ON NOT WORSHIPPING FACTS
[From The Philosophical Quarterly, 8, 1958, pp.144-156.]
My sights in this paper are trained on facts. Most people think that they know what facts are; that while their friends often, and themselves occasionally, are ignorant of the facts, at least they know what sort of things facts are---they can recognise a fact when they see it. Facts, in the popular philosophy of today, are good, simple souls; there is no guile in them, nor any room for subjective bias, and once we have made ourselves acquainted with them, we have reached the beginning and summit of all wisdom.
This view is false; and not only false, but dangerously false. Facts are not at all what people think they are; they are not the simple solid elements out of which the whole fabric of our knowledge is constructed, and the belief that they are is responsible for many of the obsessions which afflict academics, not least the historians.
Facts are not simple definite entities, because the word `fact' is not a simple substantive, but rather one that is systematically ambiguous. The words `possible', `impossible' and `necessary', `can', `cannot' and `must', are systematically ambiguous. When we say ``A prime number can be even but cannot be a perfect square", we are using can' and `can-not' in the sense of `logically can' and `logically cannot' When we say that entropy can increase, but not decrease, or that bats can fly but pigs cannot, we are speaking of physical possibilities and impossibilities. And there are many other senses: political---A government can break an election pledge, but cannot survive a vote of no confidence; social---one can tell a white lie, but cannot give one's real reasons for refusing invitations; and moral---one can give away money which is one's own, but cannot appropriate for charity what belongs to another. In each of these examples the words `can', `cannot', `possible', `impossible' have a different sense, and therefore no contradiction arises from our being able to say that it is physically impossible for pigs to fly, and, in the same breath, that it is logically possible---it would not be a contradiction in terms to talk of a pig's flying as it would to talk of a prime number's being a perfect square. Similarly, the word `fact' is systematically ambiguous: its meaning, though definite, varies with its context; and because it depends on its context in a very complex way, the concept of a fact is not viable when withdrawn from its normal habitat. We should be chary of making facts the cornerstone of our thinking, since at the higher levels of abstraction where philosophers talk about Acquaintance with Facts, and discuss the relations between True Propositions and Facts, too little context remains for any determinate meaning to survive. There are other words better suited to be philosophical specimens, which can stand being isolated and examined in vacuo, without  losing their sense: it is in terms of these that a philosopher, when tempted to say something about Facts, should re-phrase what he has to say, since then he is less likely to be misled.
That the meaning of the word `fact' is not the same in all cases can be brought out by Aristotle's Method of Opposites. We ask ``What is it being contrasted with? Is it a fact as opposed to a fiction? Or as opposed to a theory? Or as opposed to an interpretation? Or a question of fact as opposed to a question of law?'' This in itself is enough to show that there is no one unitary concept of a fact, but rather a whole sheaf of concepts, bound together indeed, but distinct. More formal proofs can be produced: the fact/non-fact contrast may be used twice in the same sentence to make different distinctions; so that what are correctly described as facts according to one contrast are equally correctly not described as facts according to the other. Thus, in science fiction we read of experiments to determine the curvature of time; the patient labours of numerous research workers in the lunar laboratories yield a vast mass of data, but it is left for Professor Bronsheim to organise these facts into a coherent theory by his brilliant hypothesis of the aberration of the feminine constant. Again, when we are passing moral judgements, both questions of law and questions of fact are facts in contrast to the set of values according to which we judge the situation. It was a fact that I was proceeding down the Cherry Hinton Road at 48 mph, and it is a fact that it is a built-up area, and that there is a law forbidding speeds in excess of 30 mph in built-up areas. Whether I am morally at fault is another matter, depending not merely on the facts of the case but on questions of moral principle as well; whether, for instance, it is necessary to obey the law absolutely at all times, or whether there are some enactments which in some circumstances need not be too rigorously observed. These two examples give us apparent contradictions in terms; fictional facts, which are facts because they are not theories, and not facts because they are fiction; and legal facts, which are facts because they are not moral principles or evaluations, and are not facts because they are laws or interpretations of laws. Hence, using the Law of Non-contradiction and a Reductio ad Absurdum, we conclude that in each case the two instances of the word `fact' are being used in different senses. This completes the formal proof of my claim that the word `fact' is ambiguous, with the consequence that there is no unitary concept of a Fact.
The word `fact' is ambiguous, but it is obviously not coincidentally ambiguous like the word `jet', which can mean either a propulsive mechanism or a black semi-precious mineral. There is evidently some system in the ambiguities of the word `fact', which I shall attempt to lay bare. I  shall do it in two stages: first I shall give a schematic account of how we use the word `fact', which is only approximately correct, but contains, I believe, the nerve of its meaning; then I shall go on to show how the simple model can be extended to give us all the present senses of the word.
To explain my first crude version of its meaning, I shall need to set it in the context of a dialogue, a discussion, or a dispute. I make no apology for introducing this among the concepts of philosophy, although it is unhallowed by tradition. Language is used normally as a means of communication between people; apart from the monologues of bores and the highly studied activity of literary composition, it is a reciprocal activity. Philosophers, who spend most of their time communing with themselves, are naturally tempted to forget that other people exist, and if talked to will talk back. They too readily assume that language was created by each individual philosopher separately for his private purposes alone, and functions to portray or represent for his solitary delectation such experiences as he can conjure up in his imagination, or has in the past enjoyed. For them language is a calculus or a map. Many errors in the theory of meaning have resulted, and many misapprehensions of the sense of particular words, one such being the word `fact', whose functioning can be correctly understood, I claim, only in the context of a dialogue or dispute.
In a dispute there is always some point at issue: but in any discussion that is to continue, there must also be many points which are common ground. It is fruitless to argue if we do not disagree about something: it is fruitless, too, to argue if we disagree about everything. We must start with some points of agreement if our discussion is to get us anywhere, and only by taking them as agreed and unquestionably true can we hope to reach agreement over the point at issue. These points of agreement we call the facts. On the basis of these we argue and may succeed in reaching a conclusion. If we do reach a conclusion, then this point now agreed between us will be a fact in any further dispute the pair of us may have. That is to say, a fact is a fact relative to a given dispute, or relative to two or more persons at a given time arguing about a given point. The points that both sides accept as true, each side will describe by the word `fact': points whose truth one side would challenge should not be called facts, unless their truth can be established on the basis of other facts, premisses that is, which are conceded as unquestionably true. The word `fact'is an incomplete symbol; the complete locution being `facts in respect of such and such a dispute'. Before we can answer the question ``What are the facts?'' we need to know, either from the context or by being told explicitly, with respect to what dispute the question is being asked. This is the fundamental reason why we cannot talk of Facts with a big F: the word `fact' is an incomplete symbol, and as the issue in dispute varies, so also will the facts.
Let me take some examples. In a court of law there may be two sorts of issue in dispute: issues of fact and issues of law. In most criminal cases, it is a question only of determining what actually happened---a question  of fact; while in many civil cases, the sole difficulty is to determine the correct interpretation of the law---a question of law; and there are yet other cases in which the court has to decide both a question of fact and a question of law. There will be rival accounts of what happened, but some things, different in each case, will not be contested by either party: both plaintiff and defendant, for example, might agree that plaintiff had chartered a ship from the defendants upon certain terms, and the ship had taken a cargo on board at Buenos Aires and discharged it at Pembroke Dock. They disagree on other points---the condition of the cargo before it was taken on board, whether it was properly examined in South America, when the deterioration took place, and what was the cause; but there are some points they do agree about; and on the basis of these and other facts, the court reaches a conclusion about what actually happened. This question having been settled, it is no longer a point in issue but becomes a fact with respect to the next question the court has to address itself to, namely what the legal consequences are; that is to say, how the law lays it down that cases like this are to be treated.
In history there are many issues. If we are arguing about the date of the battle of Marathon, the relevant facts are the date of Salamis, chronological statements made by Herodotus and Thucydides, an odd phrase in the Persae of Aesehylus, a scandal reported by Plutarch; but if the question be different the facts will be different too: in establishing the text of the Persae, the evidence of papyri from Oxyrrhynchus and of palimpsests from Stuttgart might be relevant, and it might need careful consideration to decide, in the light of them, how the text originally read; and while we were discussing this point, we would be keeping as an open, though decidable, question, the point which we took for granted in discussing the date of Marathon. The historian's fact is merely the textual critic's conclusion. Similarly in the other direction, in a discussion of Miltiades' life or Themistocles' policy, the date of Marathon might be assumed without argument; and it would not be in this discussion, as it would be when the date itself was the topic of dispute, a case of begging the question.
A similar variableness appears in value-judgements. Very seldom is the distinction between facts and values either as sharp as philosophers like to think, or drawn where they think it ought to be drawn. Often the facts which we adduce to support an evaluative conclusion, are not absolutely non-evaluative themselves. In a dispute about a man's moral worth we claim, and it will be conceded to us, that at least this action was generous and that just, and it will be claimed against us, and we shall concede, that some other deed was inexcusable and yet another difficult to defend. Again, in discussing the morality of euthanasia, we do not begin from facts purely `descriptive'; our starting point will be a mixture of `factual' facts that there are some painful diseases that are known to be incurable, some  general moral principles---that it is wrong to kill and wrong not to alleviate unnecessary suffering, (these are sometimes called `moral facts'), and some more specific common moral judgements---that to kill a person in such and such circumstances would be to murder him, and all these are usually described by non-philosophers as facts. Only in the abstract examples invented by philosophers is the sense of the word `fact' fixed, and only there can it be equated to that which `descriptive' statements describe.
Even scientists determine what constitutes a fact by reference to the questions in dispute. The Theory of Evolution, The Theory of Special Relativity, and Quantum Mechanics, started by being speculations, then became hypotheses, then were well-founded theories, and now can be described as facts. They are the starting point for further discussion. All, or nearly all, fruitful discussions in the natural sciences must start by assuming them; we should never make any progress if we devoted all our energies to establishing that which needed no establishing, since it had already been established and never been brought in question again. To say this is not to rule out the possibility of reopening any of the questions that have already been solved. A competent biologist, impressed by certain difficulties and certain features the theory fails to account for, e.g. the phenomenon of Pre-adaptation, might legitimately express his doubts, and even bring the whole theory into question. Similarly, if the Michelson-Morley experiment, repeated with more exact instruments of finer discrimination, showed that after all there was a difference in the velocities of light in different directions, we should cease to regard Relativity as an established fact. Meanwhile, in the absence of counter-evidence, there is no need to re-open the question of the truth of these theories, and it remains proper to refer to them, normally, as facts. When, however, the discussion is somewhat abnormal, we may have an altered view of what constitutes a fact. When we are illustrating scientific method we do not say that the theories of evolution, relativity, or wave-mechanics, are facts, however well-established they are, for there we are concerned with the nature of scientific argument, and the facts in that context are not the conclusions of our inferences---the theories, but the observations on which those conclusions were based and from which we are able to argue. The facts there are contrasted with what we can argue about; they are our starting point, which we can not profitably talk about in that discussion. They may be observable general features of the world, which we can verify by using the correct procedures and observing for ourselves---here it is no use talking, we can only look and see. The facts of chemistry are like this. Often however, the facts are not easily reproducible; sometimes, as in astronomy, not reproducible at all. But they have been made by honest and competent investigators, and published in reputable journals, and so it is usually fruitless to question their results. It is, however, always possible to question them, and occasionally necessary. Then the discussion becomes either one of experimental technique---``from his own account it is obvious that he forgot to allow for diurnal variations in the earth's magnetic field''  or one of historical fact---``they say that when they measured it they found that the bending of the star's rays by the sun was 1' 75'' of an are, but can we trust them? There is evidence that we can not''---or, perhaps, for a philosopher, a discussion in the philosophy of history---``On this printed page before me, which is headed 1919, it says . . .; from this I infer . . . ''.
Philosophers who have wanted to reconstruct our scientific disciplines in terms of `basic' facts, have been worried by a type-token ambiguity in the scientist's use of the word fact. Are we right in calling it a fact that pure water is most dense at 4oC, or ought we rather to talk of the several distinct facts that this and this and this specimen of pure water are each at their most dense at 4oC? If we say the latter we find it difficult to carry scientists with us, and difficult to maintain our claim to clear and simple common sense: if the former, we are subscribing to general facts of a most un-nominalistic kind, and creating as many difficulties as we hoped to solve. The worry disappears once we cease to regard facts as entities of some special sort, and see them as agreed starting points for an argument. Some arguments, as in school chemistry, start from general, reproducible, features others, as in astronomy, from features which are not reproducible at all and many from features which are in principle reproducible, but are so difficult or tedious to reproduce that we are content to accept the word of other, reputable men, who report their results in the learned journals.
The word `fact' has often been construed in terms of the word `true'. Facts, it is said, are what true statements state. I want as a first approximation to amend the definition by the addition of the qualification `unquestioned', or `established', or `accepted'. The facts in a dispute are the established and accepted truths, what those true statements state which are unquestioned by either side. Since my definition still contains the evaluative word `true', we must guard against the Naturalistic Fallacy. It does not follow from the two disputants' agreeing with each other on some point, that that point is true. Two Fascists disputing whether England or America was the more deplorable place might do this on the agreed basis that both countries were run by Jews; and we would not call that a fact. Though both parties believing it to be true and agreeing to its truth between each other would call it a fact, we know it is not true and therefore cannot really be a fact. Agreement is not enough to establish truth. The difficulty here is that which besets all attempts to lay down rules for the use of evaluative words: no rules can completely exhaust the meaning of the word, and however many criteria have been satisfied it may yet be that the word has been incorrectly applied. Evaluative terms, such as `good' (in its moral contexts) and `true', are infinitely complex, and any particular application is always logically open to revision. Whenever we say of anything that it is a fact it is always logically possible that it might not be true and therefore not a fact; though we are not on that account unjustified in believing to be true what we believe to be true. We can no other. But to be justified in  calling something a fact, it is not enough that we should believe it to be true; it must be conceded to be true by the other party also.
I shall now go beyond my first approximation and alter my definition in two ways to make it more exact. First, I replace the reference to actual parties in actual disputes by ideal parties in actual or ideal disputes. A fact is what a disputant would concede as true if he were a reasonable man living at that time. Not what an unquestioned true statement states---some people will question anything---but what an unquestionably true statement states. This alteration weakens the dependence of the facts upon the issue in dispute, and substitutes an implicit reference to time. Many issues are dead. The reasonable man does not spend much time reassuring himself that the battle of Hastings was fought in 1066 A.D. or the battle of Marathon in 490 B.C.. Although it is logically possible to impugn those dates, although indeed the date of Marathon has been impugned, yet the arguments in favour of the traditional dates are strong, and the consensus of opinion among those who have gone into the matter is so overwhelming that the reasonable man may feel fairly confident that he would not disagree with their findings.
Scientists, for all their belief in experiments, do not conduct an experiment without first consulting the `literature' in case the experiment has already been carried out, in which case it need not be carried out again unless there is some reason to suspect that the findings would not be confirmed. Therefore whatever the dispute, those findings may be cited without fear of challenge. They are facts with respect to any dispute that is likely to arise, and so with them it is not necessary to specify with respect to what dispute they are said to be facts. The incompleteness of the word `fact' fails to matter in such cases, since the incomplete symbol can be completed in any way we please without making any alteration to its sense. Only in the peculiar case, when we are throwing a well-established truth in doubt as an academic exercise in order to practise re-establishing it, may we no longer regard those facts as facts. When for our weekly essay we are set the task of determining the date of Marathon, then, for obvious reasons, the normal conditions do not apply, and we cannot beg the question by citing the findings of nearly all reputable historians as an established fact. At all other times, when the questions set us are not set us as academic exercises, we are at liberty to assume all that a reasonable man, well informed and abreast of current discoveries, would concede us. This, of course, changes over long periods of time. The store of established truth is increased by the labours of each generation; and also is winnowed by the criticism of careful minds, so that some of what to one generation seemed established truths are seen by subsequent generations to be less well assured, and even sometimes to be false. Therefore in extending the sense of the word `fact' to cover the corpus of indisputable propositions accessible not merely to a particular pair of disputants, but to disputants in general, we retain an implicit time reference. Truth is timeless, and we cannot ask when a true  statement became true: but being established is a temporal process, and we can ask when a truth became so well established that it could not reasonably be questioned. Hence it is that the conjectures of one generation can become the facts of the next, and what the last generation asserted as facts may be accounted their received falsehoods in this.
The first way in which the word `fact' has been extended in meaning has been such that it should cover the corpus of assured knowledge in the possession of each generation; the second way in which it is extended is so as to mark certain philosophical, and particularly epistemological, distinctions. Here there is no time-reference. Here we are seizing upon and making firmer the epistemological distinction we came across earlier, between those issues about which it is worth arguing, and those where the proper method of verification is something else, and argument is out of place. In some of the natural sciences this gives us the distinction between fact and theory. The fact versus value distinction reflects the characteristic feature of evaluative disputes, that with them there is, because of the autonomy of our value judgements, always a possibility of further argument. Plato makes this point in the Phaedrus. Over some questions, as whether the given specimen is iron or silver, there is little possibility of dispute, and it can be settled beyond doubt in a finite number of steps; over others, such as whether the action under consideration was right or good, there is no certain agreement. With silver I can perform tests; dissolve it in nitric acid, form a precipitate with a chloride, and watch it go blue in the light; if, having observed all these, I still doubt whether it is really silver, I show that I do not know the meaning of the word `silver' perhaps I was meaning by `silver' what everybody else means by `gold' whereas if I agree with you completely about what the government did, but maintain that they were right in spite of your strong conviction that they were wrong, I do not show that I do not know the meaning of the word `right'. The word `factual' here describes those questions whose truth or falsity can be definitely established, as opposed to those for which, though they may be true or false, no agreed decision-procedure exists. Similarly in history, there are some questions, such as the date of the battle of Marathon, which, though we may lack the evidence actually to decide them, are of a sort for which a definite decision is possible. Other questions, the imputation of motives, the attribution of causes, the distribution of emphasis, cannot be so definitely settled by historians. The former we call questions of fact, in contrast to the interpretations, assessments and judgements over which reasonable men might well differ. It is in this sense we can talk of `questions of fact', which would have been a contradiction in terms according to my first explication of the meaning of the word. A dispute over a question of fact is a relatively simple kind of dispute; we know how it could be settled in principle, only unfortunately it cannot in practice be settled here and now, either because the evidence required happens to be missing, or because though extant it is  not accessible at the time and place of the argument. Whether Danzig was or was not a town of the Hanseatic League is a question of fact. I am not sure of the answer, but in labelling it as such I am saying that it is not a matter we should argue about, but rather look up. The date of the Lelantine war is also a question of fact. Here I am not saying we should not discuss it and merely look it up, because we cannot look it up, as there is no settled answer, and the answer of each historian will depend on his whole reconstruction of early Greek history; rather, I am saying that the question is in the same class, from a philosopher's point of view, as questions about the date of Marathon or Hastings, the only difference being the philosophically uninteresting one that we have sufficient evidence to be sure about the latter, but not about the former. To put it another way, the date of the Lelantine war, if it were known, could be learned by a schoolboy; whereas questions about motives, causes and significance can be asked and answered only by mature men who have some training in history, and some understanding of men and affairs. Some questions in history are schoolboy questions, and these are questions of fact: but only a man with deep insight and wide sympathies can begin to understand the mind of a Pericles or a Pitt.
In most cases where the word `fact' is used of a definite type of statement, an epistemological distinction between the more and the less certain is being made, with the facts being those that are more certain and less disputable than statements of some other, less favoured, type. One exception is the contrast between questions of fact and questions of law. There the contrast is between matters of fact, which can be characterised in nontechnical and fairly neutral language, and questions of law, which are questions partly of deductive logic and the subsumption of particular instances under general rules, and partly of the correct characterisation of cases in a highly technical language. It is, of course, a philosophically interesting distinction between different sorts of argument, but the reason here for applying the word `fact' is not so much to flag the side of the contrast to which a greater degree of certitude attaches, as that the professional interest of lawyers leads them to mark off those, purely legal, questions for which subtlety and legal acumen are required, from those questions of fact, which, though important in each individual case, do not generalise and are not discussable. In this one instance the word `fact' has more the force of our first than of our final definition; it is the discussability of the non-facts that makes the contrast: in all the others---fact v. value, fact v. interpretation, fact v. fiction, fact v. theory---the non-discussability and non-questionability of the facts is the leading logical spirit.
``Comment is free: facts are sacred''. We are now in a position to understand why. It is not a recommendation, but a tautology. It is not a precept telling editors how to behave, but a concealed definition of the meaning of the word `fact'. Facts are what are sacred; what an honest man cannot reasonably deny, question or refuse to concede. Comment is free, fiction is free, conjecture is free, interpretation is free; in all these  there is latitude for disagreement. You may think the government's policy disastrous, I that it was wise and courageous; we cannot both be right, but it is not necessary that either of us should be a fool or a rogue. Anthony Trollope killed Mrs. Proudie to oblige a chance acquaintance; the Last Chronicle of Barset may be the worse for it; but Anthony Trollope is not a liar because he says that Mrs. Proudie died. A cosmologist who disbelieves in the theory of continuous creation may incur the wrath of other cosmologists, but we should not be justified in describing him as dishonest or depriving him of his chair. I can conclude that King John was a much maligned patriot or Richard III a sensitive soul sincerely anxious for the welfare of his wards: if I do, I cannot be dismissed out of hand, as I can if I think that Magna Carta was issued by Charlemagne, or the battle of Bosworth was a naval victory of Drake's. In each case we are contrasting the sort of statement where there is latitude and two well-informed and honest men could reasonably differ, with the sort of statement where the external controls are much more rigid, and where a man cannot have an opinion of his own without thereby revealing himself either unreasonable or dishonest. But, and this is the important point, the connection goes in the opposite way to what we thought it did. Facts do not make the reasonable man, the reasonable man makes the facts. They are what the reasonable man cannot but concede.
I have now completed my two accounts of the meaning of the word `fact', the first giving, as I believe, the basic sense of the word, the second describing the two extensions of the word to make it less dependent on context and to mark distinctions of philosophic interest.
The extensions have increased the utility of the word; also its dangers. We are liable to be torn by the different meanings of the word, in attempting, when using it in the one sense, to conform with the requirements of the other. The philosopher, anxious to achieve certitude and not to rest his case on the subjective opinions of fallible men, decides to eschew unproven conclusions and to confine himself to truths that are indubitable, that is to say, to facts. Then, conscious of the meaning of the word according to its extended sense, he concludes that facts are propositions of a certain type. They alone are worthy of his attention, and all other topics can be thrown out as so much metaphysical lumber. But then even within the selected range he finds some things that are dubious, and therefore, reverting to the first definition, not facts. So he must select even more stringently the chosen class. Unconsciously construing the word `fact' at one moment as incomplete symbol, and at the next as a term denoting a definite class of propositions, the philosopher finds endless contradictions confronting him, his whole conceptual structure distorted and absurd, and himself swept into a sea of scepticism and despair. Many false philosophies are generated here: for instance, the squirrel disease, which afflicts historians, and phenomenalism, to which philosophers are prone.
The squirrel theory of history is the theory which holds that it is the duty of the historian to gather up facts from county record libraries and  then bury them again in university libraries. It is based on the argument that since history is a reputable discipline like the natural sciences, it is itself a science, and concerned with the discovery of facts. Historians who discover facts are doing their job: historians who do not discover new facts are not really historians at all, but merely popularisers, pedagogues or journalists. Since the number of non-controversial interesting statements on historical matters is exceedingly small, it follows that most historians of the scientific school are condemned to lead dull and dismal lives, as they continue their patient accumulation of irrelevant detail. The laundry lists of a German Count in Italy in 1784, the Manorial system in three villages in South West Kent in the latter part of the 14th century, Local Government in Barnstaple 1633-1637, show that the mine of interesting subjects is perilously near exhaustion. I do not wish to decry the historian's scrupulous regard for facts: inaccurate history is not history at all: no historian can be any good without mastering large masses of detailed facts, and some historians may be right in concentrating their efforts on accumulating and establishing more facts. What I deplore is the heresy that this is the sole business of historical research, that if I succeed in showing that a charter has been mis-dated by two years, I have done well, but if I re-write the history of England over a century or more, however convincingly I recreate the motives and personalities of that time, however soundly I judge of the relative significance of the different tendencies and movements then, however coherently I work together the whole, I am not an original historian, and my work is only derivative, secondary , and second-rate. Why should a historian confine himself so? Only if he believes both that there is a special class of entities called facts, and that they alone are sacred and objective, is the squirrel theory tenable at all. If he believes that the past is naturally articulated into a number of discrete atomic facts which are objectively true, and which can be established by a non-controversial technique, then it will be reasonable for the impartial historian to record just these without embellishment or ornament, and leave it to others to gloss them and explain them as they choose. The facts are there; the pattern is in the eye of the beholder. Less consciously, a historian, distressed by the personal factors that obviously influence his colleagues' judgement, and not realising that history, being a humanity, is different in kind from the natural sciences, may yearn for the impersonal objectivity of the scientist, and decide not to stick his neck out but to confine himself to what cannot be impugned by even the youngest of his colleagues. Therefore no generalisations, only details. Then the primary meaning of the word `fact' reasserts itself, and since not all details are certain, they too cannot always be facts, and therefore, by virtue of the second meaning, never can be. Instead, the facts are the sources; the original sources. And so we get the final reduction of history to scissors and paste; the only safe thing for the historian to do is to produce an anthology of original sources, together with an index for the convenience of the reader.  it would be wrong for me to suggest that a false view of facts was the sole cause of error in our thinking about the nature of history. The bad state of the study of the classics and of modern literary criticism shows that there are other causes at work too. And Sir Lewis Namier's jaundiced view of the nature of man is as much responsible as his logical atomism for his granular approach. Nevertheless, the attempt to construe the relative term `fact' as though it were an absolute is partly responsible; in particular for the obsession that History has room for Tycho Brahes, but not for Kepler.
Some arguments for Phenomenalism turn on the unrecognised assumption that there are basic facts, a definite class of indubitable entities. Then, as in History, the reductive argument begins. Any ordinary set of statements cannot state facts, since men, being fallible, sometimes make mistakes. Besides, merely if we think hard enough about any given statement, the original sense of the word `fact' reappears. If I concentrate hard enough on why I believe there is a tree in the Quad, I cease to assume that there is a tree in the Quad, and treat that statement instead as a proposition to be proved rather than a premiss that is given. In having the question of whether there is a tree in the Quad or not brought to my notice I am being begged not to beg the question, and the courtesies of argument demand that I put in doubt what I normally know to be true. There is also what I might call the Yellow-Spot phenomenon in philosophy, namely, that if we focus our attention too hard on any matter for too long, we cease to see it straight. In the dark night of the intellect, which is the philosopher's usual state of mind, it is wise for him occasionally to distract his thoughts and look away, that he may see what he is looking at the better; more especially when he is dealing with facts and certainty. For facts are essentially what is peripheral to the question under examination, what can be taken for granted on this occasion; and therefore by being asked sufficiently earnestly to consider any question sufficiently closely I can be cajoled into giving up fact-status for this occasion for almost any statement: courtesy compels. Only if I am making the minimum possible statement can I be pushed no further: only if I say that there is in my visual field at this moment a red rectangular patch on a cream background, am I safe from possible error: hence, if there are basic facts, only the simplest facts of sense-experience can fill the bill. By attempting to make rigid and absolute the flexible standard, which depends on the circumstances, of what the honest man cannot reasonably refuse to concede, we have ensnared ourselves in a reductionist spiral, demanding an ever lower standard of reasonableness until we reach the phenomenalist's goal, the lowest common denominator of what must be conceded by every reasonable man in any circumstances whatever,that is, what must be conceded by a barely sentient being.
It is an interesting way of doing philosophy. We start by assuming that a fact is what a true statement states: from this it is a natural inference that since the conclusions of ethical debate and scientific theorising are not facts, they are not true either. By restricting our criterion of truth to  that of agreed truth, we are able to eliminate all doubt and dubiety within the province of philosophy; nor can the opponent of this view fault the examples given of what is to be allowed as really true, for only those truths that cannot reasonably be contested are put forward as examples. One weakness alone attaches to the method: as there are few facts, if any, that we cannot in our metaphysical moments be uncertain of, our concept of truth is regressive; our criterion grows progressively and indefinitely more stringent. At first we exclude those propositions of morals, theology and metaphysics, whose elimination is welcome to many of the enlightened; but the more we think, the more nice we become as to what are unquestionable truths; and so the truths of logic, mathematics, and natural science, of common sense and everyday life, join the procession to the guillotine.
For there are no basic facts: there are only facts relative to a dispute. Since there is nothing that cannot on some occasion be reasonably doubted, there can be no truths established beyond doubt to all comers, no elemental facts which we just have to accept and on which all else is based. Nothing is never doubtful, though this is not to say that everything is always doubtful. In every dispute we have to start somewhere, though there is nowhere that is the starting point for every dispute.
This is a little too strong: though there is nothing we cannot doubt, there are many things that, apart from our metaphysical moments, we do not doubt: there is a core of accepted truths that are unquestioned by all people living at a given time, and unquestionable by any reasonable man at that time, not engaged in philosophy; and on these established platitudes, accepted by the many though not by the philosophers, we can base all our reasonable and practical contentions. These facts, adequate for our nonphilosophical constructions, prove, however, shifting sands when we try to build a theory of knowledge or theory of truth upon them: because then we try to have our facts as basic facts, neutral elemental atoms, facts not with regard to this or that specifiable issue, but with regard to any conceivable issue; facts not in the context of a dispute between two actual or likely disputants, but in the context of any argument between any possible disputants whatsoever, or rather, facts in no context at all. We think too much of facts as hard, brute facts, existing independently of us and ineluctable, as things that are what they are, and whose consequences will be what they will be, And about which we must not seek to be deceived. Having hypostatized them, we bow down to them, and prostrate ourselves before them. It is unnecessary. It is impossible. Facts are not sacred: they are not worth worshipping: they do not exist: they are not even things.
Corpus Christi College, Cambridge.Click here to return to home page
Click here to return to bibliography