|The following correction was subsequently made to this article in issue 20 (volume 7.1):|
|Contrary to a report in issue XVIII, Galileo has not been "pardoned" by the pope; rather, his views have been more or less pardoned or condoned a bit apologetically without opening the case to full legalistic and awkward review. Among other details, his "conviction" was a muddy affair which did not challenge heliocentrism, as such, although that was the ostensible issue. It has been suggested that his conviction was in fact a plea bargain which got a modernizing pope off the hook along with Galileo who spent the rest of his life under house arrest rather than executed for heresy.|
The single most important outcome of the intensive work on the history and philosophy of science that has been done over the past twenty-five years is that science must be understood first and foremost as a process, not as a system of proven results. Scientists pose hypotheses about various aspects of nature and test these hypotheses by examining nature. When a hypothesis is well supported, they use it as a basis for learning more about nature, and, when a hypothesis fails to give correct results, it is revised or rejected and replaced.
There is a wide variety of different sorts of scientific hypotheses, and I will try to indicate how wide this variety is, but I will not attempt to illustrate all of the different kinds of hypotheses that occur in science. Some hypotheses deal with the occurrence of specific events—for example, that a total eclipse of the sun will occur at a particular time in the future, that a total eclipse did occur at a particular time in the past, or that Earth was hit by an asteroid at the end of the Cretaceous. Contrary to a widespread view, there is no special problem in testing hypotheses about past events, because such events leave currently available evidence. Such testing is, of course, indirect, but we will see that this is the normal situation in science. Other hypotheses make claims about the basic properties of a kind of entity—for example, the charge and mass of the electron or the melting point of copper. Some hypotheses make claims about how objects behave under specified circumstances—that planets moving in the gravitational field of the sun travel in elliptical orbits. Some hypotheses make quite general claims that apply to a wide variety of situations—for example, the claim that energy is conserved in all physical processes or that the velocity of light in a vacuum is the same for all observers. And in some disciplines we find systems of hypotheses which, taken together, describe the fundamental features of a particular portion of the natural world; these are known as fundamental theories, for example, Newton's mechanics, quantum theory, relativity theory, the theory of evolution by natural selection, or plate tectonics.
This is a small sampling of theses that do play, or have played, an important role in science. Some of the claims I mentioned are theses which, at present, we have good reason for believing correct, and some we have reason for believing incorrect, but the central point that concerns me about them is captured in the term hypotheses, because, no matter how powerful the evidence in favor of a scientific thesis, we never reach the point at which that claim has been unequivocally proven. Sometimes evidence in favor of a theory is so powerful and comprehensive that the theory becomes accepted as the basis for research in a discipline, and for a substantial period of time that theory is not seriously open to question. Still, the theory remains a hypothesis, open to reconsideration or rejection under appropriate circumstances.
I have been emphasizing the testability and refutability of scientific hypotheses, but this point must not be understood in a simplistic fashion. The process of testing scientific theories is often complex, and it is not always obvious just when a theory should be rejected. There are situations in which there appears to be evidence against a theory but in which further research shows that the evidence is either irrelevant or in fact supports the theory. Just as it is important for scientists to be prepared to reject hypotheses, so it is important that hypotheses not be rejected prematurely, and the tenacious defense of a hypothesis can play a positive role in the development of science. Science operates in terms of what Thomas Kuhn has called an "essential tension" between conservatism and innovation, between testing and defending theories, between acceptance of hypotheses and their rejection and replacement (1977).
In order to understand these ideas more fully, it will be useful to have an actual example from the history of science before us, so I want to sketch some aspects of the career of Newton's mechanics. This is one of the most powerful and successful scientific theories ever developed; it is a theory which provided the basis for virtually all physical research for a period of over 150 years; it is a theory which was supported by overwhelming evidence and which overcame a number of apparent refutations but which was ultimately seen to be incorrect.
Newton's Mathematical Principles of Natural Philosophy was published in 1687 and, contrary to the impression one gets from modern physics textbooks, it did not immediately sweep the scientific world; it took approximately fifty years before the theory was generally accepted as correct. Originally, there were two kinds of objections to the theory. One of these was conceptual: the key idea of universal gravitation—that bits of matter exercise a force on each other even when they are not in physical contact—seemed absurd to both the common sense and the accepted science of the day. Newton himself was never happy with this idea and tried, unsuccessfully, to replace gravitational attractions with some system of particles pushing on each other. This problem was solved in a familiar way: many new ideas seem obviously absurd when first proposed, but, when they are shown to be fruitful and when they become familiar, we soon begin to wonder how we were ever able to think in a different fashion.
The second class of problems was straightforwardly empirical, for, in spite of its striking successes, there were also cases in which Newton's theory gave the wrong results. The most important of these had to do with a particular aspect of the orbit of the moon (the motion of the line of upsides), for which Newton's calculated value was half the observed value. Now there is a very important respect in which the derivation of an incorrect prediction from a theory is vastly more revealing than the derivation of a correct prediction. For if a theory yields correct predictions, that shows that the theory may be right but offers no guarantees, while the derivation of an incorrect prediction does guarantee that something is wrong somewhere. This is central to the logic of theory testing, and I want to leave the story of Newton's mechanics for a moment in order to develop this idea.
The best way to clarify this point is to use a simplified model. Suppose I have a machine in which I keep coins; I deposit coins into a slot, and a digital display gives the total amount of money in the machine. You visit me one day, and you offer the hypothesis that I have three quarters in the machine. A brief calculation yields the prediction that the display will read seventy-five cents, and, when you check the display, your prediction is confirmed. You have evidence that supports your hypothesis. But has your hypothesis been proven? Of course not, for there are many other hypotheses that would yield the same result: I might have put seventy-five pennies in my machine or seven dimes and a nickle. Suppose, however, that you offer the same hypothesis—that there are three quarters in the machine—and, on checking, you find that the display reads sixty-two cents. In this case, it is clear that your hypothesis is wrong. We do not know what the correct hypothesis is, but three quarters do not add up to sixty-two cents, and that hypothesis is eliminated.
Now there are many respects in which this model does illustrate the key features of scientific theory testing. Just as in my model you are not permitted to look into the machine and see what coins are there, so it is not possible to test most scientific theories directly. Rather, we deduce results from those theories and check to see if those results are correct. If the results are correct, we have some evidence that confirms our theory, and, if the results are incorrect, we have powerful evidence against our theory. To be sure, interesting scientific theories are not nearly as trivial as my model. Newton's theory, for example, gave dozens of correct predictions. Still, the key point I want to illustrate with the model does hold: you can begin with an incorrect hypothesis, reason from it correctly, and come up with correct predictions; but you cannot begin with a correct hypothesis, reason correctly, and end up with incorrect predictions. If a hypothesis yields correct results, it is supported by those results but is not proven; if a hypothesis yields incorrect results, we have proof that something is wrong.
Unfortunately, there is one complication that I have not yet mentioned. Go back to the case in which you are offering hypotheses about the coins in my machine. You proposed that there are three quarters in the machine, and the display reads sixty-two cents. Clearly your hypothesis is wrong, unless the machine is broken. That is, although a result which is different from what we predicted does show that something is wrong, it does not show what is wrong, and it will not always be clear just where our theory needs to be modified. This may take a great deal of further research, and the results will often be surprising. This is analogous to what happened in the case of Newton's incorrect prediction, and I want to return now to that story, keeping in mind the points I have just made about the logic of theory testing.
The key figure in the resolution of this story was Clairaut. For a while, Clairaut thought that the observed motion of the moon showed that something was wrong in Newton's theory, although he believed that only a small modification was required (the addition of an inverse fourth power term to the law of gravitation). But then, around 1750, Clairaut made a surprising discovery. The calculations required to get from Newton's theory to a predicted orbit for the moon are quite complex. In the course of these computations, mathematicians had been making what seemed to be a series of correct approximations, and Clairaut found that the error was located in these approximations, not in the theory. For some sixty years it had seemed that Newton's theory gave an incorrect result, but it would have been premature to reject the theory for that reason. Appearances to the contrary, the theory was capable of giving the correct result, provided it was developed in the correct way. An important point about science now emerges: sometimes it just takes time before a question can be resolved, and, although people get impatient and want the answers to their questions now, it is not always possible. All of these points can be further illustrated by following the career of Newton's theory for the next hundred fifty years or so.
For about a century, there were no serious challenges to Newton's theory, but a great deal of work was done on and in the theory. Superior mathematical techniques were developed, leading to more precise predictions, while better observational techniques were also being developed, which led to tougher tests of these predictions. As a result of these developments, two new problems arose. The power of Newton's work had shown itself, first of all, in its ability to yield precise calculations for the orbits of the planets, and, by the middle of the nineteenth century, it became clear that the theory was not giving correct results for Mercury, the closest planet to the sun, and for Uranus, the most distant of the planets known at that time. I want to consider the outcome of each of these problems, because the two problems eventually had quite different resolutions.
The problem of Uranus was resolved first. One approach, taken by Airy, the British Astronomer Royal, was to suggest, once again, that Newton's gravitation law was wrong, but this view did not prevail. The alternative approach was developed independently by two scientists, Urban Leverrier and John Adams. These two set out to solve the problem within the framework of Newtonian mechanics by proposing that, rather than Newton's theory being incorrect, the mistake lay in the belief that all the planets were known. If, however, there were an eighth planet, this planet would exercise a gravitational attraction on Uranus and affect its orbit. In other words, the orbit of Uranus was calculated by first calculating what the orbit would be if only Uranus and the sun existed and by then correcting the orbit to take account of the much smaller attractions of each of the known planets. This calculated orbit did not quite match the observed orbit, and Leverrier and Adams considered the hypothesis that the difference was caused by the gravitational attraction of an as yet unobserved planet. Knowing the supposed effects of this planet on Uranus, they were able to compute the mass and orbit that the planet must have, and the planet was found by an observer, Galle, the first night he looked for it, within one degree of the predicted location. That planet is Neptune.
Several points require emphasis. First, we must be clear that cases in which theory disagrees with observation must be taken very seriously indeed, for they do guarantee that something is wrong somewhere in our currently accepted body of beliefs, but it is not always obvious just where the error is. A great deal of time, research, and effort may be required to isolate the problem, and it may turn out to be relatively minor, as in the two cases cited thus far. We do not reject a powerful, successful theory just because it faces problems, as the discovery of Neptune illustrates. What seems to be a refutation may turn into a major triumph for the theory in question. Still, even the most successful scientific theory is open to reconsideration and may be rejected. This eventually occurred in the case of Newton's theory, and that brings us to the story of Mercury.
You will, perhaps, not find it surprising that, after the discovery of Neptune, Leverrier, one of the two scientists who had predicted the existence of this planet, attempted to solve the problem posed by Mercury's orbit in a similar way. That is, Leverrier hypothesized another planet, Vulcan, located between Mercury and the sun and attempted to compute its mass and orbit from the known differences between the observed orbit of Mercury and the previously computed orbit. But this time the approach failed. There is no evidence that such a planet exists, and there are now very good reasons for believing that it is not possible to calculate the orbit of Mercury correctly using Newton's theory of gravitation—although it is possible to get this information, and that for the other planets as well, by using Einstein's vastly more complex theory of gravitation.
To be sure, we do still use Newton's theory for a wide variety of situations, and this is legitimate because we know that, for those situations, Newton's theory gives the correct results. But we must keep two points clearly in mind. The first is that Newton's theory also gives wrong results for a large variety of cases. This is important because the theory is supposed to give us a picture of how the universe is built, and, if this picture is correct, all results derived from the theory ought to be correct. Again, this is the fundamental fact behind the logic of theory testing: if a theory yields incorrect results, something is wrong, no matter how many correct results it yields. The second point is that, if we turn to Einstein's theory for an alternative picture of how the universe is built, we get a picture that is completely different from the Newtonian picture and that makes use of ideas that would never have occurred to Newton writing in the seventeenth century. This is not a criticism of Newton. Newton's work is brilliant and stands as a major landmark in the development of science. But, since Newton's time, we have learned a lot, and fundamentally new ideas have been invented with the result that the current scientific picture of the structure of the universe is quite different from anything our ancestors could have imagined.
It would be hard to find a clearer or more dramatic illustration of what it means to say that all scientific theories are hypotheses. No matter how well they are supported, no matter how many objections they overcome, no matter how many tests they pass, scientific claims are always open to reconsideration. It is, I think, impossible to overemphasize this point. There are, no doubt, scientists who speak and write as if they have achieved the final word in their discipline, and this is unfortunate. But the way to get a proper perspective on science is by looking both at its historical development and at what contemporary scientists do. Fifteen years ago, for example, astronomers believed that the question of how the sun produces its energy had been answered; it became a closed, finished subject. Now they are not so sure. Results from an experiment still in progress (Bahcall and Davis, 1978) have opened the question up again. The experiment, known as the solar neutrino experiment, is extremely complex and difficult, and the question we must ask ourselves is: why, if scientists were sure that they had the correct theory of the sun, was the experiment tried at all? The answer, I think, is clear: all scientific views are open to reconsideration, and, when someone thinks of a new way to test an old theory, one can expect that the tests will be done, and, in many cases, the outcome will surprise us. Note also that, although this experiment has given results that are different from those predicted, we cannot yet say that the previously accepted theory of the sun has been refuted. The problem is that there are a number of theories involved in the design and the evaluation of the experiment, and what is clear now is that something is wrong somewhere. Just what is wrong, or how to fix it up, is not yet clear.
This brings me to another aspect of science that must be considered: science takes time, often long periods of time in comparison to a human life span. Recall that it took approximately sixty years from the publication of Newton's theory to its general acceptance and another 175 years until it became clear that the theory was not right. Why wasn't the problem with Mercury's orbit discovered at once? The problem involved an extremely tiny shift in the location of the point in the planet's orbit at which it is closest to the sun, and the difference between the observed motion of this point and the motion computed from Newton's theory was approximately twenty-two seconds of arc per century. It just took time to discover that disagreement. Take another example. Pluto, the most recently discovered planet, was first observed about fifty years ago. As it takes Pluto almost 250 years to make one orbit of the sun, we have not yet observed a single orbit and will not do so in our lifetime; therefore, we have no idea what small discrepancies may appear after a few dozen orbits have been recorded. In fact, it takes Neptune about 165 years to complete an orbit. Again, we have not yet observed an entire orbit, but a good deal of data is available, and in the past few years some astronomers have begun to suspect that there is a discrepancy between the observed and calculated orbits for Neptune.
This is all very frustrating to those of us who would like to know—now—how the universe is built. The only response available here is that that is tough, this is just a part of human life. Seventy-five years ago, the major cause of death in the developed world was pneumonia. It is no longer a major killer at all compared to cancer and heart disease because we have learned to cure pneumonia. This does not help those who died of pneumonia in the past century, but it took time to find the drugs that cure this disease. Hopefully, sometime in the future, we will find cures for cancer and heart disease. But the development of the necessary knowledge takes time, the process if full of errors and false starts, and it may turn out that we will not cure these ills until someone comes up with a radical new idea—an idea that will seem as absurd, at first glance, as the notion that diseases can be caused by microbes seemed to people not so very long ago.
Let me summarize the upshot of my discussion so far. In particular, I want to emphasize three points about the development of science that I have been trying to illustrate and explain:
I am now ready to pose a most important question: why are so many deeply religious people opposed to treating the biblical story of creation as science? Part of the answer should, by now, be obvious. To claim that the biblical story of creation is a scientific claim is to hold that this story is a tentative hypothesis, subject to test and evaluation and subject to possible replacement by a quite different hypothesi, s. Very few religious people are prepared to hold their religious beliefs as tentative hypotheses, nor are the people who describe themselves as creation scientists vigorously pursuing new kinds of observations which might refute their hypotheses. Let me press this point, because, if we do decide to take biblical statements as hypot, heses about events that have occurred, then there are some striking questions that should be asked. For example, many creationists insist on the literal truth of the story of Noah and the flood. But consider how we would respond to the story if we encountered it, outside of its religious context, in a science textbook, presented as a hypothesis to be evaluated against the evidence. What kinds of questions would we ask? We are told, for example, that in the course of seven days Noah collected a mating pair of each of the "unclean" beasts, seven pairs of the "clean" beasts, and seven pairs of each kind of bird. Did he really get polar bears and penguins in just seven days without the use of jet aircraft or other such devices? What did the gazelle do for exercise during the 150 days the ark was afloat? How was fresh meat for the tigers stored during those five months? If one thinks that questions of this kind are inappropriate when dealing with a biblical text, I will not argue the point—as long as the text is not offered as science.
Creationists, of course, proceed in a different way. Rather than developing and testing their own theory, they criticize what they consider their major opponent, evolution. Now in one respect this is most desirable. Evolution is a scientific theory; it has its successes. It has overcome what seemed to be devastating criticism—for example, the physicist Kelvin's argument in the past century that the sun could not have been burning long enough to provide the required time span. The theory of evolution has been modified in the period since Darwin's first formulation, and it may some day be abandoned and replaced. To the extent, then, that creationists, or anyone else, point up actual problems for an existing theory, they make a genuine contribution to the development of science. But it is a mistake to think that finding problems in evolution theory, or even finding a knockdown, drag-out refutation of evolution, would provide evidence in favor of creationism. It would support creationism if this were the only possible alternative, but one need only look at the world's religious literature to find many tales about the origin of the universe which do not include seven days of creation, creation from nothing, the story of the ark, and so forth. If a disproof of evolution would support some alternative theory now available, it gives no more support to biblical creationism than to the stories found in Hindu texts or in various African or American Indian traditions. We could, no doubt, begin seeking evidence which would allow us to compare the scientific merits of these different viewpoints. For example, the story of creation from nothing, considered as a scientific hypothesis, clearly violates the laws of conservation of energy and conservation of matter; stories of creation from a cosmic egg, or from some previously existent material, do not obviously violate these principles. But, once again, I submit that this is not the way that many people view their own religious commitments.
We have, then, part of the answer to our question of why religious people would oppose treating the biblical story of creation as science: they do not think of their religious beliefs as tentative hypotheses open to evaluation and possible refutation. But there is more to the story, and this requires a bit more historical reflection. There have been cases in which people tried to settle scientific issues by appealing to scripture, and the result has been harmful to both science and religion—although I think that religion has suffered more than science as a result. The most famous example is the conflict between Galileo and the Roman Catholic Church in the seventeenth century over the question of the motion of Earth. I will summarize this story briefly.
In 1543, Copernicus published a book in which he developed in detail an astronomical theory based upon the assumption that Earth spins daily on its axis and moves, in the course of a year, around the sun. The earlier view, according to which Earth stands still at the center of the universe, was by no means foolish, and there was a great deal of straightforward scientific evidence to support it. But it seemed to many that the older view was required by scripture as well. Discussion continued for a time, but eventually the Church decided to look into this new view. In 1616, when the discussion was coming to a head, Galileo went to Rome to try to convince Church authorities that it would be a mistake to make the outcome of a scientific debate a matter of Church doctrine. I want to emphasize that Galileo was not attacking the Church. He always considered himself a devout Catholic, and his only aim was to try to prevent his Church from making a decision that would hamper science and embarrass Catholicism. Galileo lost this battle, portions of Copernicus' work were declared by the Church to be not only wrong but heretical, and Galileo was called before Cardinal Bellarmine, head of the Inquisition, and ordered to no longer hold, teach, or defend the Copernican view. For a long time Galileo held his peace, but in 1632 he published a book in which he defended Copernicanism. As a result, the following year, aged sixty-nine, ill, and almost blind, he was tried in Rome for holding, teaching, and defending the Copernican view, was forced under threat of torture and execution to recant publicly, and was confined to house arrest for the remaining nine years of his life. It is worthwhile to recall some of the biblical passages that were cited in the early seventeenth century to prove that Earth did not move. Joshua, for example, commanded the sun to stand still: "Sun stand thou still on Gideon, and thou, Moon, in the valley of Ajalon" (10:12), and in the next verse we are told, "And the sun stood still and the moon stayed. . . ." It was regularly pointed out that Joshua commanded the sun to stand still; he did not command Earth to stop turning, as he surely would have done if Copernicus had been correct. In Ecclesiastes we read, "The sun also riseth, and the sun goeth down, and hasteth to the place where he rose" (1:5), and the Psalms speak of the Lord who, among other things, "laid the foundations of the earth, that it should not be moved forever" (103:5, Douay; 104:5, King James). It is also worth noting that back in 1533, when a brief summary of Copernicus' views was being discussed, Martin Luther also referred to the story of Joshua to show that Earth does not move.
These examples should suffice to make the point: there are passages in the Bible which we have reason to believe are wrong if read literally and taken as scientific claims about the world. Galileo's point, on the other hand, was right: the attempt to settle scientific issues by appeal to scripture is likely to end up in serious theological embarrassment. This particular embarrassment still lingers, to the point that Galileo's trial was recently reopened and the judgment reversed. It is in this context that I think we can fully understand why so many thoughtful religious people insist on keeping science and religion distinct.
There is one more aspect of science that must be considered before I conclude these remarks. A central aim of science is the attempt to understand nature in terms of natural processes, and it is inappropriate to invoke miracles or other forms of supernatural intervention in the course of developing scientific explanations. To the extent, then, that creationists postulate divine acts to account for features of the natural world, they are just not doing science. Similarly, presumed revelations are not a legitimate source of scientific knowledge. A large part of the challenge of science consists of the attempt to understand the world around us by the use of human intelligence. How far is it possible to go in attempting to carry out this project—that is, are there limits to science? It seems to me that the only appropriate answer to this question, at the present stage of the development of science, is that we do not know what the limits of science are or if there are any at all. Many limits which people once thought had been established have been surpassed. For example, it was once thought that organic molecules could not, by their very nature, be produced in the laboratory; but it has been done. Many nineteenth century physicists thought that a theory of atomic and subatomic particles was beyond the capacity of the human intellect; they were wrong. As recently as 1960, the eminent astronomer Fred Hoyle argued that, because of the intrinsic limits of telescopes, we had reached the end of our ability to observe distant astronomical objects and that no fundamentally new data would ever be available. It did not occur to Hoyle at that time to note that he was talking about optical telescopes; yet, while he was writing, radio telescopes were being developed and deployed, and these new observing devices have provided an impressive array of new discoveries. We have expanded our knowledge to an incredible degree in the past hundred years, and we have no idea what will be accomplished in the next thousand years. There may be limits to science, there may be limits to human knowledge, but we do not know if there are such limits or what they are. If this question can be answered at all, it can only be answered by further research. And if the history of science provides any indication as to what we should expect as a result of further research, it is that we should expect to be surprised.
Bahcall, J., and Davis, R. 1976. "Solar Neutrinos: A Scientific Puzzle." Science, 191:264.
Burchfield, J. 1975. Lord Kelvin and the Age of the Earth. New York: Science History Publications.
Grosser, M. 1962. The Discovery of Neptune. Cambridge: Harvard University Press.
Hoyle, F. 1960. The Nature of the Universe. New York: Signet Books.
Kuhn, T. 1977. The Essential Tension. Chicago: University of Chicago Press.