itself through the whole body of living things; trees grow and yield fruit in season, because food is diffused through the whole from the very roots over the stem and all the boughs. Voices pass through walls and fly through houses shut, stiffening frost pierces to the bones. Now if there are no void parts, by what way can the bodies severally pass? You would see it to be quite impossible. Once more, why do we see one thing surpass another in weight though not larger in size? For if there is just as much body in a ball of wool as there is in a lump of lead, it is natural it should weigh the same, since the property of body is to weigh all things downwards, while on the contrary the nature of void is ever without weight. Therefore when a thing is just as large, yet is found to be lighter, it proves sure enough that it has more of void in it; while on the other hand that which is heavier shows that there is in it more of body and that it contains within it much less of void. Therefore that which we are seeking with keen reason exists sure enough, mixed up in things; and we call it void.

   ― LUCRETIUS, On the Nature of Things, Bk. 1




8. Scientists have recently embarked on their most ambitious effort yet to find and exploit one of the most elusive of the predicted phenomena of nature: gravitational waves. As long ago as 1915, Albert Einstein predicted in his general theory of relativity that the violent birth and death of stars in the universe would give off gravitational waves, bathing the earth in a unique kind of radiation. But so far, after decades of searching with increasingly sensitive detectors, scientists have not found one. A new facility, known as the laser interferometer gravitational wave observatory, has been designed to bring this search to successful culmination.

   The discovery of gravitational waves would rank as one of the most important observational feats in modern physics and astronomy. It would provide a new confirmation of Einstein's General Theory of Relativity, a foundation of modern science whose validity is difficult to prove experimentally. That theory changed the concept of space from an empty void into a curving fabric of space and time. When stars collapse, the theory suggests, ripples in the space time fabric move off in all directions at the speed of light; they do not transmit the force of gravity, but are distortions of that force.

   In theory, gravity waves from distant cosmic events should move objects on earth an infinitesimal amount, measured in distances that are much smaller than the nucleus of an atom. The new observatory is designed to be sensitive enough to measure these waves and help pinpoint their cosmic origin. The risk, say skeptics, is that the apparatus will not be sensitive enough to detect the waves.

   ― Ambitious Effort Aims to Find Gravity Waves,"
   The New York Times, 27 February 1990




9. Like multiple sclerosis, poliomyelitis in its paralytic form was a disease of the more advanced nations rather than of the less advanced ones, and of economically better off people rather than of the poor. It occurred in northern Europe and North America much more frequently than in southern Europe or the countries of Africa, Asia or South America. Immigrants to South Africa from northern Europe ran twice the risk of contracting paralytic poliomyelitis than South African born whites ran, and the SouthAfrican born whites ran a much greater risk than nonwhites. Among the Bantu of South Africa paralytic poliomyelitis was rarely an adult disease. During World War II in North Africa cases of paralytic poliomyelitis were commoner among officers in the British and American forces than among men in the other ranks. At the time various wild hypotheses for the difference were proposed; it was even suggested that it arose from the fact that the officers drank whiskey whereas men in the other ranks drank beer!

   We now understand very well the reason for the strange distribution of paralytic poliomyelitis. Until this century poliomyelitis was a universal infection of infancy, and infants hardly ever suffered paralysis from it. The fact that they were occasionally so affected is what gave the disease the name "infantile paralysis." With the improvement of hygiene in the advancing countries of the world, more and more people missed infection in early childhood and contracted the disease for the first time at a later age, when the risk that the infection will cause paralysis is much greater.

   This explains why the first epidemics of poliomyelitis did not occur until this century, and then only in the economically advanced countries.

   ― GEOFFREY DEAN, "The Multiple Sclerosis Problem,"
   Scientific American, July 1970




10. Since Venus rotates so slowly, we might be tempted to conclude that Venus, like Mercury, keeps one face always toward the Sun. If this hypothesis were correct we should expect that the dark side would be exceedingly cold. Pettit and Nicholson have measured the temperature of the dark side of Venus. They find that the temperature is not low, its value being only 9°F., much warmer than our stratosphere in broad daylight. Ir is unlikely that atmospheric currents from the bright side of Venus could perpetually heat the dark side. The planet must rotate fairly often to keep the dark side from cooling excessively.

    ― FRED L. WHIPPLE, Earth, Moon and Planets




11. A large rock balanced on a small protuberance is an object of a certain wonder. Such rocks are not rare; for example, in Goblin Valley in southern Utah there are more than 1,000 of them. But how do the rocks stay balanced?

    Balanced rocks originate when a bed of sediments is dissected by erosion until a column is formed. If the strata at the top of the column are harder than the strata farther down, erosion will whittle the softer rock down to a pillar narrower than the capstone.

    Nothing about the erosion process, however, guarantees that the end product will be symmetrical, and so what keeps the capstone in place? Two investigators at Kansas State University, Wilson Tripp, an engineer, and Frederic C. Appl, whose specialty is rock mechanics, suggest that a dynamic process is responsible, that it starts when the capstone first begins to tilt in any direction and that the point of contact between the capstone and its supporting pillar continuously shifts, thereby remaining exactly under the capstone's center of gravity. The principle that underlies the process is simply that rock under the stress of compression is more resistant to erosion than unstressed rock.

    When the capstone first begins to tilt, Tripp and Appl note, the movement will shift the stress of compression from one section of the supporting pillar to another. Thereafter the unstressed section will erode more rapidly than before and the stressed section will erode more slowly. Successive tilts in other directions will stress successive sections of the pillar, and the differential erosion that results will make the process self leveling. As a consequence the capstone will remain poised on the pillar until the inevitable day when the area of contact becomes too small for the self leveling to continue, and the balancing work, ceasing its apparent defiance of the laws of statistical mechanics, crashes satisfyingly to the ground.

    ― "Science and the Citizen," Scientific American, March 1974




12. Find a sociological puzzle, and one gets, usually, a slew of esoteric explanations, couched, of course, in opaque sociological jargon. Take the question which has befuddled a number of commentators recently: Why is it that women today seem to be marrying later than before? We shall not try to list all the ingenious explanations that have been advanced, from the rise of women's liberation to the increasing proportion of open homosexuality, male and female. Suffice to say that simple statistics, once understood, provide the most likely explanations. Here is Paul C. Glick, of the U.S. Bureau of the Census, writing in Current Population Reports:

    One of the tangible factors that probably helps to explain the increasing postponement of marriage is the 5 to 10 percent excess of women as compared with men during recent years in those ages when most first marriages occur (18 to 24 years for women and 20 to 26 years for men). This imbalance is a consequence of past fluctuations in the birth rate. For example, women born in 1947 after the baby boom had begun were ready to marry in 20 years, but the men they were most likely to marry were born in 1944 or 1945 (about one half in each year) when the birth rate was still low; these men were about 8 percent less numerous than the 20 year old women. (By contrast, girls who were born during the last 15 years, while the birth rate has been declining, will be scarce as compared with eligible men when they reach the main age for marriage.)

    ― VICTOR R. FUCHS, "The Economics of Health in a
    Post Industrial Society," The Public Interest, Summer 1979




13. Early in the eighteenth century Edmund Halley asked: "Why is the sky dark at night?" This apparently naive question is not easy to answer, because if the universe had the simplest imaginable structure on the largest possible scale, the background radiation of the sky would be intense. Imagine a static infinite universe, that is, a universe of infinite size in which the stars and galaxies are stationary with respect to one another. A line of sight in any direction will ultimately cross the surface of a star, and the sky should appear to be made up of overlapping stellar disks. The apparent brightness of a star's surface is independent of its distance, so that everywhere the sky should be as bright as the surface of an average star. Since the sun is an average star, the entire sky, day and night, should be about as bright as the surface of the sun. The fact that it is not was later characterized as Olbers' paradox (after the 18th century German astronomer Heinrich Olbers). The paradox applies not only to starlight but also to all other regions of the electromagnetic spectrum. It indicates that there is something fundamentally wrong with the model of a static infinite universe, but it does not specify what.

    ― ADRIAN WEBSTER, "The Cosmic Radiation Background,"
    Scientific American, August 1974




14. The vast majority of fish species are cold blooded, meaning that their body temperatures fluctuate with the surrounding water. Some 25 species, however, can keep their eyes, brains, or entire bodies warm, independent of ambient temperatures, as birds and mammals do.

    For years scientists have debated which of two competing theories better explains this finny ability. One was that the fish acquired their warming techniques primarily so they could expand their ranges into colder regions of the ocean, which promised new sources of food. The other hypothesis was that the techniques allowed the fish to increase their aerobic capacity so they could be more active.

    Now Dr. Barbara A. Block, an animal physiologist at the University of Chicago, has concluded that new genetic evidence makes a strong case for the first theory, called niche expansion. She discovered that the warming techniques evolved not just once but on three separate occasions. "If it had evolved only once, it might not seem important," said Dr. Block, "but we're getting a clear message that keeping the central nervous system and eyes warm gives an advantage. We can see a clear correlation. The endothermic species have broad thermal niches, and the only other thing they have that others [closely related] don't have is the ability to warm their heads."

    ― Reported in Science, April 1993




15. Dr. Konrad Buettner of the University of California at Los Angeles has recently advanced the hypothesis that, during the lifetime of the moon, the everlasting influx of cosmic rays has slowly ground the upper surface layers of rocks into fine dust. That the moon's skin cannot consist of solid rocks has been demonstrated through temperature measurements during lunar eclipses. As soon as the shadow of the earth creeps over the measuring area the temperature drops steeply, and after half an hour it is over 200º F. lower than it was in the full sun. When the shadow has passed by, the temperature again rises at a similarly steep rate. No solid piece of rock can cool down and heat up so quickly. These drastic temperature changes can be explained only by the existence of a thick layer of heat insulating dust as fine as face powder. The thickness of the layer must be at least several inches. The sandblasting of meteoric dust also grinds at the moon's surface, but cosmic rays can be expected to do a much better job.

    ― HEINZ HABER, Man in Space

Summary of Chapter 15

In this chapter, we saw how the method of science is used to identify and formulate the principles that underlie the patterns encountered in our observation of the world; to resolve problematic situations by explaining the facts that create the problem; and to formulate general truths and causal laws that give us some measure of control over our world. The nature of this scientific method, and the hypotheses or theories produced by it, is the subject of this chapter.

In section 15.1, we examined the values of science: its practical usefulness and its satisfaction of our human desire to know and understand.

In section 15.2, we distinguished scientific explanations, always hypothetical and uncertain in some degree, from unscientific explanations, which are characterized by dogmatism and a spirit of finality. Above all, a scientific explanatio