The Idea of History

Epilegomena: 3: Historical Evidence

R. G. Collingwood

 Next | Previous


' History,' said Bury, 'is a science ; no less, and no more.'

Perhaps it is no less : that depends on what you mean by a science. There is a slang usage, like that for which ' hall' means a music-hall or 'pictures' moving pictures, according to which 'science' means natural science. Whether history is a science in that sense of the word,. however, need not be asked ; for in the tradition of European speech, going back to the time when

Latin speakers translated the Greek episthmh by their own word scientia, and continuing unbroken down to the present day, the word 'science' means any organized body of knowledge. If that is what the word means Bury is so far incontestably right, that history is a science, nothing less.

But if it is no less, it is certainly more. For anything that is a science at all must be more than merely a science, it must be a science of some special kind. A body of knowledge is never merely organized, it is always organized in some particular way. Some bodies of knowledge, like meteorology, are organized by collecting observations concerned with events of a certain kind which the scientist can watch as they happen, though he cannot produce them at will. Others, like chemistry, are organized not only by observing events as they happen, but by making them happen under strictly controlled conditions. Others again are organized not by observing events at all, but by making certain assumptions and proceeding with the utmost exactitude to argue out their consequences.

History is organized in none of these ways. Wars and revolutions, and the other events with which it deals, are not deliberately produced by historians under laboratory conditions in order to be studied with scientific precision. Nor are they even observed by historians, in the sense in which events are observed by natural scientists. Meteorologists and astronomers will make

( 250) arduous and expensive journeys in order to observe for themselves events of the kinds in which they are interested, because their standard of observation is such that they cannot be satisfied with descriptions by inexpert witnesses ; but historians do not fit out expeditions to countries where wars and revolutions are going on. And this is not because historians are less energetic or courageous than natural scientists, or less able to obtain the money such expeditions would cost. It is because the facts which might be learned through such expeditions, like the facts which might be learned through the deliberate fomenting of a war or a revolution at home, would not teach historians anything they want to know.

The sciences of observation and experiment are alike in this, that their aim is to detect the constant or recurring features in all events of a certain kind. A meteorologist studies one cyclone in order to compare it with others ; and by studying a number of them he hopes to find out what features in them are constant, that is, to find out what cyclones as such are like. But the historian has no such aim. If you find him on a certain occasion studying the Hundred Years War or the Revolution of 1688, you cannot infer that he is in the preliminary stages of an inquiry whose ultimate aim is to reach conclusions about wars or revolutions as such. If he is in the preliminary stages of any inquiry, it is more likely to be a general study of the Middle Ages or the seventeenth century. This is because the sciences of observation and experiment are organized in one way and history is organized in another. In the organization of meteorology, the ulterior value of what has been observed about one cyclone is conditioned by its relation to what has been observed about other cyclones. In the organization of history, the ulterior value of what is known about the Hundred Years War is conditioned, not by its relation to what is known about other wars, but by its relation to what is known about other things that people did in the Middle Ages.

Equally obvious is the difference between the organization of history and that of the 'exact' sciences. It is true that in history, as in exact science, the normal process of thought is inferential ; that is to say, it begins by asserting this or that, and goes on to ask what it proves. But the starting-points are of very different kinds. In exact science they are assumptions, and the

( 251) traditional way of expressing them is in sentences beginning with a word of command prescribing that a certain assumption be made : ' Let ABC be a triangle, and let AB = AC.' In history they are not assumptions, they are facts, and facts coming under the historian's observation, such as, that on the page open before him there is printed what purports to be a charter by which a certain king grants certain lands to a certain monastery. The conclusions, too, are of different kinds. In exact science, they are conclusions about things which have no special habitation in space or time : if they are .anywhere, they are everywhere, and if they are at any time they are at all times. In history, they are conclusions about events, each having a place and date of its own. The exactitude with which place and date are known to the historian is variable ; but he always knows that there were both a place and a date, and within limits he always knows what they were ; this knowledge being part of the conclusion to which he is led by arguing from the facts before him.

These differences in starting-point and conclusion imply a difference in the entire organization of the respective sciences. When a mathematician has made up his mind what the problem is which he desires to solve, the next step before him is to make assumptions which will enable him to solve it ; and this involves an appeal to his powers of invention. When an historian has similarly made up his mind, his next business is to place himself in a position where he can say: 'The facts which I am now observing are the facts from which I can infer the solution of my problem.' His business is not to invent anything, it is to discover something. And the finished products, too, are differently organized. The scheme upon which exact sciences have been traditionally arranged depends on relations of logical priority and posteriority: one proposition is placed before a second, if understanding of the first is needed in order that the second should be understood ; the traditional scheme of arrangement in history is a chronological scheme, in which one event is placed before a second if it happened at an earlier time.

History, then, is a science, but a science of a special kind. It is a science whose business is to study events not accessible to our observation, and to study these events inferentially, arguing to them from something else which is accessible to our

( 252) observation, and which the historian calls ' evidence' for the events in which he is interested.

(i) History as inferential

History has this in common with every other science: that the historian is not allowed to claim any single piece of knowledge, except where he can justify his claim by exhibiting to himself in the first place, and secondly to any one else who is both able and willing to follow his demonstration, the grounds upon which it is based. This is what was meant, above, by describing history as inferential. The knowledge in virtue of which a man is an historian is a knowledge of what the evidence at his disposal proves about certain events. If he or somebody else could have the very same knowledge of the very same events by way of memory, or second sight, or some Wellsian machine for looking backwards through time, this would not be historical knowledge ;and the proof would be that he could not produce, either to himself or to any other critic of his claims, the evidence from which he had derived it. Critic, not sceptic ; for a critic is a person able and willing to go over somebody else's thoughts for himself to see if they have been well done ; whereas a sceptic is a person who will not do this ; and because you cannot make a man think, any more than you can make a horse drink, there is no way of proving to a sceptic that a certain piece of thinking is sound, and no reason for taking his denials to heart. It is only by his peers that any claimant to knowledge is judged.

This necessity of justifying any claim to knowledge by exhibiting the grounds upon which it is based is a universal characteristic of science because it arises from the fact that a science is an organized body of knowledge. To say that knowledge is inferential is only another way of saying that it is organized. What memory is, and whether it is a kind of knowledge or not, are questions that need not be considered in a book about history: for this at least is clear, in spite of what Bacon and others have said, that memory is not history, because history is a certain kind of organized or inferential knowledge, and memory is not organized, not inferential, at all. If I say ' I remember writing a letter to So-and-so last week', that is a statement of memory, but it is not an historical statement. But if I can add ' and my memory is not deceiving me ; because here

( 253) is his reply', then I am basing a statement about the past on evidence ; I am talking history. For the same reason, there is no need in an essay like this to consider the claims of people who say that when they are in a place where a certain event has recurred they can in some way see the event going on before their eyes. What actually happens on occasions like this, and whether the people to whom it happens thereby obtain knowledge of the past, are certainly interesting questions, but this is not the right place to discuss them ; for even if these people do obtain knowledge of the past, it is not organized or inferential knowledge ; not scientific knowledge ; not history.

(ii) Different kinds of inference

Different kinds of science are organized in different ways ; and it should follow (indeed, this would seem to be only the same thing in other words) that different kinds of science are characterized by different kinds of inference. The way in which knowledge is related to the grounds upon which it is based is in fact not one and the same for all kinds of knowledge. That this is so, and that therefore a person who has studied the nature of inference as such—let us call him a logician—can correctly judge the validity of an inference purely by attending to its form, although he has no special knowledge of its subject-matter, is a doctrine of Aristotle ; but it is a delusion, although it is still believed by many very able persons who have been trained too exclusively in the Aristotelian logic and the logics that depend upon it for their chief doctrines.[1]

The main scientific achievement of the ancient Greeks lay in mathematics ; their main work on the logic of inference was naturally, therefore, devoted to that form of inference which occurs in exact science. When at the end of the Middle Ages the modern natural sciences of observation and experiment began

( 253) to take shape, a revolt against Aristotelian logic was inevitable ; in particular, a revolt against the Aristotelian theory of demonstration, which could by no manner of means be made to cover the technique actually used in the new sciences. Thus, by degrees, there came into existence a new logic of inference, based on analysis of the procedure used in the new natural sciences. The text-books of logic in use to-day still bear the marks of this revolt in the distinction they draw between two kinds of inference, ' deductive' and ' inductive'. It was not until late in the nineteenth century that historical thought reached a stage of development comparable with that reached by natural science about the beginning of the seventeenth ; but this event has not yet begun to interest those philosophers who write text-books of logic.

The chief characteristic of inference in the exact sciences, the characteristic of which Greek logicians tried to give a theoretical account when they formulated the rules of the syllogism, is a kind of logical compulsion whereby a person who makes certain assumptions is forced, simply by so doing, to make others. He has freedom of choice in two ways: he is not compelled to make the initial assumption (a fact technically expressed by saying that 'the starting-points of demonstrative reasoning are not themselves demonstrable'), and when once he has done so he is still at liberty, whenever he likes, to stop thinking. What he cannot do is to make the initial assumption, to go on thinking, and to arrive at a conclusion different from that which is scientifically correct.

In what is called 'inductive' thinking there is no such compulsion. The essence of the process, here, is that having put certain observations together, and having found that they make a pattern, we extrapolate this pattern indefinitely, just as a man who has plotted a few points on squared paper and says to himself 'the points I have plotted suggest a parabola', proceeds to draw as much of the parabola as he likes in either direction. This is technically described as 'proceeding from the known to the unknown', or 'from the particular to the universal'. It is essential to 'inductive' thinking, though the logicians who have tried to construct a theory of such thinking have not always realized this, that the step so described is never taken under any kind of logical compulsion. The thinker who takes it is logically

( 255) free to take it or not to take it, just as he pleases. There is nothing in the pattern formed by the observations he or some-one else has actually made which can oblige him to extrapolate in that particular way, or indeed to extrapolate at all. The reason why this very obvious truth has been so often overlooked is that people have been hypnotized by the prestige of Aristotelian logic into thinking that they see a closer resemblance than actually exists between 'deductive' and 'inductive' thinking, that is, between exact science and the sciences of observation and experiment. In both cases there are, for any given piece of thinking, certain starting-points, traditionally called premisses, and a certain terminal point, traditionally called a conclusion ; and in both cases the premisses 'prove' the conclusion. But whereas in exact science this means that they enforce the conclusion, or make it logically obligatory, in the sciences of observation and experiment it means only that they justify it, that is, authorize anybody to think it who wishes to do so. What they provide, when they are said to 'prove' a certain conclusion, is not compulsion to embrace it, but only permission ;a perfectly legitimate sense of the word 'prove' (approuver, probare), as there should be no need to show.

If in practice this permission, like so many permissions, amounts to virtual compulsion, that is only because the thinker who avails himself of it does not regard himself as free to extrapolate or not, just as he pleases. He regards himself as under an obligation to do so, and to do it in certain ways: obligations which, when we inquire into their history, we find to have their roots in certain religious beliefs about nature and its creator God. It would be out of place to develop this statement more fully here ; but not, perhaps, to add that if to-day it seems to some readers paradoxical, that is only because the facts have been obscured by a smoke-screen of propagandist literature, beginning with the 'illuminist' movement of the eighteenth century and prolonged by the 'conflict between religion and science' in the nineteenth, whose purpose was to attack Christian theology in the supposed interests of a 'scientific view of the world' which in fact is based upon it and could not for a moment survive its destruction. Take away Christian theology, and the scientist has no longer any motive for doing what inductive thought gives him permission to do. If he goes on doing it at all,

( 256) that is only because he is blindly following the conventions of the professional society to which he belongs.

(iii) Testimony

Before trying to describe the special characteristics of historical inference positively, we shall find it useful to describe them negatively: to describe something that is very often, but mistakenly, identified with it. Like every science, history is autonomous. The historian has the right, and is under an obligation, to make up his own mind by the methods proper to his own science as to the correct solution of every problem that arises forhim in the pursuit of that science. He can never be under any obligation, or have any right, to let someone else make up his mind for him. If anyone else, no matter who, even a very learned historian, or an eyewitness, or a person in the confidence of the man who did the thing he is inquiring into, or even the man who did it himself, hands him on a plate a ready-made answer to his question, all he can do is to reject it: not because he thinks his informant is trying to deceive him, or is himself deceived, but because if he accepts it he is giving up his autonomy as an historian and allowing someone else to do for him what, if he is a scientific thinker, he can only do for himself. There is no need for me to offer the reader any proof of this statement. If he knows anything of historical work, he already knows of his own experience that it is true. If he does not already know that it is true, he does not know enough about history to read this essay with any profit, and the best thing he can do is to stop here and now.

When the historian accepts a ready-made answer to some question he has asked, given him by another person, this other person is called his ' authority', and the statement made by such an authority and accepted by the historian is called ' testimony'.In so far as an historian accepts the testimony of an authority and treats it as historical truth, he obviously forfeits the name of historian ; but we have no other name by which to call him.

Now, I am not for a moment suggesting that testimony ought never to be accepted. In the practical life of every day, we constantly and rightly accept the information that other people offer us, believing them to be both well informed and truthful, and having, sometimes, grounds for this belief. I do not even

( 257) deny, though I do not assert it, that there may be cases in which, as perhaps in some cases of memory, our acceptance of such testimony may go beyond mere belief and deserve the name of knowledge. What I assert is that it can never be historical knowledge, because it can never be scientific knowledge. It is not scientific knowledge because it cannot be vindicated by appeal to the grounds on which it is based. As soon as there are such grounds, the case is no longer one of testimony. When testimony is reinforced by evidence, our acceptance of it is no longer the acceptance of testimony as such ; it is the affirmation of something based upon evidence, that is, historical knowledge.

(iv) Scissors, and paste

There is a kind of history which depends altogether upon the testimony of authorities. As I have already said, it is not really history at all, but we have no other name for it. The method by which it proceeds is first to decide what we want to know about,and then to go in search of statements about it, oral or written, purporting to be made by actors in the events concerned, or by eyewitnesses of them, or by persons repeating what actors or eyewitnesses have told them, or have told their informants, or those who informed their informants, and so on. Having found in such a statement something relevant to his purpose, the historian excerpts it and incorporates it, translated if necessary and recast into what he considers a suitable style, in his own history. As a rule, where he has many statements to draw upon, he will find that one of them tells him what another does not ; so both or all of them will be incorporated. Sometimes he will find that one of them contradicts another ; then, unless he can find a way of reconciling them, he must decide to leave one out ;and this, if he is conscientious, will involve him in a critical consideration of the contradictory authorities' relative degree of trustworthiness. And sometimes one of them, or possibly even all of them, will tell him a story which he simply cannot believe, a story characteristic, perhaps, of the superstitions or prejudices of the author's time or the circle in which he lived, but not credible to a more enlightened age, and therefore to be omitted.

History constructed by excerpting and combining the testimonies of different authorities I call scissors-and-paste history. I repeat that it is not really history at all, because it does not

( 258) satisfy the necessary conditions of science ; but until lately it was the only kind of history in existence, and a great deal of the history people are still reading to-day, and even a good deal of what people are still writing, belongs to this type. Consequently people who know little about history (some of whom, in spite of my recent farewell, may still be reading these pages) will say with some impatience : `Why, this thing that you say is not history, is just history itself ; scissors and paste, that is what history is ; and that is why history is not a science, which is a fact that everybody knows, in spite of groundless claims by professional historians magnifying their office'. I shall therefore say a little more about the vicissitudes of scissors-and-paste history.

Scissors and paste was the only historical method known to the later Greco-Roman world or the Middle Ages. It existed in its simplest form: An historian collected testimony, spoken or written, using his own judgement as to its trustworthiness, and put it together for publication: the work which he did on it being partly literary—the presentation of his material as a connected, homogeneous, and convincing narrative—and partly rhetorical, if I may use that word to indicate the fact that most ancient and medieval historians aimed at proving a thesis, in particular some philosophical or political or theological thesis.

It was only in the seventeenth century, when the post-medieval reform of natural science had attained completion, that historians began to think their house also needed to be set in order. Two new movements in historical method now began. One was a systematic examination of authorities, in order to determine their relative credibility, and in particular to establish principles according to which this determination should be carried out. The other was a movement to broaden the basis of history by making use of non-literary sources, such as coins and inscriptions and suchlike relics of antiquity which hitherto had been of interest not to historians but only to collectors of curiosities.

The first of these movements did not overstep the limits of scissors-and-paste history, but it permanently altered its character. As soon as it became understood that a given statement, made by a given author, must never be accepted for historical truth until the credibility of the author in general and

( 259) of this statement in particular had been systematically inquired into, the word `authority' disappeared from the vocabulary of historical method, except as an archaistic survival ; for the man who makes the statement came henceforth to be regarded not as someone whose word must be taken for the truth of what he says, which is what was meant by calling him an authority, but as someone who has voluntarily placed himself in the witness-box for cross-examination. The document hitherto called an authority now acquired a new status, properly described by calling it a `source', a word indicating simply that it contains the statement, without any implications as to its value. That is sub judice; and it is the historian who judges.

This is `critical history', as it was worked out from the seventeenth century onwards, and officially acclaimed in the nineteenth as the apotheosis of the historical consciousness. There are two things to observe about it : that it was still only a form of scissors and paste ; and that it had already, in principle, been superseded by something very different.

(1)     The problem of which historical criticism offers a solution is a problem interesting to nobody but the practitioner of scissors-and-paste history. The presupposition of the problem is that in a certain source we have found a certain statement which bears on our subject. The problem is: Shall we incorporate this statement in our own narrative or not ? The methods of historical criticism are intended to solve this problem in one or other of two ways : affirmatively or negatively. In the first case, the excerpt is passed as fit for the scrapbook ; in the second, it is consigned to the waste-paper basket. 

(2)     But many historians in the nineteenth century, and even in the eighteenth, were aware that this dilemma was fallacious. It was by now a commonplace that if in some source you found a statement which for some reason could not be accepted as literally true, you must not on that account reject it as worthless. It might be a way, perhaps a well-established way according to the custom of the time when it was written, of saying something which you, through ignorance of that custom, did not recognize as its meaning. 

The first person to make this point was Vico, at the beginning of the eighteenth century. It is true that in Germany, the home of `critical history' in the late eighteenth and early nineteenth

( 260) centuries, the importance of Vico's work was not as widely recognized as it ought to have been ; buts he was not entirely unknown there ; indeed, some very famous German scholars, like F. A. Wolf, actually borrowed some of his ideas. Now, anyone who had read Vico, or even a secondhand version of some of his ideas, must have known that the important question about any statement contained in a source is not whether it is true or false, but what it means. And to ask what it means is to step right outside the world of scissors-and-paste history into a world where history is not written by copying out the testimony of the best sources, but by coming to your own conclusions.

Critical history is of interest to the student 'of historical method to-day only as the final form taken by scissors-and-paste history on the eve of its dissolution. I will not venture to name any historian, or even any historical work, as one from which the last traces of it have disappeared. But I will venture to say that any historian (if there is any) who practises it consistently, or any historical work written entirely on this method, is at least a century out of date.

So much for one of the two movements which gave new life to history in the seventeenth century. The other, the archaeological movement, was totally hostile to the principles of scissors-and-paste history, and could have arisen only when those principles were moribund. No very profound knowledge of coins and inscriptions is needed in order to realize that the assertions they make are by no means uniformly trustworthy, and indeed are to be judged more as propaganda than as statements of fact. Yet this gives them an historical value of their own; for propaganda, too, has its history.

If any reader still thinks that history as practised to-day is a scissors-and-paste affair, and is willing to go to a little trouble in order to settle the question, let him take the history of Greece down to the end of the Peloponnesian War, which I mention as an example peculiarly favourable to himself because Herodotus and Thucydides have there maintained the position of ' authorities' to a quite peculiar degree, and compare in detail the account of it given by Grote with that given in the Cambridge Ancient History. Let him mark in each book every sentence of which he can find the original in Herodotus or Thucydides ; and by the time he is through with the job he will have learnt something

( 261) about how historical method has changed in the last hundred years.

(v) Historical inference

In (ii) it was pointed out that proof might be either compulsive, as in exact science, where the nature of inference is such that nobody can affirm the premisses without being obliged to affirm the conclusion also, or permissive, as in 'inductive' science, where all a proof can do is to justify the thinker in affirming its conclusion, granted that he wishes to do so. An inductive argument with a negative conclusion is compulsive, that is to say it absolutely forbids the thinker from affirming what he wishes to affirm ; with a positive conclusion, it is never more than permissive.

If history means scissors-and-paste history, the only kind of proof known to the historian is of this latter kind. For the scissors-and-paste historian, there is only one kind of problem which is capable of being settled by any sort of argument. This is the problem whether to accept or reject a certain piece of testimony bearing upon the question in which he is interested. The sort of argument by which he settles a problem of this kind is, of course, historical criticism. If criticism leads him to a negative conclusion, viz. that the statement or its author is untrustworthy, this forbids him to accept it, just as a negative result in an ' inductive' argument (for example, a result showing that events of the kind in which he is interested happen in the absence of that kind of event which he hopes to identify as their cause) forbids the inductive scientist to affirm the view he hoped to affirm. If criticism leads him to a positive conclusion, the most it gives him is a nihil obstat. For the positive conclusion is in effect that the man who made the statement is not known to be either ignorant or mendacious, and that the statement itself bears upon it no recognizable marks of being untrue. But it maybe untrue for all that : and the man who made it, though in general he bears a good name for being well informed and honest, may on this one occasion have fallen a victim to misinformation about his facts, misunderstanding of them, or a desire to suppress or distort what he knew or believed to be the truth.

To avert a possible misunderstanding, it may be added here

( 262) that one might think there was another kind of problem for the scissors-and-paste historian, beside the kind which consists in whether to accept or reject a given piece of testimony, which therefore has to be settled by methods other than those of historical criticism : the problem, namely, of what implications follow from a piece of testimony that he has accepted, or would follow if he did accept it. But this is not a problem specially belonging to scissors-and-paste history ; it is a problem which arises in history or pseudo-history of any kind whatever, and indeed in any kind of science or pseudo-science. It is simply the general problem of implication. When it occurs in scissors-and-paste history, however, it presents one peculiar feature. If a certain statement coming to the historian by way of testimony has a certain implication, and if this implicational relation is a compulsive one, nevertheless if the inference which leads him to accept the testimony is only permissive the same permissive character attaches to his assertion of its implication. If he has only borrowed his neighbour's cow, and she has a calf in his field, he cannot claim the calf as his own property. Any answer to the question whether the scissors-and-paste historian is obliged or only permitted to accept certain testimony carries with it a corresponding answer to the question whether he is obliged or only permitted to accept the implications of that testimony.

One hears it said that history is ' not an exact science'. The meaning of this I take to be that no historical argument ever proves its conclusion with that compulsive force which is characteristic of exact science. Historical inference, the saying seems to mean, is never compulsive, it is at best permissive ; or, as people sometimes rather ambiguously say, it never leads to certainty, only to probability. Many historians of the present writer's generation, brought up at a time when this proverb was accepted by the general opinion of intelligent persons (I say nothing of the few who were a generation ahead of their time), must be able to recollect their excitement on first discovering that it was wholly untrue, and that they were actually holding in their hands an historical argument which left nothing to caprice, and admitted of no alternative conclusion, but proved its point as conclusively as a demonstration in mathematics. Many of these, again, must be able to recollect the shock of dis-

( 263) -covering on reflection that the proverb was not, strictly speaking, an error about history, history as they were practising it, the science of history, but a truth about something else, namely scissors-and-paste history.

If any reader wishes to rise here on a point of order and protest that a philosophical question, which ought therefore to be settled by reasoning, is being illegitimately disposed of by reference to the authority of historians, and quote against me the good old story about the man who said ' I'm not arguing, I'm telling you', I can only admit that the cap fits. I am not arguing ; I am telling him.

Is this wrong of me ? The question I want settled is whether an inference of the kind used in scientific history, as distinct from scissors-and-paste history, yields compulsion or only permission to embrace its conclusion. Suppose the question had been not about history but about mathematics. Suppose somebody had wanted to know whether Euclid's proof of what is called Pythagoras' theorem compels or merely permits a man to adopt the view that the square on the hypotenuse is equal to the sum of the squares on the other two sides. I speak with submission ; but for myself I can think of only one thing that a sensible man in that situation would do. He would try to find somebody whose mathematical education had got as far as Euclid I. 47, and ask him. And if he did not like his answer, he would look for other people similarly qualified to give one, and ask them. If all else failed to convince him, he would have to get down to it and study the elements of plane geometry for himself.

The one thing that he will not do, if he is a man of any intelligence, is to say ' This is a philosophical question, and the only answer I will be satisfied with is a philosophical answer'. He can call it anything he pleases ; he cannot alter the fact that the only way of knowing whether a given type of argument is cogent or not is to learn how to argue that way, and find out. Meanwhile, the second best thing is to take the word of people who have done so for themselves.

(vi) Pigeon-holing

Scissors-and-paste historians who have become disgusted with the work of copying out other people's statements, and,

( 264) conscious of having brains, feel a laudable desire to use them, are often found satisfying this desire by inventing a system of pigeon-holes in which to arrange their learning. This is the origin of all those schemes and patterns into which history has again and again, with surprising docility, allowed itself to be forced by such men as Vico, with his pattern of historical cycles based on Greco-Roman speculations ; Kant, with his proposal for a `universal history from a cosmopolitan point of view'; Hegel, who followed Kant in conceiving universal history as the progressive realization of human freedom ; Comte and Marx, two very great men who followed Hegel's lead each in his own way ; and so on down to Flinders Petrie, Oswald Spengler, and Arnold Toynbee in our own time, whose affinities are less with Hegel than with Vico.

Although we find it as late as the twentieth century and as early as the eighteenth, not to mention isolated occurrences even earlier, this impulse towards arranging the whole of history in a single scheme (not a chronological scheme merely, but a qualitative scheme, in which 'periods' each with its own pervasive character follow one another in time, according to a pattern which may be necessary a priori on logical grounds, or may be forced upon our minds by the fact of its frequent repetition, or may be a bit of both) is in the main a nineteenth-century phenomenon. It belongs to the period when scissors-and-paste history was on its last legs ; when people were becoming dissatisfied with it but had not yet broken away from it. This is why the people who have indulged it have been, in general, men with a high degree of intelligence and a real talent for history, but a talent which has been to some extent thwarted and baffled by the limitations of scissors and paste.

It is typical of this condition that some of them described their pigeon-holing enterprise as 'raising history to the rank of a science'. History as they found it meant scissors-and-paste history ; that, obviously, was no science, because there was nothing autonomous, nothing creative, about it ; it was merely the transhipment of ready-made information from one mind into another. They were conscious that history might be something more than this. It might have, and it ought to have, the characteristics of a science. But how was this to be brought about ? At this point the analogy of the natural sciences came,

( 265) they thought, to their aid. It had been a commonplace ever since Bacon that a natural science began by collecting facts, and then went on to construct theories, that is, to extrapolate the patterns discernible in the facts already collected. Very well: let us put together all the facts that are known to historians, look for patterns in them, and then extrapolate these patterns into a theory of universal history.

It proved to be not at all a difficult task for anybody with an active mind and a taste for hard work. For there was no need to collect all the facts known to historians. Any large collection of facts, it was found, revealed patterns in plenty ; and extrapolating such patterns into the remote past, about which there was very little information, and into the future, about which there was none, gave the 'scientific' historian just that sense of power which scissors-and-paste history denied him. After being taught to believe that he, as an historian, could never know anything except what his authorities told him; he found himself discovering, as he fancied, that this lesson had been a fraud ; that by converting history into a science he could ascertain, entirely for himself, things that his authorities had concealed from him or did not know.

This was a delusion. The value of each and all of these pigeon-holing schemes, if that means their value as means for discovering historical truths not ascertainable by the interpretation of evidence, was exactly nil. And in fact none of them ever had any scientific value at all ; for it is not enough that science should be autonomous or creative, it must also be cogent or objective ; it must impress itself as inevitable on anyone who is able and willing to consider the grounds upon which it is based, and to think for himself what the conclusions are to which they point. That is what none of these schemes can do. They are the offspring of caprice. If any of them has ever been accepted by any considerable body of persons beside the one who invented it, that is not because it has struck them as scientifically cogent, but because it has become the orthodoxy of what is in fact, though not necessarily in name, a religious community. This was to some extent achieved by Comtism, and to a much greater extent by Marxism. In these cases, or at any rate in the case of Marxism, historical schemes of the kind in question proved to have an important magical value, as

( 264) providing a focus for emotions and in consequence an incentive to action. In other cases they have had an amusement value, not without its function in the life of a jaded scissors-and-paste man.

And the delusion was not complete. The hope that scissors-and-paste history would one day be replaced by a new kind of history that should be genuinely scientific was a well-grounded hope, which has in fact been realized. The hope that this new kind of history would enable the historian to know things that his authorities could not or would not tell him was also well grounded, and has also been fulfilled. How these things have happened, we shall very soon see.

(vii) Who killed John Doe?

When John Doe was found, early one Sunday morning, lying across his desk with a dagger through his back, no one expected that the question who did it would be settled by means of testimony. It was not likely that anyone saw the murder being done. It was even less likely that someone in the murderer's confidence would give him away. It was least likely of all that the murderer would walk into the village police-station and denounce himself. In spite of this, the public demanded that he should be brought to justice, and the police had hopes of doing it ; though the only clue was a little fresh green paint on the handle of the dagger, like the fresh green paint on the iron gate between John Doe's garden and the rector's.

This was not because they hoped that, in time, testimony would be forthcoming. On the contrary, when it did come, in the shape of a visit from an elderly neighbouring spinster asserting that she killed John Doe with her own hand because he had made a dastardly attempt upon her virtue, even the village constable (not an exceptionally bright lad, but kindly) advised her to go home and have some aspirin. Later in the day the village poacher came along and said that he had seen the squire's gamekeeper climbing in at John Doe's study window ; testimony which was treated with even less deference. Finally the rector's daughter, in a state of great agitation, rushed in and said she had done it herself ; the only effect of which was to make the village constable ring up the local Inspector and remind him that the girl's young man, Richard Roe, was a medical student,

( 267) and presumably knew where to find a man's heart ; and that he had spent Saturday night at the rectory, within a stone's throw of the dead man's house.

There had been a thunderstorm that night, with heavy rain, between twelve and one ; and the Inspector, when he questioned the rectory parlour-maid (for the living was a good one), was told that Mr. Roe's shoes had been very wet in the morning. Questioned, Richard admitted having gone out in the middle of the night, but refused to say where or why.

John Doe was a blackmailer. For years he had been black-mailing the rector, threatening to publish the facts about a certain youthful escapade of his dead wife. Of this escapade the rector's supposed daughter, born six months after marriage, was the fruit ; and John Doe had letters in his possession that proved it. By now he had absorbed the whole of the rector's private fortune, and on the morning of the fatal Saturday he demanded an instalment of his wife's, which she had left to him in trust for her child. The rector made up his mind to end it. He knew that John Doe sat at his desk late into the night ; he knew that behind him, as he sat, there was a french window on the left and a trophy of Eastern weapons on the right ; and that on hot nights the window was left open until he went to bed. At midnight, wearing gloves, he slipped out ; but Richard, who had noticed his state of mind and was troubled about it, happened to be leaning out of his window and saw the rector cross the garden. He hurried into his clothes and followed ; but by the time he reached the garden the rector was gone. At this moment the thunderstorm broke. Meanwhile the rector's plan had succeeded perfectly. John Doe was asleep, his head fallen forward on a pile of old letters. Only after the dagger had reached his heart did the rector look at them, and see his wife's handwriting. The envelopes were addressed ' John Doe, Esq.' Until that moment, he had never known who his wife's seducer had been.

It was Detective Inspector Jenkins of Scotland Yard, called in by the Chief Constable at the entreaty of his old friend's little girl, who found in the rectory dustbin a lot of ashes, mostly from writing paper, but including some from leather, probably a pair of gloves. The wet paint on John Doe's garden gate—he had painted it himself that day, after tea—explained why the gloves

( 268) might have been destroyed ; and among the ashes were metal buttons bearing the name of a famous glove-maker in Oxford Street whom the rector always patronized. More of John Doe's paint was found on the right cuff of a jacket, ruined as to shape by a recent wetting, which on Monday the rector bestowed on a deserving parishioner. The Detective-Inspector was severely blamed, later on, for allowing the rector to see in what direction his inquiries were tending, and thus giving him an opportunity to take cyanide and cheat the hangman.

The methods of criminal detection are not at every point identical with those of scientific history, because their ultimate purpose is not the same. A criminal court has in its hands the life and liberty of a citizen, and in a country where the citizen is regarded as having rights the court is therefore bound to do something and do it quickly. The time taken to arrive at a decision is a factor in the value (that is, the justice) of the decision itself. If any juror says: `I feel certain that a year hence, when we have all reflected on the evidence at leisure, we shall be in a better position to see what it means,' the reply will be: `There is something in what you say ; but what you propose is impossible. Your business is not just to give a verdict ; it is to give a verdict now ; and here you stay until you do it.' This is why a jury has to content itself with something less than scientific (historical) proof, namely with that degree of assurance or belief which would satisfy it in any of the practical affairs of daily life.

The student of historical method will hardly find it worth his while, therefore, to go closely into the rules of evidence, as these are recognized in courts of law. For the historian is under no obligation to make up his mind within any stated time. Nothing matters to him except that his decision, when he reaches it, shall be right : which means, for him, that it shall follow inevitably from the evidence.

So long as this is borne in mind, however, the analogy between legal methods and historical methods is of some value for the understanding of history ; of sufficient value, I think, to justify my having put before the reader in outline the above sample of a literary genre which in the absence of any such motive it would, of course, be beneath his dignity to notice.

( 269)

(viii) The question

Francis Bacon, lawyer and philosopher, laid it down in one of his memorable phrases that the natural scientist must `put Nature to the question'. What he was denying, when he wrote this, was that the scientist's attitude towards nature should be one of respectful attentiveness, waiting upon her utterances and building his theories on the basis of what she chose to vouchsafe him. What he was asserting was two things at once : first, that the scientist must take the initiative, deciding for himself what he wants to know and formulating this in his own mind in the shape of a question ; and secondly, that he must find means of compelling nature to answer, devising tortures under which she can no longer hold her tongue. Here, in a single brief epigram, Bacon laid down once for all the theory of experimental science.

It is also, though Bacon did not know this, the true theory of historical method. In scissors-and-paste history the historian takes up a pre-Baconian position. His attitude towards his authorities, as the very word shows, is one of respectful attentiveness. He waits to hear what they choose to tell him, and lets them tell it in their own way and at their own time. Even when he has invented historical criticism, and his authorities have become mere sources, this attitude is at bottom unchanged. There is a change, but it is only superficial. It consists merely in the adoption of a technique for dividing witnesses into sheep and goats. One class is disqualified from giving testimony ; the other is treated exactly as authorities were treated under the old dispensation. But in scientific history, or history proper, the Baconian revolution has been accomplished. The scientific historian no doubt spends a great deal of time reading the same books that the scissors-and-paste historian used to read—Herodotus, Thucydides, Livy, Tacitus, and so forth—but he reads them in an entirely different spirit ; in fact, a Baconian spirit. The scissors-and-paste historian reads them in a simply receptive spirit, to find out what they said. The scientific historian reads them with a question in his mind, having taken the initiative by deciding for himself what he wants to find out from them. Further, the scissors-and-paste historian reads them on the understanding that what they did not tell him in so many words he would never find out from them at all ; the scientific

( 270) historian puts them to the torture, twisting a passage ostensibly about something quite different into an answer to the question he has decided to ask. Where the scissors-and-paste historian said quite confidently ' There is nothing in such-and-such an author about such-and-such a subject ', the scientific or Baconian historian will reply ' Oh, isn't there ? Do you not see that in this passage about a totally different matter it is implied that the author took such-and-such a view of the subject about which you say his text contains nothing ? '

To illustrate from my fable. The village constable does not arrest the rector's daughter and beat her periodically with a rubber truncheon until she tells him that she thinks Richard did the murder. What he tortures is not her body, but her statement that she killed John Doe. He begins by using the methods of critical history. He says to himself : ' The murder was done by somebody with a good deal of strength and some knowledge of anatomy. This girl certainly hasn't the first, and probably hasn't the second ; at any rate, I know she has never attended ambulance classes. Further, if she had done it she wouldn't be in such a hurry to accuse herself. The story is a lie.'

At this point the critical historian would lose interest in the story and throw it in the waste-paper basket : the scientific historian begins to be interested in it, and tests it for chemical reactions. This he is able to do because, being a scientific thinker, he knows what questions to ask. ' Why is she telling a lie ? Because she is shielding someone. Whom is she shielding ?Either her father or her young man. Is it her father ? No ; fancy the rector ! Therefore it is her young man. Are her suspicions of him well founded ? They might be ; he was here at the time ; he is strong enough ; and he knows enough anatomy.' The reader will recollect that in criminal detection probability is required, of a degree sufficient for the conduct of daily life, whereas in history we demand certainty. Apart from that, the parallel is complete. The village constable (not a clever lad, as I explained ; but a scientific thinker does not have to be clever, he has to know his job, that is, know what questions to ask) has been trained in the elements of police work, and this training enables him to know what questions to ask and thus to interpret the untrue statement that she did it herself into evidence for the true conclusion that she suspects Richard Roe.

( 271)

The constable's only mistake was that in the excitement of answering the question `Whom does this girl suspect ? ' he lost sight of the question 'Who killed John Doe ? ' This is where Inspector Jenkins, not so much because he was a cleverer man as because he had learned the job more thoroughly, had the advantage of him. The way I see the Inspector going to work is like this.

'Why does the rector's daughter suspect Richard Roe ? Probably because she knows that he was involved in something queer which happened at the rectory that night. We know that one queer thing happened at the rectory: Richard was out in the storm, and that was quite enough to make the girl suspicious. But what we want to know is, did he kill John Doe ? If he did, when did he do it ? After the thunderstorm broke, or before ? Not before, because here are his tracks going both ways in the mud of the rectory garden path : you see them beginning a few yards from the garden door, going away from the house ; so that is where he was, and that is the direction he was going in, when the downpour began. Well, did he carry mud into John Doe's study ? No : none there. Did he take off his shoes before going in ? Think a moment. What position was John Doe in when he was stabbed ? Was he leaning back or sitting upright in his chair ? No ; because the chair would have protected his back. He must have been leaning right forward. Possibly, indeed probably, asleep in the position in which he still lies. How exactly did the murderer proceed? If Doe was asleep, nothing easier: step quietly inside, take the dagger and in it goes. If Doe was awake and merely leaning forward, the same might be done, but not so easily. Now, did the murderer pause outside to take off his shoes ? Impossible. In either case, speed was the first thing necessary: the job had to be done before he leaned back, or woke up. So the absence of mud in the study lets Richard out.

'Then, once more, why did he go into the garden ? For a walk ? Not with that thunderstorm growling about. For a smoke ? They smoke all over the house. To meet the girl ? No signs that she was in the garden ; and why should he ? They had had the drawing-room to themselves ever since dinner, and the rector isn't one to shoo young people off to bed. Broadminded sort of chap. Had trouble, I shouldn't wonder. Now, why did young Richard go into that garden ? Something must have been

( 272) going on there. Something queer. A second queer thing that night at the rectory, one we don't know about.

'What could it have been? If the murderer had come from the rectory, which that paint suggests he did, and if Richard saw him from his window, it might have been that ; because the murderer got to Doe's house before the rain began, and Richard was caught in it ten yards from the garden door. Just time. Let's see what would follow, if the murderer did come from the rector. Probably he went back there afterwards. No tracks in the mud ; why? Because he knew the garden well enough to keep on the grass all the way, even in that pitch darkness. If so, he knew the rectory very well and also spent the night there. Was it the rector himself ?

'Now why does Richard refuse to say what made him go into the garden? It must be to keep somebody out of trouble ; almost certainly, trouble about the murder. Not himself, because I've told him we know he didn't do it. Somebody else. Who ? Might be the rector. Can't think of anybody else it might be. Suppose it was the rector ; how would he have worked it? Very easy. Go out about midnight, in tennis shoes and gloves. Quite silent on the rectory paths—no gravel on them. Reach that little iron gate into John Doe's garden. Does he know it's wet paint? Probably not ; it was only painted after tea. So he grabs it. Paint on glove. Probably paint on jacket too. Walk on the grass to Doe's study window. Doe is leaning forward in his chair, or likelier asleep. Now fora bit of quick work, easy for a good tennis-player. Left foot inside, right foot to the right, grab that dagger thing, left foot forward, in it goes.

'But what had John Doe been doing at that desk? Nothing on it, you know. Queer. Does a man spend the evening sitting at an empty desk ? There must have been something there. What do we know about the chap at the Yard? Blackmailer, that's it. Had he been blackmailing the rector? and gloating over the letters, or what not, all evening? And did the rector, if it was the rector, find him asleep on top of them? Well, that's not our business. We'll pass it on to the defence, —for what it's worth. I'd rather not use a motive like that in prosecution.

'Now then, Jonathan, don't go ahead too fast. You've got him in there, you've got to get him out again. What exactly does he do? About now it begins to rain cats and dogs. Back he

( 273) goes through it. More paint at the gate. Walk on grass, no mud brought in. Back in the house. All soaked: gloves covered with paint, too. Wipe paint off door-knob. Lock up. Put letters (if it was letters), and anyhow gloves, in the hot-water furnace—the ashes may be in the dustbin now. Put all clothes in the bath-room cupboard ; they will be dry by morning. And so they are ; but the jacket will be hopelessly out of shape. Now what did he do with that jacket ? First, he'd look for paint on it. If he found paint, he'd have to destroy the thing ; and I pity the man who tries to destroy a jacket in a house overrun with women. If he didn't find any, he would certainly give it away on the quiet to a poor man.

'Well, well: there's a pretty story for you ; but how can we tell whether it's true or not? There are two questions we've got to ask. First : can we find the ashes of those gloves ? And the metal buttons, if they are like most of his gloves? If we can, the story is true. And if we can find a lot of writing-paper ash as well, the blackmail bit is true, too. Second: where is that jacket? Because if we can find the tiniest speck of John Doe's paint on it, there's our case.'

I have gone to some length in this analysis because I wish to bring home to the reader the following points about the questioning activity which is the dominant factor in history, as it is in all scientific work.

(i) Every step in the argument depends on asking a question. The question is the charge of gas, exploded in the cylinder-head, which is the motive force of every piston-stroke. But the metaphor is not adequate, because each new piston-stroke is produced not by exploding another charge of the same old mixture but by exploding a charge of a new kind. No one with any grasp of method will go on asking the same question all the time, 'Who killed John Doe?' He asks a new question every time. And it is not enough to cover the ground by having a catalogue of all the questions that have to be asked, and asking every one of them sooner or later : they must be asked in the right order. Descartes, one of the three great masters of the Logic of Questioning (the other two being Socrates and Bacon), insisted upon this as a cardinal point in scientific method, but so far as modern works on logic are concerned, Descartes might never have lived. Modern logicians are in a conspiracy to

(274) pretend that a scientist's business is to 'make judgements', or 'assert propositions', or 'apprehend facts', and also to 'assert, or 'apprehend' the relations between them ; suggesting that they have no experience whatever of scientific thinking, and wish to palm off, as an account of science, an account of their own haphazard, unsystematic, unscientific consciousness.

(2) These questions are not put by one man to another man, in the hope that the second man will enlighten the first man's ignorance by answering them. They are put, like all scientific questions, to the scientist by himself. This is the Socratic idea which Plato was to express by defining thought as 'the dialogue of the soul with itself', where Plato's own literary practice makes it clear that by dialogue he meant a process of question and answer. When Socrates taught his young pupils by asking them questions, he was teaching them how to ask questions of themselves, and showing them by examples how amazingly the obscurest subjects can be illuminated by asking oneself intelligent questions about them instead of simply gaping at them, according to the prescription of our modern anti-scientific epistemologists, in the hope that when we have made our minds a perfect blank we shall 'apprehend the facts'.

(ix) Statement and evidence

It is characteristic of scissors-and-paste history, from its least critical to its most critical form, that it has to do with ready-made statements, and that the historian's problem about any one of these statements is whether he shall accept it or not : where accepting it means reasserting it as a part of his own historical knowledge. Essentially, history for the scissors-and-paste historian means repeating statements that other people have made before him. Hence he can get to work only when he is supplied with ready-made statements on the subjects about which he wants to think, write, and so forth. It is the fact that these statements have to be found by him ready-made in his sources that makes it impossible for the scissors-and-paste historian to claim the title of a scientific thinker, for this fact makes it impossible to attribute to him that autonomy which is everywhere essential to scientific thought ; where by autonomy I mean the condition of being one's own authority, making statements or taking action on one's own initiative and not

( 275) because those statements or actions are authorized or prescribed by anyone else.

It follows that scientific history contains no ready-made statements at all. The act of incorporating a ready-made statement into the body of his own historical knowledge is an act which, for a scientific historian, is impossible. Confronted with a ready-made statement about the subject he is studying, the scientific historian never asks himself: 'Is this statement true or false?', in other words 'Shall I incorporate it in my history of that subject or not?' The question he asks himself is : 'What does this statement mean ? ' And this is not equivalent to the question 'What did the person who made it mean by it ? ', although that is doubtless a question that the historian must ask, and must be able to answer. It is equivalent, rather, to the question 'What light is thrown on the subject in which I am interested by the fact that this person made this statement, meaning by it what he did mean ? ' This might be expressed by saying that the scientific historian does not treat statements as statements but as evidence: not as true or false accounts of the facts of which they profess to be accounts, but as other facts which, if he knows the right questions to ask about them, may throw light on those facts. Thus in my fable the rector's daughter tells the constable that she killed John Doe. As a scientific historian, he begins attending seriously to this statement at the point where he stops treating it as a statement, that is, as a true or false account of her having done the murder, and begins treating the fact that she makes it as a fact which may be of service to him. It is of service to him because he knows what questions to ask about it, beginning with the question : 'Now why does she tell this story?' The scissors-and-paste historian is interested in the 'content', as it is called, of statements: he is interested in what they state. The scientific historian is interested in the fact that they are made.

A statement to which an historian listens, or one which he reads, is to him a ready-made statement. But the statement that such a statement is being made is not a ready-made statement. If he says to himself 'I am now reading or hearing a statement to such and such effect', he is himself making a statement ; but it is not a second-hand statement, it is autonomous. He makes it on his own authority. And it is this autonomous

( 276) statement that is the scientific historian's starting-point. The evidence from which the constable infers that the rector's daughter suspects Richard Roe is not her statement ' I killed John Doe', but his own statement 'the rector's daughter tells me that she killed John Doe'.

If the scientific historian gets his conclusions not from the statement that he finds ready-made, but from his own autonomous statement of the fact that such statements are made, he can get conclusions even when no statements are made to him. The premisses of his argument are his own autonomous statements: there is no need for these autonomous statements to be themselves statements about other statements. To illustrate once more from the story of John Doe. The premisses from which the Detective-Inspector argued to the innocence of Richard Roe were all premisses of the Detective-Inspector's own stating, autonomous statements resting on no authority but his own: and not one of them was a statement about statements made by anybody else. The essential points were that Richard Roe had got his shoes muddy while going away from the rectory, that no mud was to be seen in John Doe's study, and that the circumstances of the murder had been such that he would not have stopped to clean or remove his shoes. Each of these three points, in its turn, was the conclusion of an inference, and the statements upon which they severally rested were no more statements about other people's statements than were these three points themselves. Again : the ultimate case against the rector did not logically depend upon any statements made by the Detective-Inspector about statements made by other persons. It depended upon the presence of certain objects in a certain dustbin, and of certain paint-smears on the cuff of a jacket made in the conventional clerical style and shrunk by wetting; and these facts were vouched for by his own observation. I do not mean that the scientific historian can work better when no statements are made to him about the subjects on which he is working ; it would be a pedantical way of avoiding scissors-and-paste history, to avoid occasions of this type which might be a trap for the weaker brethren ; what I mean is that he is not dependent on such statements being made.

This is important because it settles by appeal to principle a controversy which, even if it is no longer so urgent as it was, has

( 277) not yet ceased to echo in the minds of historians. This was the controversy between those who maintained that history was ultimately dependent on 'written sources', and those who maintained that it could also be constructed from ' unwritten sources'. The terms were unhappily chosen. ' Written sources' were not conceived as excluding oral sources, or as having any special connexion with handwriting as distinct from chiselling in stone or the like. ' Written sources', in fact, meant sources containing ready-made statements asserting or implying alleged facts belonging to the subject in which the historian was interested. ' Unwritten sources' meant archaeological material, potsherds, and so forth, connected with the same subject. Of course, the word 'source' was in no sense applicable to these, for a source means something from which water or the like is drawn ready made ; in the case of history, something from which the historian's statements are drawn ready made, and the point of describing potsherds as ' unwritten sources' was to indicate that, not being texts, they contained no ready-made statements and were therefore not written sources. (Inscribed potsherds or ' ostraka' were, of course, 'written sources'.

In effect, this was a controversy between people who believed that scissors-and-paste history was the only possible kind and people who, without impugning the validity of scissors-and-paste methods, claimed that there could be history without them. According to my own recollection the controversy was alive, though giving one an impression of obsolescence, in academic circles in this country thirty years ago ; all statements of the issue, so far as I can recall them, were extremely confused, and the philosophers of the time, though it gave them an excellent opportunity for doing a useful job of work on a subject of high philosophical interest, cared for none of these things. My impression is that the controversy fizzled out in the feeblest of compromises, the partisans of scissors-and-paste history accepting the principle that 'unwritten sources' could give valid results, but insisting that this could happen only on a verysmall scale and when they were used as an auxiliary arm to ' written sources' ; and only about low matters like industry and commerce, into which an historian with the instincts of a gentleman would not inquire. This amounted to saying that historians brought up to regard history as an affair of scissors

( 278) and paste were beginning, very timidly, to recognize the possibility of something quite different ; but that when they tried to convert this possibility into an actuality they were still too incompletely fledged for any but the shortest flights.

(x) Question and evidence

If history means scissors-and-paste history, where the historian depends on ready-made statements for all his knowledge about his subject, and where the texts in which he finds these statements are called his sources, it is easy to define a source in away which has some practical utility. A source is a text containing a statement or statements about the subject; and this definition has some practical utility because it helps the historian to divide the whole of extant literature, once he has determined his subject, into texts which might serve him as sources, and must therefore be looked at, and those which cannot, and may therefore be ignored. What he has to do is to run over his library shelves, or his bibliography of the period, asking himself at every title: 'Could this contain anything about my subject ?' And, in case he cannot give the answer out of his head, aids of several kinds have been provided: notably indexes and specialized or classified bibliographies. Even with all these aids, he may still miss an important piece of testimony, and thus provide sport for his friends ; but on any given question the amount of testimony that exists is a finite quantity, and it is theoretically possible to exhaust it.

Theoretically, but not always practically : for the amount may be so large, and some parts of it so difficult of access, that no historian can hope to see it all. And one sometimes hears people complaining that nowadays so much raw material for history is being preserved that the task of using it is becoming impossible ; and sighing for the good old days when books were few and libraries small, and an historian could hope to master his subject. What these complaints mean is that the scissors-and-paste historian is on the horns of a dilemma. If he possesses only a small amount of testimony about his subject, he wants more; because any new piece of testimony about it would, if really new, throw' new light on it, and might make the view he is actually putting forward untenable. So, however much testimony he has, his zeal as an historian makes him want more.

( 279)

But if he has a large amount of testimony, it becomes so difficult to manipulate and work up into a convincing narrative that, speaking as a mere weak mortal, he wishes he had less.

Consciousness of this dilemma has often driven men into scepticism about the very possibility of historical knowledge. And quite rightly, if knowledge means scientific knowledge and history means scissors-and-paste history. Scissors-and-paste historians who brush the dilemma aside with the blessed word 'hypercriticism' are only confessing that in their own professional practice they do not find that it troubles them, because they work to such a low standard of scientific cogency that their consciences become anaesthetized. Such cases in contemporary life are highly interesting, because in the history of science one often meets with them and wonders how such extraordinary blindness was possible. The answer is that the people who exhibit it have committed themselves to an impossible task, in this case the task of scissors-and-paste history, and since for practical reasons they cannot back out of it they have to blind themselves to its impossibility. The scissors-and-paste historian protects himself from seeing the truth about his own methods by carefully choosing subjects which he is able to 'get away' with, exactly as the nineteenth-century landscape-painter protected himself from seeing that his theory of landscape was all wrong by choosing what he called paintable subjects. The subjects must be those about which a certain amount of testimony is accessible, not too little and not too much ; not so uniform as to give the historian nothing to do, not so divergent as to baffle his endeavours to do it. Practised on these principles, history was at worst a parlour game, and at best an elegant accomplishment. I have used the past tense ; I leave it to the conscience of historians who are capable of self-criticism to decide how far I might justly have used the present.

If history means scientific history, for 'source' we must read 'evidence'. And when we try to define 'evidence' in the same spirit in which we defined 'sources', we find it very difficult. There is no short and easy test by which we can decide whether a given book is or is not capable of providing evidence about a given subject, and indeed no reason why we should limit our search to books. Indexes and bibliographies of sources are of no use at all to a scientific historian. This is not to say that he

( 280) cannot use indexes and bibliographies ; he can and does ; but they are indexes and bibliographies not of sources but of monographs or the like : not of evidence, but of previous discussions which he can take as a starting-point for his own. Consequently, whereas the books mentioned in a bibliography for the use of a scissors-and-paste historian will be, roughly speaking, valuable in direct proportion to their antiquity, those mentioned in a bibliography for the use of a scientific historian will be, roughly speaking, valuable in direct proportion to their newness.

In my fable there is only one obvious characteristic common to all the pieces of evidence used by the Detective-Inspector in his argument : they are all things observed by himself. If we ask what kind of things, it is not easy to give an answer. They include such things as the existence of certain footprints in certain mud, their number, position, and direction, their resemblance to prints produced by a certain pair of shoes, and the absence of any others ; the absence of mud on the floor of a certain room ; the position of a dead body, the position of a dagger in its back, and the shape of the chair in which it is sitting ; and so on, a most variegated collection. This, I think, we can safely say about it : that no one could possibly know what could or could not find a place in it until he had got all his questions not only formulated but answered. In scientific history anything is evidence which is used as evidence, and no one can know what is going to be useful as evidence until he has had occasion to use it.

Let us put this by saying that in scissors-and-paste history, if we allow ourselves to describe testimony—loosely, I admit—by the name of evidence, there is potential evidence and there is actual evidence. The potential evidence about a subject is all the extant statements about it. The actual evidence is that part of these statements which we decide to accept. But in scientific history the idea of potential evidence disappears ; or, if we like to put the same fact in these other words, everything in the world is potential evidence for any subject whatever. This will be a distressing idea to anyone whose notions of historical method are fixed in a scissors-and-paste mould ;,for how, he will ask, are we to discover what facts are actually of service to us, unless we can first of all round up the facts that might be of service to us ? To a person who understands the nature of scientific thinking,

( 281) whether historical or any other, it will present no difficulty. He will realize that, every time the historian asks a question, he asks it because he thinks he can answer it : that is to say, he has already in his mind a preliminary and tentative idea of the evidence he will be able to use. Not a definite idea about potential evidence, but an indefinite idea about actual evidence. To ask questions which you see no prospect of answering is the fundamental sin in science, like giving orders which you do not think will be obeyed in politics, or praying for what you do not think God will give you in religion. Question and evidence, in history, are correlative. Anything is evidence which enables you to answer your question—the question you are asking now. A sensible question (the only kind of question that a scientifically competent man will ask) is a question which you think you have or are going to have evidence for answering. If you think you have it here and now, the question is an actual question, like the question 'What position was John Doe in when he was stabbed ? ' If you think you are going to have it the question is a deferred question, like the question ' Who killed John Doe ? '

It was a correct understanding of this truth that underlay Lord Acton's great precept, 'Study problems, not periods'. Scissors-and-paste historians study periods ; they collect all the extant testimony about a certain limited group of events, and hope in vain that something will come of it. Scientific historians study problems : they ask questions, and if they are good historians they ask questions which they see their way to answering. It was a correct understanding of the same truth that led Monsieur Hercule Poirot to pour scorn on the 'human blood-hound' who crawls about the floor trying to collect everything, no matter what, which might conceivably turn out to be a clue ;and to insist that the secret of detection was to use what, with possibly wearisome iteration, he called ' the little grey cells'. You can't collect your evidence before you begin thinking, he meant : because thinking means asking questions (logicians, please note), and nothing is evidence except in relation to some definite question. The difference between Poirot and Holmes in this respect is deeply significant of the change that has taken place in the understanding of historical method in the last forty years. Lord Acton was preaching his doctrine in the heyday of Sherlock Holmes, in his inaugural lecture at Cambridge in 1895 ;

( 282) but it was caviare to the general. In Monsieur Poirot's time, to judge by his sales, the general cannot have too much of it. The revolution which dethroned the principles of scissors-and-paste history, and replaced them by those of scientific history, had become common property.


  1. The reader will perhaps forgive me a personal reminiscence here. I was still a young man when a very distinguished visitor addressed an academic society on an archaeological subject that came within my special field of studies. The point he made was new and revolutionary, and it was easy for me to see that he had proved it up to the hilt. I imagined, foolishly enough, that so lucid and cogent a piece of reasoning must convince any hearer, even one who previously knew nothing about its subject-matter. I was at first much disconcerted, but in the long run greatly instructed, by finding that the demonstration had quite failed to convince the (very learned and acute) logicians in the audience.

Valid HTML 4.01 Strict Valid CSS2