The Idea of History
Epilegomena: 2: The Historical Imagination
R. G. Collingwood
An inquiry into the nature of historical thinking is among the tasks which philosophy may legitimately undertake ; and at the present time  there are reasons, as it seems to me, for thinking such an inquiry not only legitimate but necessary. For there is a sense in which, at particular periods of history, particular philosophical problems are, as it were, in season, and claim the special attention of a philosopher anxious to be of service to his age. In part, the problems of philosophy are unchanging ; in part, they vary from age to age, according to the special
( 232) characteristics of human life and thought at the time ; and in the best philosophers of every age these two parts are so interwoven that the permanent problems appear sub specie saeculi, and the special problems of the age sub specie aeternitatis. Whenever human thought has been dominated by some special interest, the most fruitful philosophy of the age has reflected that domination ; not passively, by mere submission to its influence, but actively, by making a special attempt to understand it and placing it in the focus of philosophical inquiry.
In the Middle Ages theology was the interest that served in this way to focus philosophical speculation. In the seventeenth century it was physical science. To-day, when we conventionally date the beginnings of modern philosophy to the seventeenth century, we mean, I think, that the scientific interest which then began to dominate human life still dominates it. But if we compare the seventeenth-century mind, in its general orientation, with that of to-day, by comparing the subjects dealt with in their literature, we can hardly fail to be struck by one important difference. Since the time of Descartes, and even since the time of Kant, mankind has acquired a new habit of thinking historically. I do not mean that there were no historians worthy of the name until a century and a half ago ; that would be untrue : I do not even mean that since then the bulk of historical knowledge and the output of historical books have enormously increased; that would be true but relatively unimportant. What I mean is that during this time historical thought has worked out a technique of its own, no less definite in its character and certain in its results than its elder sister, the technique of natural science ;and that, in thus entering upon the sichere Gang einer Wissenschaft, it has taken a place in human life from which its influence, has permeated and to some extent transformed every department of thought and action.
Among others, it has profoundly influenced philosophy ; but on the whole the attitude of philosophy towards this influence has been more passive than active. Some philosophers are inclined to welcome it ; others to resent it ; comparatively few have thought philosophically about it. Attempts have been made, chiefly in Germany and Italy, to answer the questions: What is historical thinking ? and What light does it throw on the traditional problems of philosophy ? and by answering these
( 233) questions to do for the historical consciousness of to-day what Kant's transcendental analytic did for the scientific consciousness of the eighteenth century. But for the most part, and especially in this country, it has been usual to ignore all such questions, and to discuss the problems of knowledge in seeming unawareness that there is such a thing as history. This custom can of course be defended. It may be argued that history is not knowledge at all, but only opinion, and unworthy of philosophical study. Or it may be argued that, so far as it is knowledge, its problems are those of knowledge in general, and call for no special treatment. For myself, I cannot accept either defence. If history is opinion, why should philosophy on that account ignore it ? If it is knowledge, why should philosophers not study its methods with the same attention that they give to the very different methods of science ? And when I read the works of even the greatest contemporary and recent English philosophers, admiring them deeply and learning from them more than I can hope to acknowledge, I find myself constantly haunted by the thought that their accounts of knowledge, based as they seem to be primarily on the study of perception and of scientific thinking, not only ignore historical thinking but are actually inconsistent with there being such a thing.
No doubt, historical thought is in one way like perception. Each has for its proper object something individual. What I perceive is this room, this table, this paper. What the historian thinks about is Elizabeth or Marlborough, the Peloponnesian War or the policy of Ferdinand and Isabella. But what we perceive is always the this, the here, the now. Even when we hear a distant explosion or see a stellar conflagration long after it has happened, there is still a moment at which it is here and now perceptible, when it is this explosion, this new star. Historical thought is of something which can never be a this, because it is never a here and now. Its objects are events which have finished happening, and conditions no longer in existence. Only when they are no longer perceptible do they become objects for historical thought. Hence all theories of knowledge that conceive it as a transaction or relation between a subject and an object both actually existing, and confronting or compresent to one another, theories that take acquaintance as the essence of knowledge, make history impossible.
In another way history resembles science: for in each of them knowledge is inferential or reasoned. But whereas science lives in a world of abstract universals, which are in one sense everywhere and in another nowhere, in one sense at all times and in another at no time, the things about which the historian reasons are not abstract but concrete, not universal but individual, not indifferent to space and time but having a where and a when of their own, though the where need not be here and the when cannot be now. History, therefore, cannot be made to square with theories according to which the object of knowledge is abstract and changeless, a logical entity towards which the mind may take up various attitudes.
Nor is it possible to give an account of knowledge by combining theories of these two types. Current philosophy is full of such combinations. Knowledge by acquaintance and knowledge by description; eternal objects and the transient situations into which they are ingredient ; realm of essence and realm of matter ; in these and other such dichotomies, as in the older dichotomies of matters of fact and relations between ideas, or truths of fact and truths of reason, provision is made for the peculiarities both of a perception which grasps the here and now, and of the abstract thought that apprehends the everywhere and always: the aisqhsiz and nohsiz of philosophical tradition. But just as history is neither aisqhsiz nor nohsiz , so it is not a combination of the two. It is a third thing, having some of the characteristics of each, but combining them in a way impossible to either. It is not partly acquaintance with transient situations and partly reasoned knowledge of abstract entities. It is wholly a reasoned knowledge of what is transient and concrete.
My purpose here is to offer a brief account of this third thing which is history ; and I will begin by stating what may be called the common-sense theory of it, the theory which most people believe, or imagine themselves to believe, when first they reflect on the matter.
According to this theory, the essential things in history are memory and authority. If an event or a state of things is to be historically known, first of all some one must be acquainted with it ; then he must remember it ; then he must state his recollection of it in terms intelligible to another ; and finally that other must accept the statement as true. History is thus
( 235) the believing some one else when he says that he remembers something. The believer is the historian ; the person believed is called his authority.
This doctrine implies that historical truth, so far as it is at all accessible to the historian, is accessible to him only because it exists ready made in the ready-made statements of his authorities. These statements aie to him a sacred text, whose value depends wholly on the unbrokenness of the tradition they represent. He must therefore on no account tamper with them. He must not mutilate them ; he must not add to them ; and, above all, he must not contradict them. For if he takes it upon himself to pick and choose, to decide that some of his authority's statements are important and others not, he is going behind his authority's back and appealing to some other criterion ; and this, on the theory, is exactly what he cannot do. If he adds to them, interpolating in them constructions of his own devising, and accepting these constructions as additions to his knowledge, he is believing something for a reason other than the fact that his authority has said it ; and this again he has no right to do. Worst of all, if he contradicts them, presuming to decide that his authority has misrepresented the facts, and rejecting his statements as incredible, he is believing the opposite of what he has been told, and committing the worst possible offence against the rules of his craft. The authority may be garrulous, discursive, a gossip and a scandal-monger ; he may have overlooked or forgotten or omitted facts ; he may have ignorantly or wilfully misstated them ; but against these defects the historian has no remedy. For him, on the theory, what his authorities tell him is the truth, the whole accessible truth, and nothing but the truth.
These consequences of the common-sense theory have only to be stated in order to be repudiated. Every historian is aware that on occasion he does tamper in all these three ways with what he finds in his authorities. He selects from them what he thinks important, and omits the rest ; he interpolates in them things which they do not explicitly say ; and he criticizes them by rejecting or amending what he regards as due to misinformation or mendacity. But I am not sure whether we historians always realize the consequences of what we are doing. In general, when we reflect on our own work, we seem to accept
( 236) what I have called the common-sense theory, while claiming our own rights of selection, construction, and criticism. No doubt these rights are inconsistent with the theory ; but we attempt to soften the contradiction by minimizing the extent to which they are exercised, thinking of them as emergency measures, a kind of revolt into which the historian may be driven at times by the exceptional incompetence of his authorities, but which does not fundamentally disturb the normal peaceful régime in which he placidly believes what he is told because he is told to believe it. Yet these things, however seldom they are done, are either historical crimes or facts fatal to the theory: for on the theory they ought to be done, not rarely, but never. And in fact they are neither criminal nor exceptional. Throughout the course of his work the historian is selecting, constructing, and criticizing ; it is only by doing these things that he maintains his thought upon the sichere Gang einer Wissenschaft. By explicitly recognizing this fact it is possible to effect what, again borrowing a Kantian phrase, one might call a Copernican revolution in the theory of history: the discovery that, so far from relying on an authority other than himself, to whose statements his thought must conform, the historian is his own authority and his thought autonomous, self-authorizing, possessed of a criterion to which his so-called authorities must conform and by reference to which they are criticized.
The autonomy of historical thought is seen at its simplest in the work of selection. The historian who tries to work on the common-sense theory, and accurately reproduce what he finds in his authorities, resembles a landscape-painter who tries to work on that theory of art which bids the artist copy nature. He may fancy that he is reproducing in his own medium the actual shapes and colours of natural things ; but however hard he tries to do this he is always selecting, simplifying, schematizing, leaving out what he thinks unimportant and putting in what he regards as essential. It is the artist, and not nature, that is responsible for what goes into the picture. In the same way, no historian, not even the worst, merely copies out his authorities ;even if he puts in nothing of his own (which is never really possible), he is always leaving out things which, for one reason or another, he decides that his own work does not need or cannot use. It is he, therefore, and not his authority, that is
( 237) responsible for what goes in. On that question he is his own master: his thought is to that extent autonomous.
An even clearer exhibition of this autonomy is found in what I have called historical construction. The historian's authorities tell him of this or that phase in a process whose intermediate phases they leave undescribed ; he then interpolates these phases for himself. His picture of his subject, though it may consist in part of statements directly drawn from his authorities, consists also, and increasingly with every increase in his competence as an historian, of statements reached inferentially from those according to his own criteria, his own rules of method, and his own canons of relevance. In this part of his work he is never depending on his authorities in the sense of repeating what they tell him ; he is relying on his own powers and constituting himself his own authority ; while his so-called authorities are now not authorities at all but only evidence.
The clearest demonstration of the historian's autonomy, however, is provided by historical criticism. As natural science finds its proper method when the scientist, in Bacon's metaphor, puts Nature to the question, tortures her by experiment in order to wring from her answers to his own questions, so history fin$its proper method when the historian puts his authorities in the witness-box, and by cross-questioning extorts from them information which in their original statements they have withheld, either because they did not wish to give it or because they did not possess it. Thus, a commander's dispatches may claim a victory ; the historian, reading them in a critical spirit, will ask: `If it was a victory, why was it not followed up in this or that way ? ' and may thus convict the writer of concealing the truth. Or, by using the same method, he may convict of ignorance a less critical predecessor who has accepted the version of the battle given him by the same dispatches.
The historian's autonomy is here manifested in its extremest form, because it is here evident that somehow, in virtue of his activity as an historian, he has it in his power to reject something explicitly told him by his authorities and to substitute something else. If that is possible, the criterion of historical truth cannot be the fact that a statement is made by an authority. It is the truthfulness and the information of the so-called authority that are in question ; and this question the historian has to answer
( 238) for himself, on his own authority. Even if he accepts what his authorities tell him, therefore, he accepts it not on their authority but on his own ; not because they say it, but because it satisfies his criterion of historical truth.
The common-sense theory which bases history upon memory and authority needs no further refutation. Its bankruptcy is evident. For the historian there can never be authorities, because the so-called authorities abide a verdict which only he can give. Yet the common-sense theory may claim a qualified and relative truth. The historian, generally speaking, works at a subject which others have studied before him. In proportion as he is more of a novice, either in this particular subject or in history as a whole, his forerunners are, relatively to his incompetence, authoritative ; and in the limiting case where his incompetence and ignorance were absolute, they could be called authorities without qualification. As he becomes more and more master of his craft and his subject, they become less and less his authorities, more and more his fellow students, to be treated with respect or contempt according to their deserts.
And as history does not depend on authority, so it does not depend upon memory. The historian can rediscover what has been completely forgotten, in the sense that no statement of it has reached him by an unbroken tradition from eyewitnesses. He can even discover what, until he discovered it, no one ever knew to have happened at all. This he does partly by the critical treatment of statements contained in his sources, partly by the use of what are called unwritten sources, which are increasingly employed as history becomes increasingly sure of its own proper methods and its own proper criterion.
I have spoken of the criterion of historical truth. What is this criterion ? According to the common-sense theory, it is the agreement of the statements made by the historian with those which he finds in his authorities. This answer we now know to be false, and we must seek another. We cannot renounce the search. Some answer to the question there must be, for without a criterion there can be no criticism. One answer to this question was offered by the greatest English philosopher of our time in his pamphlet on The Presuppositions of Critical History. Bradley's essay was an early work with which in his maturity he was dissatisfied ; but, unsatisfactory though it cer-
( 239) -tainly is, it bears the stamp of his genius. In it Bradley faces the question how it is possible for the historian, in defiance of the common-sense theory, to turn the tables on his so-called authorities and to say 'This is what our authorities record, but what really happened must have been not this but that'.
His answer to this question was that our experience of the world teaches us that some kinds of things happen and others do not ; this experience, then, is the criterion which the historian brings to bear on the statements of his authorities. If they tell him that things happened of a kind which, according to his experience, does not happen, he is obliged to disbelieve them ; if the things which they report are of a kind which according to his experience does happen, he is free to accept their statements.
There are many obvious objections to this idea, on which I shall not insist. It is deeply tinged with the empiricist philosophy against which Bradley was soon so effectively to rebel. But apart from this there are certain special points in which the argument appears to me defective.
First, the proposed criterion is a criterion not of what did happen but of what could happen. It is in fact nothing but Aristotle's criterion of what is admissible in poetry; and hence it does not serve to discriminate history from fiction. It would no doubt be satisfied by the statements of an historian, but it would be satisfied no less adequately by those of an historical novelist. It cannot therefore be the criterion of critical history.
Secondly, because it can never tell us what did happen, we are left to rely for that on the sheer authority of our informant. We undertake, when we apply it, to believe everything our informant tells us so long as it satisfies the merely negative criterion of being possible. This is not to turn the tables on our authorities ; it is blindly to accept what they tell us. The critical attitude has not been achieved.
Thirdly, the historian's experience of the world in which he lives can only help him to check, even negatively, the statements of his authorities in so far as they are concerned not with history but with nature, which has no history. The laws of nature have always been the same, and what is against nature now was against nature two thousand years ago ; but the historical as distinct from the natural conditions of man's life differ so much at different times that no argument from analogy will hold.
( 240) That the Greeks and Romans exposed their new-born children in order to control the numbers of their population is no less true for being unlike anything that happens in the experience of contributors to the Cambridge Ancient History. In point of fact Bradley's treatment of the subject grew not out of the ordinary course of historical study but out of his interest in the credibility of the New Testament narratives, and in particular their miraculous element ; but a criterion which only serves in the case of miracle is of sadly little use to the weekday historian.
Bradley's essay, inconclusive though it is, remains memorable for the fact that in it the Copernican revolution in the theory of historical knowledge has been in principle accomplished. For the common-sense theory, historical truth consists in the historian's beliefs conforming to the statements of his authorities ; Bradley has seen that the historian brings with him to the study of his authorities a criterion of his own by reference to which the authorities themselves are judged. What it is, Bradley failed to discover. It remains to be seen whether, sixty years later, his problem, which in the meantime I believe no English-speaking philosopher has discussed in print, can be advanced beyond the point at which he left it.
I have already remarked that, in addition to selecting from among his authorities' statements those which he regards as important, the historian must in two ways go beyond what his authorities tell him. One is the critical way, and this is what Bradley has attempted to analyse. The other is the constructive way. Of this he has said nothing, and to this I now propose to return. I described constructive history as interpolating, between the statements borrowed from our authorities, other statements implied by them. Thus our authorities tell us that on one day Caesar was in Rome and on a later day in Gaul ; they tell us nothing about his journey from one place to the other, but we interpolate this with a perfectly good conscience.
This act of interpolation has two significant characteristics. First, it is in no way arbitrary or merely fanciful: it is necessary or, in Kantian language, a priori. If we filled up the narrative of Caesar's doings with fanciful details such as the names of the persons he met on the way, and what he said to them, the construction would be arbitrary: it would be in fact the kind of construction which is done by an historical novelist. But if our
( 241) construction involves nothing that is not necessitated by the evidence, it is a legitimate historical construction of a kind without which there can be no history at all.
Secondly, what is in this way inferred is essentially something imagined. If we look out over the sea and perceive a ship, and five minutes later look again and perceive it in a different place, we find ourselves obliged to imagine it as having occupied intermediate positions when we were not looking. That is already an example of historical thinking ; and it is not otherwise that we find ourselves obliged to imagine Caesar as having travelled from Rome to Gaul when we are told that he was in these different places at these successive times.
This activity, with this double character, I shall call a priori imagination ; and, though I shall have more to say of it hereafter, for the present I shall be content to remark that, however unconscious we may be of its operation, it is this activity which, bridging the gaps between what our authorities tell us, gives the historical narrative or description its continuity. That the historian must use his imagination is a commonplace ; to quote Macaulay's Essay on History, 'a perfect historian must possess an imagination sufficiently powerful to make his narrative affecting and picturesque' ; but this is to underestimate the part played by the historical imagination, which is properly not ornamental but structural. Without it the historian would have no narrative to adorn. The imagination, that 'blind but indispensable faculty' without which, as Kant has shown, we could never perceive the world around us, is indispensable in the same way to history: it is this which, operating not capriciously as fancy but in its a priori form, does the entire work of historical construction.
Two misunderstandings may here be forestalled. First, it may be thought that by imagining we can present to ourselves only what is imaginary in the sense of being fictitious or unreal. This prejudice need only be mentioned in order to be dispelled. If I imagine the friend who lately left my house now entering his own, the fact that I imagine this event gives me no reason to believe it unreal. The imaginary, simply as such, is neither unreal nor real.
Secondly, to speak of a priori imagination may seem a paradox, for it may be thought that imagination is essentially capricious,
( 242) arbitrary, merely fanciful. But in addition to its historical function there are two other functions of a priori imagination which are, or ought to be, familiar to all. One is the pure or free, but by no means arbitrary, imagination of the artist. A man writing a novel composes a story where parts are played by various characters. Characters and incidents are all alike imaginary ; yet the whole aim of the novelist is to show the characters acting and the incidents developing in a manner determined by a necessity internal to themselves. The story, if it is a good story, cannot develop otherwise than as it does ; the novelist in imagining it cannot imagine it developing except as it does develop. Here, and equally in all other kinds of art, the a priori imagination is at work. Its other familiar function is what may be called the perceptual imagination, supplementing and consolidating the data of perception in the way so well analysed by Kant, by presenting to us objects of possible perception which are not actually perceived : the under side of this table, the inside of an unopened egg, the back of the moon. Here again the imagination is a priori : we cannot but imagine what cannot but be there. The historical imagination differs from these not in being a priori, but in having as its special task to imagine the past: not an object of possible perception, since it does not now exist, but able through this activity to become an object of our thought.
The historian's picture of his subject, whether that subject be a sequence of events or a past state of things, thus appears as a web of imaginative construction stretched between certain fixed points provided by the statements of his authorities ; and if these points are frequent enough and the threads spun from each to the next are constructed with due care, always by the a priori imagination and never by merely arbitrary fancy, the whole picture is constantly verified by appeal to these data, and runs little risk of losing touch with the reality which it represents.
Actually, this is very much how we do think of historical work, when the common-sense theory has ceased to satisfy us, and we have become aware of the part played in it by the constructive imagination. But such a conception is in one way seriously at fault: it overlooks the no less important part played by criticism. We think of our web of construction as pegged down, so to speak, to the facts by the statements of authorities,
( 243) which we regard as data or fixed points for the work of construction. But in so thinking we have slipped back into the theory, which we now know to be false, that truth is given us ready made in these statements. We know that truth is to be had, not by swallowing what our authorities tell us, but by criticizing it ; and thus the supposedly fixed points between which the historical imagination spins its web are not given to us ready made, they must be achieved by critical thinking.
There is nothing other than historical thought itself, by appeal to which its conclusions may be verified. The hero of a detective novel is thinking exactly like an historian when, from indications of the most varied kinds, he constructs an imaginary picture of how a crime was committed, and by whom. At first, this is a mere theory, awaiting verification, which must come to it from without. Happily for the detective, the conventions of that literary form dictate that when his construction is complete it shall be neatly pegged down by a confession from the criminal, given in such circumstances that its genuineness is beyond question. The historian is less fortunate. If, after convincing himself by a study of the evidence already available that Bacon wrote the plays of Shakespeare or that Henry VII murdered the Princes in the Tower, he were to find an autograph document confessing the fact, he would by no means have verified his conclusions; the new document, so far from closing the inquiry, would only have complicated it by raising a new problem, the problem of its own authenticity.
I began by considering a theory according to which everything is given : according to which all truth, so far as any truth is accessible to the historian, is provided for him ready made in the ready-made statements of his authorities. I then saw that much of what he takes for true is not given in this way but constructed by his a priori imagination ; but I still fancied that this imagination worked inferentially from fixed points given in the same sense. I am now driven to confess that there are for historical thought no fixed points thus given: in other words, that in history, just as there are properly speaking no authorities, so there are properly speaking no data.
Historians certainly think of themselves as working from data; where by data they mean historical facts possessed by them ready made at the beginning of a certain piece of historical
( 244) research. Such a datum, if the research concerns the Peloponnesian War, would be, for example, a certain statement of Thucydides, accepted as substantially true. But when we ask what gives historical thought this datum, the answer is obvious :historical thought gives it to itself, and therefore in relation to historical thought at large it is not a datum but a result or achievement. It is only our historical knowledge which tells us that these curious marks on paper are Greek letters ; that the words which they form have certain meanings in the Attic dialect ; that the passage is authentic Thucydides, not an interpolation or corruption ; and that on this occasion Thucydides knew what he was talking about and was trying to tell the truth. Apart from all this, the passage is merely a pattern of black marks on white paper: not any historical fact at all, but something existing here and now, and perceived by the historian. All that the historian means, when he describes certain historical facts as his data, is that for the purposes of a particular piece of work there are certain historical problems relevant to that work which for the present he proposes to treat as settled ; though, if they are settled, it is only because historical thinking has settled them in the past, and they remain settled only until he or some one else decides to reopen them.
His web of imaginative construction, therefore, cannot derive its validity from being pegged down, as at first I described it, to certain given facts, That description represented an attempt to relieve him of the responsibility for the nodal points of his fabric, while admitting his responsibility for what he constructs between them. In point of fact, he is just as responsible for the one as for the other. Whether he accepts or rejects or modifies or reinterprets what his so-called authorities tell him, it is he that is responsible for the statement which, after duly criticizing them, he makes. The criterion that justifies him in making it can never be the fact that it has been given him by an authority.
This brings me back to the question what this criterion is. And at this point a partial and provisional answer can be given. The web of imaginative construction is something far more solid and powerful than we have hitherto realized. So far from relying for its validity upon the support of given facts, it actually serves as the touchstone by which we decide whether alleged facts are genuine. Suetonius tells me that Nero at one
( 245) time intended to evacuate Britain. I reject his statement, not because any better authority flatly contradicts it, for of course none does ; but because my reconstruction of Nero's policy based on Tacitus will not allow me to think that Suetonius is right. And if I am told that this is merely to say I prefer Tacitus to Suetonius, I confess that I do: but I do so just because I find myself able to incorporate what Tacitus tells me into a coherent and continuous picture of my own, and cannot do this for Suetonius.
It is thus the historian's picture of the past, the product of his own a priori imagination, that has to justify the sources used in its construction. These sources are sources, that is to say, credence is given to them, only because they are in this way justified. For any source may be tainted: this writer prejudiced, that misinformed ; this inscription misread by a bad epigraphist, that blundered by a careless stonemason ; this potsherd placed out of its context by an incompetent excavator, that by a blameless rabbit. The critical historian has to discover and correct all these and many other kinds of falsification. He does it, and can only do it, by considering whether the picture of the past to which the evidence leads him is a coherent and continuous picture, one which makes sense. The a priori imagination which does the work of historical construction supplies the means of historical criticism as well.
Freed from its dependence on fixed points supplied from without, the historian's picture of the past is thus in every detail an imaginary picture, and its necessity is at every point the necessity of the a priori imagination. Whatever goes into it, goes into it not because his imagination passively accepts it, but because it actively demands it.
The resemblance between the historian and the novelist, to which I have already referred, here reaches its culmination. Each of them makes it his business to construct a picture which is partly a narrative of events, partly a description of situations, exhibition of motives, analysis of characters. Each aims at making his picture a coherent whole, where every character and every situation is so bound up with the rest that this character in this situation cannot but act in this way, and we cannot imagine him as acting otherwise. The novel and the history must both of them make sense ; nothing is admissible in either except what is
( 246) necessary, and the judge of this necessity is in both cases the imagination. Both the novel and the history are self-explanatory, self-justifying, the product of an autonomous or self-authorizing activity ; and in both cases this activity is the a priori imagination.
As works of imagination, the historian's work and the novelist's do not differ. Where they do differ is that the historian's picture is meant to be true. The novelist has a single task only: to construct a coherent picture, one that makes sense. The historian has a double task: he has both to do this, and to construct a picture of things as they really were and of events as they really happened. This further necessity imposes upon him obedience to three rules of method, from which the novelist or artist in general is free.
First, his picture must be localized in space and time. The artist's need not ; essentially, the things that he imagines are imagined as happening at no place and at no date. Of Wuthering Heights it has been well said that the scene is laid in Hell, though the place-names are English ; and it was a sure instinct that led another great novelist to replace Oxford by Christminster, Wantage by Alfredston, and Fawley by Marychurch, recoiling against the discord of topographical fact in what should be a purely imaginary world.
Secondly, all history must be consistent with itself. Purely imaginary worlds cannot clash and need not agree ; each is a world to itself. But there is only one historical world, and everything in it must stand in some relation to everything else, even if that relation is only topographical and chronological.
Thirdly, and most important, the historian's picture stands in a peculiar relation to something called evidence. The only way in which the historian or any one else can judge, even tentatively, of its truth is by considering this relation ; and, in practice, what we mean by asking whether an historical statement is true is whether it can be justified by an appeal to the evidence: for a truth unable to be so justified is to the historian a thing of no interest. What is this thing called evidence, and what is its relation to the finished historical work ?
We already know what evidence is not. It is not ready-made historical knowledge, to be swallowed and regurgitated by the historian's mind. Everything is evidence which the historian
( 247) can use as evidence. But what can he so use ? It must be something here and now perceptible to him : this written page, this spoken utterance, this building, this fingerprint. And of all the things perceptible to him there is not one which he might not conceivably use as evidence on some question, if he came to it with the right question in mind. The enlargement of historical knowledge comes about mainly through finding how to use as evidence this or that kind of perceived fact which historians have hitherto thought useless to them.
The whole perceptible world, then, is potentially and in principle evidence to the historian. It becomes actual evidence in so far as he can use it. And he cannot use it unless he comes to it with the right kind of historical knowledge. The more historical knowledge we have, the more we can learn from any given piece of evidence ; if we had none, we could learn nothing. Evidence is evidence only when some one contemplates it historically. Otherwise it is merely perceived fact, historically dumb. It follows that historical knowledge can only grow out of historical knowledge ; in other words, that historical thinking is an original and fundamental activity of the human mind, or, as Descartes might have said, that the idea of the past is an `innate' idea.
Historical thinking is that activity of the imagination by which we endeavour to provide this innate idea with detailed content. And this we do by using the present as evidence for its own past. Every present has a past of its own, and any imaginative reconstruction of the past aims at reconstructing the past of this present, the present in which the act of imagination is going on, as here and now perceived. In principle the aim of any such act is to use the entire perceptible here-and-now as evidence for the entire past through whose process it has come into being. In practice, this aim can never be achieved. The perceptible here-and-now can never be perceived, still less interpreted, in its entirety ; and the infinite process of past time can never be envisaged as a whole. But this separation between what is attempted in principle and what is achieved in practice is the lot of mankind, not a peculiarity of historical thinking. The fact that it is found there only shows that herein history is like art, science, philosophy, the pursuit of virtue, and the search for happiness.
It is for the same reason that in history, as in all serious matters, no achievement is final. The evidence available for solving any given problem changes with every change of historical method and with every variation in the competence of historians. The principles by which this evidence is interpreted change too ; since the interpreting of evidence is a task to which a man must bring everything he knows : historical knowledge, knowledge of nature and man, mathematical knowledge, philosophical knowledge ; and not knowledge only, but mental habits and possessions of every kind: and none of these is unchanging. Because of these changes, which never cease, however slow they may appear to observers who take a short view, every new generation must rewrite history in its own way ; every new historian, not content with giving new answers to old questions, must revise the questions themselves ; and—since historical thought is a river into which none can step twice—even a single historian, working at a single subject for a certain length of time, finds when he tries to reopen an old. question that the question has changed.
This is not an argument for historical scepticism. It is only the discovery of a second dimension of historical thought, the history of history: the discovery that the historian himself, together with the here-and-now which forms the total body of evidence available to him, is a part of the process he is studying, has his own place in that process, and can see it only from' the point of view which at this present moment he occupies within it.
But neither the raw material of historical knowledge, the detail of the here-and-now as given him in perception, nor the various endowments that serve him as aids to interpreting this evidence, can give the historian his criterion of historical truth. That criterion is the idea of history itself : the idea of an imaginary picture of the past. That idea is, in Cartesian language, innate ; in Kantian language, a priori. It is not a chance product of psychological causes ; it is an idea which every man possesses as part of the furniture of his mind, and discovers himself to possess in so far as he becomes conscious of what it is to have a mind. Like other ideas of the same sort, it is one to which no fact of experience exactly corresponds. The historian, however long and faithfully he works, can never say that his work, even in crudest outline or in this or that smallest detail, is done once
(249) for all. He can never say that his picture of the past is at any point adequate to his idea of what it ought to be. But, however fragmentary and faulty the results of his work may be, the idea which governed its course is clear, rational, and universal. It is the idea of the historical imagination as a self-dependent, self-determining, and self-justifying form of thought.