The English astronomer and mathematician Sir Arthur Stanley Eddington is famous for his work concerning the theory of relativity. He is also well known a philosopher of science and a populariser of science.

The publisher Klaus-Dieter Sedlacek, born in 1948, has lived in southern Germany since his childhood. He studied mathematics and computer sciences in addition with physics. Since the last 10 years, he published more than 80 books.

Bibliografische Information Der Deutschen Bibliothek:

Die Deutsche Bibliothek verzeichnet diese Publikation

in der Deutschen Nationalbibliografie; detaillierte

bibliografische Daten sind im Internet über


Newly published

Herstellung und Verlag: BoD – Books on Demand GmbH, Norderstedt.

ISBN: 9783749413799



This book contains the substance of the course of lectures which I delivered as Tarner Lecturer of Trinity College Cambridge in the Easter Term 1938. The lectures have afforded me an opportunity of developing more fully than in my earlier books the principles of philosophic thought associated with the modern advances of physical science.

It is often said that there is no "philosophy of science", but only the philosophies of certain scientists. But in so far as we recognize an authoritative body of opinion which decides what is and what is not accepted as present-day physics, there is an ascertainable present-day philosophy of physical science. It is the philosophy to which those who follow the accepted practice of science stand committed by their practice. It is implicit in the methods by which they advance science, sometimes without fully understanding why they employ them, and in the procedure which they accept as giving assurance of truth, often without examining what kind of assurance it can give.

There should be no conflict between the claim that a philosophy is scientifically grounded and the claim that it is, so far as it goes, a true philosophy. But in a specialised work of this kind the primary object must be to ascertain and discuss the philosophy which, whether true or not, is the present philosophy of physical science in the sense stated above. Those of us who believe that science, notwithstanding continual failures and readjustments, is slowly drawing nearer to the truth, are content that philosophic truth should be reached by the same method of progressive advance.

In order to make sure of our scientific foundations it is found necessary to enter rather deeply into the principles of relativity theory and quantum theory. Since the intention is

to give, not merely an exposition, but a justification of the views to which they lead, some parts of the book introduce matters of considerable technical difficulty. Generally I have abstained from mathematical formulae; this, however, is not wholly out of consideration for the general reader, but because those whose minds arc too much immersed in mathematical formulae are likely to miss what we are here seeking.

The discussion, although relating to the same subject matter, is mainly on different lines from that given eleven years ago in The Nature of the Physical World. The starting point in the present treatment is knowledge. The title of the earlier book might have been expanded into "the nature of die physical universe, with applications to the theory of physical knowledge"; the corresponding title of the present book would be "the nature of physical knowledge, with applications to the theory of the physical universe". The change of emphasis makes for a more logical sequence of ideas; but primarily it reflects a change which has occurred in physical science itself. It is significant of this change that the contrast between the scientific table and the familiar table, with which The Nature of the Physical World opens, had become a contrast between the scientific story and the familiar story of experience at the beginning of New Pathways in Science. The first was, I believe, the natural form of expression according to the scientific outlook of 1928; the second had become more natural six years later.

Neither the scientific advances of the last decade nor the years of reflection have altered the general trend of my philosophy. I say "my philosophy", not as claiming authorship of ideas which are widely diffused in modem thought, but because the ultimate selection and synthesis must be a personal responsibility. If it were necessary to give a short name to this philosophy, I should hesitate between "Selective subjectivism" and "Structuralism". The former name refers to the aspect most prominent in the first eight chapters; the latter refers to a more mathematical conception which dominates the rest of the book. Both can now be carried much farther than in The Nature of the Physical World. The domain of subjectivity has been extended as a consequence of our better understanding of quantum mechanics; and the conception of structure has been made more precise by the connection now recognized between the foundations of physics and the mathematical Theory of Groups.

With this "philosophy of physical science" as a nucleus, I endeavour in the last two chapters to develop the outline of a general philosophical outlook which a scientist can accept without inconsistency. 1 am not among those who think that in the search for truth all aspects of human experience are to be ignored save those which are followed up in physical science. But I find no disharmony between a philosophy which embraces the wider significance of human experience and the specialised philosophy of physical science, even though the latter relates to a system of thought of recent growth whose stability is yet to be tested.



April 1939




Between physics and philosophy there lies a debatable territory which I shall call scientific epistemology. Epistemology is that branch of philosophy which treats of the nature of knowledge. It will not be denied that a significant part of the whole field of knowledge is that which has come to us by the methods of physical science. This part takes the form of a detailed description of a world—the so-called physical universe. I give the name "scientific epistemology" to the sub-branch of epistemology which deals with the nature of this part of our knowledge, and therefore indirectly with the nature and status of the physical universe to which it formally relates.

There are two matters of definition which it is desirable to make clear at the outset.

Some writers restrict the term "knowledge" to things of which we are quite certain; others recognise knowledge of varying degrees of uncertainty. This is one of the common ambiguities of speech as to which no one is en tided to dictate, and an author can only state which usage he has himself chosen to follow. If "to know" means "to be quite certain of", the term is of little use to those who wish to be undogmatic. I therefore prefer the broader meaning; and my own usage will recognise uncertain knowledge. Anything which would be knowledge if we were assured of its truth, is still counted as knowledge (uncertain or false knowledge) if we are not assured.

It will not be necessary for us to formulate a general definition of knowledge. Our procedure will be to specify a particular collection of more or less widely accepted knowledge, and then to make an epistemological study of its nature. Especially, though not exclusively, we have to consider the knowledge acquired by the methods of physical science. For brevity I will call this physical knowledge. In principle we might identify physical knowledge with the contents of certain encyclopedic works, such as the Handbuch der Physik, which between them cover the various branches of physical science. But there are obvious objections to a slavish acceptance of a particular authority; and I will therefore define physical knowledge to be that which a right-thinking person1 would to-day accept as justified by physical science.

It should not be overlooked that physical knowledge includes a vast amount of miscellaneous information which would be out of place in scientific text-books. For example, the result of a measurement of weight is physical knowledge, whether it is made for the purpose of deciding a scientific issue or for deciding the amount of a tradesman's bill. The condition is that it shall be passed as scientifically correct (by the right-thinking person), not that it shall be scientifically important. It should also be noticed that the term is intended to refer to physical science as it stands to-day. We are not going to occupy ourselves with speculations as to possible future developments. We are to take stock of the results which the methods of physical science have yielded up to now, and see what kind of knowledge we have been acquiring.

I have said that I do not regard the term "knowledge" as implying assurance of truth. But in considering a particular body of knowledge, it may be assumed that an effort has been made to admit to that body only the more trustworthy knowledge; so that usually a reasonable degree of certainty or probability is attributable to the knowledge which we shall have occasion to discuss. But the assessment of certainty of knowledge is to be regarded as separate from the study of the nature of knowledge.

The other matter of definition is the term "physical universe". Physical knowledge (as accepted and formulated to-day) has the form of a description of a world. We define the physical universe to be the world so described. Effectively therefore the physical universe is defined as the theme of a specified body of knowledge, just as Mr Pickwick might be defined as the hero of a specified novel.

A great advantage of this definition is that it does not prejudge the question whether the physical universe—or Mr Pickwick—really exists. That is left open for discussion if we can agree on a definition of "really exists", which for most persons is a parrot-phrase whose meaning they have not troubled to consider. The few who have attempted to give it a definite meaning do not always agree on the meaning. By defining the physical universe and the physical objects which constitute it as the theme of a specified body of knowledge, and not as things possessing a property of existence elusive of definition, we free the foundations of physics from suspicion of metaphysical contamination.

This type of definition is characteristic of the epistemological approach, which takes knowledge as the starting point rather than an existent entity of which we have somehow to obtain knowledge. But in defining scientifically a term already in common use, we must be careful to avoid abuse of language. To justify the above definition of the physical universe, we ought to show that it is not in conflict with what the ordinary man (in which term I do not include philosophers) understands by the physical universe. This justification is deferred to p. 136.


The nature of physical knowledge and of the world which it professes to describe has long been a battleground for rival schools of philosophers. But physicists can scarcely be denied a hearing on a subject which concerns them so intimately. A student of physical science should be in a position to throw some light on the nature of the knowledge obtainable by the methods which he practises. Recently a number of books have been written by authors whose qualifications are purely scientific, in which scientific epistemology is developed and used as an approach to the wider problems of philosophy. I do not think that this ' intrusion" into philosophy is a matter for surprise or caustic comment.

One often finds an impression that it is an innovation for scientists to indulge in philosophy; but this is incorrect. I have noticed that some of the recent books are plentifully sprinkled with quotations from scientists of the nineteenth century which, whether they fortify the argument or not, prove at any rate that our predecessors shared the common foible of holding strong philosophic views—and expressing them. Some were out of their depth, then as now. But some were profound thinkers—Clifford, Karl Pearson, Poincaré, and others—whose writings have an honoured place in the development of scientific philosophy.

It is, however, important to recognize that about twenty-five years ago the invasion of philosophy by physics assumed a different character. Up till then traffic with philosophy had been a luxury for those scientists whose disposition happened to turn that way. I can find no indication that the scientific researches of Pearson and Poincare were in any way inspired or guided by their particular philosophical outlook. They had no opportunity to put their philosophy into practice. Conversely, their philosophical conclusions were the outcome of general scientific training, and were not to any extent dependent on familiarity with recondite investigations and theories. To advance science and to philosophize on science were essentially distinct activities, hi the new movement scientific epistemology is much more intimately associated with science. For developing the modern theories of matter and radiation a definite epistemological outlook has become a necessity; and it is the direct source of the most far-reaching scientific advances.

We have discovered that it is actually an aid in the search for knowledge to understand the nature of the knowledge which we seek.

By making practical application of our epistemological conclusions we subject them to the same kind of observational control as physical hypotheses. If our epistemology is at fault, it will lead to an impasse in the scientific developments proceeding from it; that warns us that our philosophical insight has not been deep enough, and we must cast about to find what has been overlooked. In this way scientific advances which result from epistemological insight have in turn educated our epistemological insight. Between science and scientific epistemology there has been a give and take by which both have greatly benefited.

In the view of scientists at least, this observational control gives to modern scientific epistemology a security which philosophy has not usually been able to attain. It introduces also the same kind of progressive development which is characteristic of science, but not hitherto of philosophy. We are not making a series of shots at ultimate truth, which may hit or miss. What we claim for the present system of scientific philosophy is that it is an advance on that which went before, and that it is a foundation for the advances which will come after it.

In science the observational test is valuable, not only for controlling physical hypotheses (for which it is indeed the only possible guarantee), but also for detecting fallacies of argument and unwarranted assumptions. It is the latter kind of control that an observational test applies to scientific epistemology. This may seem superfluous to those who never reason incorrectly. But perhaps even the most confident philosopher will admit that there are some of his opponents to whom such control would be salutary. I have little doubt that every one of the philosophical conclusions in this book has been anticipated by one of the schools of philosophy— and emphatically condemned by another. But to those who recognize them as familiar truisms or as long-condemned fallacies, I would point out that they are now put forward with altogether new sanctions which ought to be reckoned with.

Theoretical physicists, through the inescapable demands of their own subject, have been forced to become epistemologists, just as pure mathematicians have been forced to become logicians. The invasion of the epistemological branch of philosophy by physics is exactly parallel to the invasion of the logical branch of philosophy by mathematics. Pure mathematicians, having learnt by experience that the obvious is difficult to prove—and not always true—found it necessary to delve into the foundations of their own processes of reasoning; in so doing they developed a powerful technique which has been welcomed for the advancement of logic generally. A similar pressure of necessity has caused physicists to enter into epistemology, rather against their will. Most of us, as plain men of science, begin with an aversion to the philosophic type of inquiry into the nature of things. Whether we are persuaded that the nature of physical objects is obvious to commonsense, or whether we are persuaded that it is inscrutable beyond human understanding, we are inclined to dismiss the inquiry as unpractical and futile. But modern physics has not been able to maintain this aloofness. There can be little doubt that its advances, though applying primarily to the restricted field of scientific epistemology, have a wider bearing, and offer an effective contribution to the philosophical outlook as a whole.

Formally we may still recognise a distinction between science, as treating the content of knowledge, and scientific epistemology, as treating the nature of knowledge of the physical universe. But it is no longer a practical partition; and to conform to the present situation scientific epistemology should be included in science. We do not dispute that it must also be included in philosophy. It is a field in which philosophy and physics overlap.


So long as a scientific writer on philosophy confines himself to scientific epistemology, he is not outside the borders of his own subject. But most authors have felt that they could usefully advance farther and consider the general philosophical bearing of the new conceptions. This venturesomeness has been strongly criticised; but it seems to me that the critics have failed to grasp the situation.

It is recorded that Archbishop Davidson, in conversation with Einstein, asked him what effect he thought the theory of relativity would have on religion. Einstein answered: "None. Relativity is a purely scientific theory, and has nothing to do with religion." In those days one had to become expert in dodging persons who were persuaded that the fourth dimension was the door to spiritualism, and the hasty evasion is not surprising. But those who quote and applaud the remark as though it were one of Einstein's most memorable utterances overlook a glaring fallacy in it. Natural selection is a purely scientific theory. If in the early days of Darwinism the then Archbishop had asked what effect the theory of natural selection would have on religion, ought the answer to have been "None. The Darwinian-theory is a purely scientific theory, and has nothing to do with religion"?

The compartments into which human thought is divided are not so water-tight that fundamental progress in one is a matter of indifference to the rest. The great change in theoretical physics which began in the early years of the present century is a purely scientific development; but it must affect the general current of human thought, as at earlier times the Copernican and the Newtonian systems have done. This alone would seem to justify the scientific authors in taking a broad view of their task. It seems to me unreasonable to maintain that the working out of these wider implications of the new conception of the physical universe should be left entirely to those who do not understand it.

Not so very long ago the subject now called physics was known as "natural philosophy". The physicist is by origin a philosopher who has specialised in a particular direction. But he is not the only victim of specialisation. By the breaking away of physics the main body of philosophy suffered an amputation. In practice, if not in theory, academic philosophy has also become specialized, and is no longer coextensive with the system of thought and knowledge by which we orient ourselves towards our moral and material environment. To a man's philosophy in the broadest sense— to his religio vitae—natural philosophy, under the name of science, has continued to be a powerful, perhaps even a predominant, contributor. It would be difficult to point to any development in academic philosophy which has had so great an influence on man's outlook as the growth of the scientific theory of evolution. In the last twenty years it has been the turn of physics to reassert itself as natural philosophy; and I believe that the new contribution of physical science, if fully grasped, is not less significant than the doctrine of evolution.

We may define rather more closely the status of a scientist who writes on the philosophical outcome or modern physical theories. I do not think that any school of philosophers is prepared to wash its hands of the physical universe and leave the physicists to make what they like of it. It seems therefore to be agreed that scientific epistemology is still an integral part of philosophy. Those whose work lies in the epistemological developments of modern physics must therefore be counted as specialists in one of the departments into which philosophy is divided—a department not far from the heart of the subject. In their discussion of philosophy as a whole they are likely to display the faults of a specialist who finds himself outside his own groove; but they are not common intruders. The evils of specialization would, I think, be still more pronounced if they made no attempt to correlate with the rest of philosophy the progress that has been made in their own department.

Scientific epistemology is the main theme of these lectures. We shall consider it primarily from the scientific aspect. But we shall also at times endeavour to view it in its general setting as a region of overlap of physics and philosophy, and trace its consequences in both fields.


For the truth of the conclusions of physical science, observation is the supreme Court of Appeal. It does not follow that every item which we confidently accept as physical knowledge has actually been certified by the Court; our confidence is that it would be certified by the Court if it were submitted. But it does follow that every item of physical knowledge is of a form which might be submitted to the Court. It must be such that we can specify (although it may be impracticable to carry out) an observational procedure which would decide whether it is true or not. Clearly a statement cannot be tested by observation unless it is an assertion about the results of observation. Every item of physical knowledge must therefore be an assertion of what has been or would be the result of carrying out a specified observational procedure.

I do not think that anyone—least of all, those who are critical of the modern tendencies of physics—will disagree with the first axiom of scientific epistemology, namely that the knowledge obtained by the methods of physical science is limited to observational knowledge in the sense explained above. We do not deny that knowledge which is not of an observational nature may exist, e.g. the theory of numbers in pure mathematics; and non-committally we may allow the possibility of other forms of insight of the human mind into a world outside itself. But such knowledge is beyond the borders of physical science, and therefore does not enter into the description of the world introduced in the formulation of physical knowledge. To a wider synthesis of knowledge, of which physical knowledge is only a part, we may perhaps correlate a "world" of which the physical universe is only a partial aspect. But at this stage of our inquiry we limit the discussion to physical knowledge, and therefore to a physical universe from which, by definition, all characteristics which are not the subject of physical knowledge are excluded.

A distinction is commonly made between observational and theoretical knowledge; but in practice the terms are used so loosely as to deprive the classification of all real significance. The whole development of physical science has been a process of combining theory and observation; and in general every item of physical knowledge—or at least every item to which attention is ordinarily directed—has a partly observational and partly theoretical basis. The distinction, so far as it can be made, has reference to the mode of obtaining the knowledge —to the nature of the evidence for its truth. It does not concern the knowledge itself— what it is we intend to assert. Thus our axiom that all physical knowledge is of an observational nature is not to be understood as excluding theoretical knowledge. I know the position of Jupiter last night. That is knowledge of an observational nature; it is possible to detail the observational procedure which yields the quantities (right ascension and declination) which express my knowledge of the planet's position. As a matter of fact I did not follow this procedure, nor did I learn the position from anyone who had followed the procedure; I looked it up in the Nautical Almanac. That gave me the result of a computation according to planetary theory. Present-day physics accepts that theory and all its consequences; that is to say, it admits the calculated position as a foreknowledge of the results which would be obtained by carrying out the recognized observational procedure. Of my two pieces of knowledge, namely knowledge of the results of a mathematical computation and foreknowledge of the results of an observational procedure, it is the latter which I assert when I claim to know the position of Jupiter. If, on submission to the Court of Appeal, my foreknowledge of the result of the observational procedure proves to be incorrect, I shall have to admit that I was mistaken and did not know the position of Jupiter; it will be no use my urging that my knowledge of the result of the mathematical computation was correct.

It is the essence of acceptance of a theory that we agree to obliterate the distinction between knowledge derived from it and knowledge derived from actual observation. It may seem one-sided that the obliteration of the distinction should render all physical knowledge observational in nature. But not even the most extreme worshipper of theory has proposed the reverse—that in accepting the results of an observational research as trustworthy we elevate them to the status of theoretical conclusions. The one-sidedness is due to our acceptance of observation, not theory, as the supreme Court of Appeal.


We have seen that every item of physical knowledge, whether derived from observation or theory or from a combination of both, is an assertion of what has been or would be the result of carrying out a specified observational procedure. Generally it is an assertion of what would be the result if an observation were made; for this reason it is more accurate to describe physical knowledge as hypothetico-observational2 Occasionally the hypothetical form can be dropped—the observation has been made and the result obtained— but the proportion of knowledge to which this applies is small, and mostly uninteresting. I am not denying the importance of actual observation as a source of knowledge; but as a constituent of scientific knowledge it is almost negligible. Whenever in the process of reducing observations a "correction" is applied, observational knowledge of an actual experiment is replaced by hypothetico-observational knowledge of what would have been the result of an experiment under more ideal conditions.

Consider, for example, our knowledge that the distance of the moon is about 240,000 miles. The exact meaning of this assertion must be ascertained by reference to the definition of distance in physics and astronomy (Chapter v); but, accurately enough for present purposes, what we claim to know is that 240,000 x 1760 yard-sticks placed end to end would reach from here to the moon. This is hypothetico-observational knowledge; for certainly no one has carried out the experiment. It is true that actual observations were employed in arriving at the figure 240,000 miles; but apart from theory we should not know that the resulting quantity was the distance of the moon. There are a variety of practical methods of finding the distance; one of the most accurate involves inter alia swinging pendulums in different latitudes on the earth. Although it would be true to assert that 240,000 miles is the result of an actual observational procedure of swinging pendulums, etc., that is not what we intend to assert when we say that the distance of the moon is 240,000 miles. By employing accepted theory we have been able to substitute for the actual observational procedure a hypothetical observational procedure which would yield the same result if it were carried out. The gain is that hypothetico-observational knowledge can be systematized and gathered into a coherent whole, whereas actual observational knowledge is sporadic and desultory.

One cannot help feeling a misgiving that hypothetico-observational knowledge is not entirely satisfactory from a logical standpoint. What exactly is the status of conditional knowledge if the condition is not fulfilled? Can any sense at all be attributed to a statement that if something, which we know did not happen, had happened, then certain other things would have happened? Yet I cannot help prizing my knowledge that 240,000 x 1760 yard-sticks would reach from here to the moon, although there is no prospect that they ever will do so.


Scientific study of the facts of observation has led us to make a number of generalisations which we call laws of nature. Generalisation is the most conspicuous source of the hypothetico-observational character of physical knowledge, since it flagrantly oversteps actual observation and asserts knowledge of what would be observed on any occasion if the necessary procedure were carried out.

I think it is sometimes maintained that a law of nature is a systematization, not a generalization, of knowledge. Ideally it is possible to accept a systematization of existing observational knowledge, without prejudging whether any future observations will conform to the system. To a person holding this view it should be a complete surprise every time a new observation is found to obey the law. For example, Bode's law of planetary distances can be regarded as a systematized statement concerning the distances of the six planets known in his time, and not expected to apply to planets discovered subsequently. This may be the right attitude to adopt towards particular laws that have been enunciated; but it certainly cannot be applied generally throughout physics. We must not imagine that Bode's systematization would continue to be possible in a physics purged of generalizations Unless we accept certain prior generalizations, e.g. that light travels in straight lines, the distances of the planets cannot be determined; and Bode's law then drops out, because there is no material for it to systematize The fact is that generalization from observation has, whether adver-tendy or not, been practiced in physical science from the very beginning; and we must regard it as no less part of the scientific method than observation itself. And with the generalizations there has entered into the body of scientific knowledge a hypothetico-observational element, which has been conceded the right to remain.

Our main conclusion is that, notwithstanding the diversity of method, physical knowledge remains homogeneous in its nature; it is knowledge of what would be the result of an observational procedure if it were carried out, including as a special case the result of any observational procedure that has been carried out.

In the progress of physics individual facts have become largely merged in generalizations Would it be true to say that complete knowledge of physics would consist wholly of such generalizations? The answer is different according as we refer to physics in a narrow sense (including chemistry, but not astronomy or other observational, as distinguished from experimental, sciences) or to physical science generally. In the narrow sense physics is, I think, concerned solely with generalizations The physicist is not interested in special facts except as material for generalization If he studies a particular lump of iron, it is as a sample exhibiting the general properties of iron. The astronomer on the other hand is interested in the particular lump of matter on which we happen to live, whether or not it is a sample of planets generally. He is curious about the existence of vegetation on Mars, when the next bright comet will appear, how closely a minor planet approached the earth, and so on. It may be said that this is just an amateurish interest which the more serious-minded physicist has outgrown; an astronomer must, of course, ascertain the constants of the earth, as a physicist must ascertain the constants of his galvanometer, but he has no business to be interested in them. Astronomers will scarcely agree; but let that pass. It is sufficient to say that these special facts are knowledge acquired by the methods of physical science, and must not be neglected in scientific epistemology which we have defined as the study of the nature of the knowledge that has come to us in this way; nor are they negligible in the universe of which that knowledge forms a description.

We must therefore remember that not all our knowledge of the physical universe is comprised in knowledge of the laws of nature. The warning is not so superfluous as it seems. I have often found an impression that to explain away the laws of nature as wholly subjective is the same thing as to explain away the physical universe as wholly subjective. Such a view is altogether unfounded.

1 "Right-thinking person" is, of course, a modest way of referring to oneself.

2 "Hypothetico-observational knowledge" means knowledge of the result of a hypothetical observation, not hypothetical interpretation of die result of an actual observation




Let us suppose that an ichthyologist is exploring the life of the ocean. He casts a net into the water and brings up a fishy assortment. Surveying his catch, he proceeds in the usual manner of a scientist to systematise what it reveals. He arrives at two generalisations:

(1) No sea-creature is less than two inches long.

(2) All sea-creatures have gills. These are both true of his catch, and he assumes tentatively that they will remain true however often he repeats it.

In applying this analogy, the catch stands for the body of knowledge which constitutes physical science, and the net for the sensory and intellectual equipment which we use in obtaining it. The casting of the net corresponds to observation; for knowledge which has not been or could not be obtained by observation is not admitted into physical science.

An onlooker may object that the first generalization is wrong. "There are plenty of sea-creatures under two inches long, only your net is not adapted to catch them." The ichthyologist dismisses this objection contemptuously. "Anything uncatchable by my net is ipso facto outside the scope of ichthyological knowledge, and is not part of the kingdom of fishes which has been defined as the theme of ichthyological knowledge. In short, what my net can't catch isn't fish." Or—to translate the analogy—"If you are not simply guessing, you are claiming a knowledge of the physical universe discovered in some other way than by the methods of physical science, and admittedly unverifiable by such methods. You are a metaphysician. Bah!"The dispute arises, as many disputes do, because the protagonists are talking about different things. The onlooker has in mind an objective kingdom of fishes. The ichthyologist is not concerned as to whether the fishes he is talking about form an objective or subjective class; the property that matters is that they are catchable. His generalization is perfectly true of the class of creatures he is talking about—a selected class perhaps, but he would not be interested in making generalizations about any other class. Dropping analogy, if we take observation as the basis of physical science, and insist that its assertions must be verifiable by observation, we impose a selective test on the knowledge which is admitted as physical. The selection is subjective, because it depends on the sensory and intellectual equipment which is our means of acquiring observational knowledge. It is to such subjectively-selected knowledge, and to the universe which it is formulated to describe, that the generalizations of physics— the so-called laws of nature—apply.

It is only with the recent development of epistemological methods in physics that we have come to realise the far-reaching effect of this subjective selection of its subject matter. We may at first, like the onlooker, be inclined to think that physics has missed its way, and has not reached the purely objective world which, we take it for granted, it was trying to describe. Its generalisations, if they refer to an objective world, are or may be rendered fallacious through the selection; But that amounts to condemning observationally grounded science as a failure because a purely objective world is not to be reached by observation.

Clearly an abandonment of the observational method of physical science is out of the question. Observationally grounded science has been by no means a failure; though we may have misunderstood the precise nature of its success. Those who are dissatisfied with anything but a purely objective universe may turn to the metaphysicians, who are not cramped by the self-imposed ordinance that every assertion must be capable of submission to observation as the final Court of Appeal. But we, as physicists, shall continue to study the universe revealed by observation and to make our generalizations about it; although we now know that the universe so reached cannot be wholly objective. Of course, the great mass of physicists, who pay no attention to epistemology, would have gone on doing this in any case.

Should we then ignore the onlooker with Ins suggestion of selection? I think not; though we cannot accept his remedy. Suppose that a more tactful onlooker makes a rather different suggestion: "I realise that you are right in refusing our friend's hypothesis of uncatchable fish, which cannot be verified by any tests you and I would consider valid. By keeping to your own method of study, you have reached a generalisation of the highest importance—to fishmongers, who would not be interested in generalisations about uncatchable fish. Since these generalisations are so important, I would like to help you. You arrived at your generalisation in the traditional way by examining the fish. May I point out that you could have arrived more easily at the same generalisation by examining the net and the method of using it?"

The first onlooker is a metaphysician who despises physics on account of its limitations; the second onlooker is an epistemologist who can help physics because of its limitations. It is just because of the limited—some might say, the perverted—aim of physics that such help is possible. The traditional method of systematic examination of the data furnished by observation is not the only way of reaching the generalizations valued in physical science. Some at least of these generalizations can also be found by examining the sensory and intellectual equipment used in observation. Epistemology thus presents physics with a new method of achieving its aims. The development of relativity theory, and the transformation of quantum theory from an empirical to a rational theory arc the outcome of the new method; and in it is our great hope of further fundamental advances.


We return to our fish to illustrate another point of great importance. No suggestion was offered as to the second generalisation— that all sea-creatures have gills—and, so far as we can see, it could not have been deduced from an examination of the net and its mode of use. If the ichthyologist extends his investigations, making further catches, perhaps in different waters, he may any day bring up a sea creature without gills and upset his second generalisation. If this happens, he will naturally begin to distrust the security of his first generalisation. His fear is needless; for the net can never bring up anything that it is not adapted to catch.

Generalisations that can be reached epistemologically have a security which is denied to those that can only be reached empirically.

It has been customary in scientific philosophy to insist that the laws of nature have no compulsory character; they are uniformities which have been found to occur hitherto in our limited experience, but we have no right to assert that they will occur invariably and universally. This was a very proper philosophy to adopt as regards empirical generalisations—it being understood, of course, that no one would be so foolish as to apply the philosophy in practice. Scientists, assured by their philosophy that they had no right to expectations, continued to cherish indefensible expectations, and interpreted their observations in accordance with them. Attempts have been made by the theory of probability to justify our expectation that if an occurrence (whose cause is unknown) has happened regularly hitherto it will continue to happen on the next occasion; but I think that all that has emerged is an analysis and axiomatisation of our expectation, not a defence of it.

The situation is changed when we recognise that some laws of nature may have an epistemological origin. These are compulsory; and when their epistemological origin is established, we have a right to our expectation that they will be obeyed invariably and universally. The process of observing, of which they are a consequence, is independent of time or place.

But, it may be objected, can we be sure that the process of observing3 is unaffected by time or place? Strictly speaking, no. But if it is affected—if position in time and space or any other circumstance prevents the observational procedure from being carried out precisely according to the recognised specification—we can (and do) call the resulting observation a "bad observation". Those who resent the idea of compulsion in scientific law may perhaps be mollified by the concession that, although it can no longer be accepted as a principle of scientific philosophy that the laws of nature are uncompulsory, there is no compulsion that our actual observations shall satisfy them, for (unfortunately) there is no compulsion that our observations shall be good observations.

What about the remaining laws of nature, not of an epistemological origin, and therefore, so far as we know, non-compulsory? Must they continue to mar the scheme as a source of indefensible expectations, which nevertheless are found to be fulfilled in practice? Before worrying about them, it will be well to wait till we see what is left of the system of natural law after the part which can be accounted for epistemologically has been removed. There may not be anything left to worry about.