*  ESSAY *

Extracts from The Poverty of Historicism

These should be read alongside the summary of the whole book.




WHETHER the true motive of scientific inquiry is the desire to know, that is to say, a purely theoretical or 'idle' curiosity, or whether we should rather understand science as an instrument for solving the practical problems that arise in the struggle for life, this is a question that need not be decided here. It will be admitted that the defenders of the rights of 'pure' or 'fundamental' research deserve every support in their fight against the narrow view, unfortunately again fashionable, that scientific research is justified only if it proves to be a sound investment. But even the somewhat extreme view (to which I personally incline) that science is most significant as one of the greatest spiritual adventures that man has yet known may be combined with a recognition of the importance of practical problems and practical tests for the progress of science, whether applied or pure; for practice is invaluable in scientific speculation, both as a spur and as a bridle. One need not espouse pragmatism in order to appreciate Kant's saying: 'To yield to every whim of curiosity, and to allow our passion for inquiry to be restrained by nothing but the limits of our ability, this shows an eagerness of mind not unbecoming to scholarship. But it is wisdom that has the merit of selecting, from among the innumerable problems which present themselves, those whose solution is important to mankind.'

The application of this to the biological and perhaps even more to the social sciences is clear. Pasteur's reform of the biological sciences was carried out under the stimulus of highly practical problems, which were in part industrial and agricultural. And social research nowadays has a practical urgency surpassing even that of cancer research. As Professor Hayek says, 'economic analysis has never been the product of detached intellectual curiosity about the why of social phenomena, but of an intense urge to reconstruct a world which gives rise to profound dissatisfaction'; and some of the social sciences, other than economics, that have not yet adopted this outlook, show by the barrenness of their results how urgently their speculations are in need of practical checks.

The need for the stimulus of practical problems is equally clear when we consider inquiries into the methods of scientific research, and more especially, into the methods of the generalizing or theoretical social sciences with which we are concerned here. The more fruitful debates on method are always inspired by certain practical problems which face the research worker; and nearly all debates on method which are not so inspired are characterized by that atmosphere of futile subtlety which has brought methodology into disrepute with the practical research worker. It should be realized that methodological debates of the more practical kind are not only useful but also necessary. In the development and improvement of method, as of science itself, we learn only by trial and error, and we need the criticism of others in order to find out our mistakes; and this criticism is the more important since the introduction of new methods may mean a change of a fundamental and revolutionary character. All this may be illustrated by examples such as the introduction of mathematical methods into economics, or of the socalled 'subjective' or 'psychological' methods into value theory. A more recent example is the combination of the methods of this theory with statistical methods ('demand analysis'). This last revolution in method was partly the outcome of prolonged and largely critical debates; a fact from which the apologist for the study of method may certainly draw encouragement.

A practical approach to the study of both the social sciences and their method is advocated by many of the followers of historicism who hope that they will be able to transform, by the use of historicist methods, the social sciences into a powerful instrument in the hands of the politician. It is this recognition of the practical task of the social sciences which provides something like a common ground for discussion between the historicists and some of their opponents; and I am ready to take my own stand on this common ground in order to criticize historicism as a poor method, unable to yield the results it promises.


Although in this study my topic is historicism, a doctrine of method with which I disagree, rather than those methods which, in my opinion, have been successful, and whose further and more conscious development I recommend, it will be useful to deal briefly with the successful methods first, so as to reveal to the reader my own bias and to clarify the point of view that underlies my criticism. For convenience, I shall label these methods 'piecemeal technology'.

The term 'social technology' (and even more the term 'social engineering' 1 which will be introduced in the next section) is likely to arouse suspicion, and to repel those whom it reminds of the 'social blueprints' of the collectivist planners, or perhaps even of the 'technocrats'. I realize this danger, and so I have added the word 'piecemeal', both to off-set undesirable associations and to express my conviction that 'piecemeal tinkering' (as it is sometimes called), combined with critical analysis, is the main way to practical results in the social as well as in the natural sciences. The social sciences have developed very largely through the criticism of proposals for social improvements or, more precisely, through attempts to find out whether or not some particular economic or political action is likely to produce an expected, or desired, result. 2 This approach, which might indeed be called the classical
one, is what I have in mind when I refer to the technological approach to social science, or to 'piecemeal social technology'.

Technological problems in the field of social science may be of a 'private' or of a 'public' character. For example, investigations into the technique of business administration, or into the effects of improved working conditions upon output, belong to the first group. Investigations into the effects of prison reform or universal health insurance, or of the stabilization of prices by means of tribunals, or of the introduction of new import duties, etc., upon, say, the equalization of incomes, belong to the second group; and so do some of the most urgent practical questions of the day, such as the possibility of controlling trade cycles; or the question whether centralized 'planning', in the sense of state management of production, is compatible with an effective democratic control of the administration; or the question of how to export democracy to the Middle East.

This emphasis upon the practical technological approach does not mean that the theoretical problems which will arise from the analysis of the practical problems should be excluded. On the contrary, it is one of my main points that the technological approach is likely to prove fruitful in giving rise to significant problems of a purely theoretical kind. But besides helping us in the fundamental task of selecting problems, the technological approach imposes a discipline on our speculative inclinations (which, especially in the field of sociology proper, are liable to lead us into the region of metaphysics); for it forces us to submit our theories to definite standards, such as standards of clarity and practical testability. I can perhaps make my point about the technological approach by saying that sociology in particular--and perhaps the social sciences in general--should look, not so much for 'its Newton or its Darwin' as for its Galileo, or its Pasteur. 

This and my previous references to an analogy between the methods of the social and the natural sciences are likely to provoke as much opposition as our choice of terms like 'social technology' and 'social engineering' (this in spite of the important qualification expressed by the word 'piecemeal'). So I had better say that I fully appreciate the importance of the fight against a dogmatic methodological naturalism or 'scientism' (to use Professor Hayek's term). Nevertheless I do not see why we should not make use of this analogy as far as it is fruitful, even though we recognize that it has been badly misused and misrepresented in certain quarters. Besides, we can hardly offer a stronger argument against these dogmatic naturalists than one that shows that some of the methods they attack are fundamentally the same as the methods used in the natural sciences.

A prima facie objection against what we call the technological approach is that it implies the adoption of an 'activist' attitude towards the social order and that it is therefore liable to prejudice us against the anti-interventionist or 'passivist' view: the view that if we are dissatisfied with existing social or economic conditions, it is because we do not understand how they work and why active intervention could only make matters worse. Now I must admit that I am certainly out of sympathy with this 'passivist' view, and that I even believe that a policy of universal anti-interventionism is untenable--even on purely logical grounds, since its supporters are bound to recommend political intervention aimed at preventing intervention. Nevertheless, the technological approach as such is neutral in this matter (as indeed it ought to be), and by no means incompatible with anti-interventionism. On the contrary, I think that antiinterventionism involves a technological approach. For to assert that interventionism makes matters worse is to say that certain political actions would not have certain effects--to wit, not the desired ones; and it is one of the most characteristic tasks of any technology to point out what cannot be achieved.

It is worth while to consider this point more closely. As I have shown elsewhere, every natural law can be expressed by asserting that such and such a thing cannot happen; that is to say, by a sentence in the form of the proverb: 'You can't carry water in a sieve.' For example, the law of conservation of energy can be expressed by: 'You cannot build a perpetual motion machine'; and that of entropy by: 'You cannot build a machine which is a hundred per cent efficient.' This way of formulating natural laws is one which makes their technological significance obvious and it may therefore be called the 'technological form' of a natural law. If we now consider anti-interventionism in this light, then we see at once that it may well be expressed by sentences of the form: 'You cannot achieve such and such results', or perhaps, 'You cannot achieve such and such ends without such and such concomitant effects.' But this shows that anti-interventionism can be called a typically technological doctrine.

It is not, of course, the only one in the realm of social science. On the contrary, the significance of our analysis lies in the fact that it draws attention to a really fundamental similarity between the natural and the social sciences. I have in mind the existence of sociological laws or hypotheses which are analogous to the laws or hypotheses of the natural sciences. Since the existence of such sociological laws or hypotheses (other than so-called 'historical laws') has often been doubted, I will now give a number of examples: 'You cannot introduce agricultural tariffs and at the same time reduce the cost of living' - 'You cannot, in an industrial society, organize consumers' pressure groups as effectively as you can organize certain producers' pressure groups.' - 'You cannot have a centrally planned society with a price system that fulfils the main functions of competitive prices.' - 'You cannot have full employment without inflation.' Another group of examples may be taken from the realm of power politics: 'You cannot introduce a political reform without causing some repercussions which are undesirable from the point of view of the ends aimed at' (therefore, look out for them). - 'You cannot introduce a political reform without strengthening the opposing forces, to a degree roughly in ratio to the scope of the reform.' (This may be said to be the technological corollary of 'There are always interests connected with the status quo.') - 'You cannot make a revolution without causing a reaction.' To these examples we may add two more, which may be called ' Plato's law of revolutions' (from the eighth book of the Republic) and 'Lord Acton's law of corruption', respectively: 'You cannot make a successful revolution if the ruling class is not weakened by internal dissension or defeat in war.'--'You cannot give a man power over other men without tempting him to misuse it--a temptation which roughly increases with the amount of power wielded, and which very few are capable of resisting.' 1 Nothing is here assumed about the strength of the available evidence in favour of these hypotheses whose formulations certainly leave much room for improvement. They are merely examples of the kind of statements which a piecemeal technology may attempt to discuss, and to substantiate.


I suggested in the foregoing section that the deductive methods there analyzed are widely used and important --more so than Mill, for example, ever thought. This suggestion will now be further elaborated, in order to throw some light on the dispute between naturalism and anti-naturalism. In this section I am going to propose a doctrine of the unity of method; that is to say, the view that all theoretical or generalizing sciences make use of the same method, whether they are natural sciences or social sciences. (I postpone the discussion of the historical sciences until section 31.) At the same time, some of these doctrines of historicism which I have not yet sufficiently examined will be touched upon, such as the problems of Generalization; of Essentialism; of the role played by Intuitive Understanding; of the Inexactitude of Prediction; of Complexity; and of the application of Quantitative Methods.

I do not intend to assert that there are no differences whatever between the methods of the theoretical sciences of nature and of society; such differences clearly exist, even between the various natural sciences themselves, as well as between the various social sciences. (Compare, for example, the analysis of competitive markets and of Romance languages.) But I agree with Comte and Mill--and with many others, such asC. Menger--that the methods in the two fields are fundamentally the same (though what I understand by them may not be what they had in mind). The methods always amount to deductive causal explanation, prediction, and testing, as sketched in the foregoing section. This has sometimes been called the hypothetico-deductive method, 1 or more often the method of hypothesis, for it does not achieve absolute certainty for any of the scientific statements which it tests; rather, these statements always retain the character of tentative hypotheses, even though their character of tentativeness may cease to be obvious after they have passed a great number of severe tests.

Because of their tentative or provisional character, hypotheses were considered, by most students of method, as provisional in the sense that they have ultimately to be replaced by proved theories (or at least by theories which can be proved to be 'highly probable', in the sense of some calculus of probabilities). I believe that this view is mistaken and that it leads to a host of entirely unnecessary difficulties. But this problem 2 is of comparatively little moment here. What is important is to realize that in science we are always concerned with explanations, predictions, and tests, and that the method of testing hypotheses is always the same (see the foregoing section). From the hypothesis to be tested--for example, a universal law--together with some other statements which for this purpose are not considered as problematic--for example, some initial conditions--we deduce some prognosis. We then confront this prognosis, whenever possible, with the results of experimental or other observations. Agreement with them is taken as corroboration of the hypothesis, though not as final proof; clear disagreement is considered as refutation or falsification.

According to this analysis, there is no great difference between explanation, prediction and testing. The difference is not one of logical structure, but rather one of emphasis; it depends on what we consider to be our problem and what we do not so consider. If it is not our problem to find a prognosis, while we take it to be our problem to find the initial conditions or some of the universal laws (or both) from which we may deduce a given 'prognosis', then we are looking for an explanation (and the given 'prognosis' becomes our 'explicandum'). If we consider the laws and initial conditions as given (rather than as to be found) and use them merely for deducing the prognosis, in order to get thereby some new information, then we are trying to make a prediction. (This is a case in which we apply our scientific results.) And if we consider one of the premises, i.e. either a universal law or an initial condition, as problematic, and the prognosis as something to be compared with the results of experience, then we speak of a test of the problematic premise.

The result of tests is the selection of hypotheses which have stood up to tests, or the elimination of those hypotheses which have not stood up to them, and which are therefore rejected. It is important to realize the consequences of this view. They are these: all tests can be interpreted as attempts to weed out false theories-to find the weak points of a theory in order to reject it if it is falsified by the test. This view is sometimes considered paradoxical; our aim, it is said, is to establish theories, not to eliminate false ones. But just because it is our aim to establish theories as well as we can, we must test them as severely as we can; that is, we must try to find fault with them, we must try to falsify them. Only if we cannot falsify them in spite of our best efforts can we say that they have stood up to severe tests. This is the reason why the discovery of instances which confirm a theory means very little if we have not tried, and failed, to discover refutations. For if we are uncritical we shall always find what we want: we shall look for, and find, confirmations, and we shall look away from, and not see, whatever might be dangerous to our pet theories. In this way it is only too easy to obtain what appears to be overwhelming evidence in favour of a theory which, if approached critically, would have been refuted. In order to make the method of selection by elimination work, and to ensure that only the fittest theories survive, their struggle for life must be made severe.

This, in outline, is the method of all sciences which are backed by experience. But what about the method by which we obtain our theories or hypotheses? What about inductive generalizations, and the way in which we proceed from observation to theory? To this question (and to the doctrines discussed in section 1, so far as they have not been dealt with in section 26) I shall give two answers. (a) I do not believe that we ever make inductive generalizations in the sense that we start with observations and try to derive our theories from them. I believe that the prejudice that we proceed in this way is a kind of optical illusion, and that at no stage of scientific development do we begin without something in the nature of a theory, such as a hypothesis, or a prejudice, or a problem--often a technological one--which in some way guides our observations, and helps us to select from the innumerable objects of observation those which may be of interest. But if this is so, then the method of elimination--which is nothing but that of trial and error discussed in section 24--can always be applied. However, I do not think that it is necessary for our present discussion to insist upon this point. For we can say (b) that it is irrelevant from the point of view of science whether we have obtained our theories by jumping to unwarranted conclusions or merely by stumbling over them (that is, by 'intuition') or by some inductive procedure. The question, 'How did you first find your theory?' relates, as it were, to an entirely private matter, as opposed to the question, 'How did you test your theory?' which alone is scientifically relevant. And the method of testing described here is fertile; it leads to new observations, and to a mutual give and take between theory and observation.

Now all this, I believe, is not only true for the natural but also for the social sciences. And in the social sciences it is even more obvious than in the natural sciences that we cannot see and observe our objects before we have thought about them. For most of the objects of social science, if not all of them, are abstract objects; they are theoretical constructions. (Even 'the war' or 'the army' are abstract concepts, strange as this may sound to some. What is concrete is the many who are killed; or the men and women in uniform, etc.) These objects, these theoretical constructions used to interpret our experience, are the result of constructing certain models (especially of institutions), in order to explain certain experiences-a familiar theoretical method in the natural sciences (where we construct our models of atoms, molecules, solids, liquids, etc.). It is part of the method of explanation by way of reduction, or deduction from hypotheses. Very often we are unaware of the fact that we are operating with hypotheses or theories, and we therefore mistake our theoretical models for concrete things. This is a kind of mistake which is only too common. 1 The fact that models are often used in this way explains--and by so doing destroys--the doctrines of methodological essentialism (cp. section 10). It explains them, for the model is abstract or theoretical in character, and so we are liable to feel that we see it, either within or behind the changing observable events, as a kind of permanent ghost or essence. And it destroys them because the task of social theory is to construct and to analyse our sociological models carefully in descriptive or nominalist terms, that is to say, in terms of individuals, of their attitudes, expectations, relations, etc.--a postulate which may be called 'methodological individualism'.

The unity of the methods of the natural and social sciences may be illustrated and defended by an analysis of two passages from Professor Hayek Scientism and the Study of Society. 2 In the first passage, Professor Hayek writes:

'The physicist who wishes to understand the problems of the social sciences with the help of an analogy from his own field would have to imagine a world in which he knew by direct observation the inside of the atoms and had neither the possibility of making experiments with lumps of matter nor the opportunity to observe more than the interactions of a comparatively few atoms during a limited period. From his knowledge of the different kinds of atoms he could build up models of all the various ways in which they could combine into larger units and make these models more and more closely reproduce all the features of the few instances in which he was able to observe more complex phenomena. But the laws of the macrocosm which he could derive from his knowledge of the microcosm would always remain "deductive"; they would, because of his limited knowledge of the data of the complex situation, scarcely ever enable him to predict the precise outcome of a particular situation; and he could never verify them by controlled experiment-although they might be disproved by the observation of events which according to his theory are impossible.'

I admit that the first sentence of this passage points to certain differences between social and physical science. But the rest of the passage, I believe, speaks for a complete unity of method. For if, as I do not doubt, this is a correct description of the method of social science, then it shows that it differs only from such interpretations of the method of natural science as we have already rejected. I have in mind, more especially, the 'inductivist' interpretation which holds that in the natural sciences we proceed systematically from observation to theory by some method of generalization, and that we can 'verify', or perhaps even prove, our theories by some method of induction. I have been advocating a very different view here--an interpretation of scientific method as deductive, hypothetical, selective by way of falsification, etc. And this description of the method of natural science agrees perfectly with Professor Hayek's description of the method of social science. (I have every reason to believe that my interpretation of the methods of science was not influenced by any knowledge of the methods of the social sciences; for when I developed it first, I had only the natural sciences in mind, 1 and I knew next to nothing about the social sciences.)

But even the differences alluded to in the first sentence of the quotation are not so great as may appear at first sight. It is undoubtedly true that we have a more direct knowledge of the 'inside of the human atom' than we have of physical atoms; but this knowledge is intuitive. In other words, we certainly use our knowledge of ourselves in order to frame hypotheses about some other people, or about all people. But these hypotheses must be tested, they must be submitted to the method of selection by elimination. (Intuition prevents some people from even imagining that anybody can possibly dislike chocolate.) The physicist, it is true, is not helped by such direct observation when he frames his hypotheses about atoms; nevertheless, he quite often uses some kind of sympathetic imagination or intuition which may easily make him feel that he is intimately acquainted with even the 'inside of the atoms'--with even their whims and prejudices. But this intuition is his private affair. Science is interested only in the hypotheses which his intuitions may have inspired, and then only if these are rich in consequences, and if they can be properly tested. (For the other difference mentioned in Professor Hayek's first sentence, i.e. the difficulty of conducting experiments, see section 24.)

These few remarks may also indicate the way in which the historicist doctrine expounded in section 8 should be criticized--that is to say, the doctrine that social science must use the method of intuitive understanding.

In the second passage, speaking of social phenomena,

Professor Hayek says: '. . . our knowledge of the principle by which these phenomena are produced will rarely if ever enable us to predict the precise result of any concrete situation. While we can explain the principle on which certain phenomena are produced and can from this knowledge exclude the possibility of certain results, e.g. of certain events occurring together, our knowledge will in a sense be only negative, i.e. it will merely enable us to preclude certain results but not enable us to narrow the range of possibilities sufficiently so that only one remains'.

This passage, far from describing a situation peculiar to the social sciences, perfectly describes the character of natural laws which, indeed, can never do more than exclude certain possibilities. ('You cannot carry water in a sieve'; see section 20, above.) More especially the statement that we shall not, as a rule, be able 'to predict the precise result of any concrete situation' opens up the problem of the inexactitude of prediction (see section 5, above). I contend that precisely the same may be said of the concrete physical world. In general it is only by the use of artificial experimental isolation that we can predict physical events. (The solar system is an exceptional case--one of natural, not of artificial isolation; once its isolation is destroyed by the intrusion of a foreign body of sufficient size, all our forecasts are liable to break down.) We are very far from being able to predict, even in physics, the precise results of a concrete situation, such as a thunderstorm, or a fire.

A very brief remark may be added here on the problem of complexity (see section 4, above). There is no doubt that the analysis of any concrete social situation is made extremely difficult by its complexity. But the same holds for any concrete physical situation. (A somewhat similar argument can be found in C. Menger, Collected Works, vol. II ( 1883 and 1933), pp. 259-60.)

The widely held prejudice that social situations are more complex than physical ones seems to arise from two sources. One of them is that we are liable to compare what should not be compared; I mean on the one hand concrete social situations and on the other hand artificially insulated experimental physical situations. (The latter might be compared, rather, with an artificially insulated social situation--such as a prison, or an experimental community.) The other source is the old belief that the description of a social situation should involve the mental and perhaps even physical states of everybody concerned (or perhaps that it should even be reducible to them). But this belief is not justified; it is much less justified even than the impossible demand that the description of a concrete chemical reaction should involve that of the atomic and sub-atomic states of all the elementary particles involved (although chemistry may indeed be reducible to physics). The belief also shows traces of the popular view that social entities such as institutions or associations are concrete natural entities such as crowds of men, rather than abstract models constructed to interpret certain selected abstract relations between individuals.

But in fact, there are good reasons, not only for the belief that social science is less complicated than physics, but also for the belief that concrete social situations are in general less complicated than concrete physical situations. For in most, if not in all, social situations there is an element of rationality. Admittedly, human beings hardly ever act quite rationally (i.e. as they would if they could make the optimal use of all available information for the attainment of whatever ends they may have), but they act, none the less, more or less rationally; and this makes it possible to construct comparatively simple models of their actions and inter-actions, and to use these models as approximations.

The last point seems to me, indeed, to indicate a considerable difference between the natural and the social sciences--perhaps the most important difference in their methods, since the other important differences, i.e. specific difficulties in conducting experiments (see end of section 24) and in applying quantitative methods (see below), are differences of degree rather than of kind. I refer to the possibility of adopting, in the social sciences, what may be called the method of logical or rational construction, or perhaps the 'zero method'. By this I mean the method of constructing a model on the assumption of complete rationality (and perhaps also on the assumption of the possession of complete information) on the part of all the individuals concerned, and of estimating the deviation of the actual behaviour of people from the model behaviour, using the latter as a kind of zero co-ordinate. 2 An example of this method is the comparison between actual behaviour (under the influence of, say, traditional prejudice, etc.) and model behaviour to be expected on the basis of the 'pure logic of choice', as described by the equations of economics. Marschak interesting "Money Illusion", for example, may be interpreted in this way. 3 An attempt at applying the zero method to a different field may be found in P. Sargant Florence's comparison between the 'logic of large-scale operation' in industry and the 'illogic of actual operation'.

In passing I should like to mention that neither the principle of methodological individualism, nor that of the zero method of constructing rational models, implies in my opinion the adoption of a psychological method. On the contrary, I believe that these principles can be combined with the view that the social sciences are comparatively independent of psychological assumptions, and that psychology can be treated, not as the basis of all social sciences, but as one social science among others.

In concluding this section, I have to mention what I consider to be the other main difference between the methods of some of the theoretical sciences of nature and of society. I mean the specific difficulties connected with the application of quantitative methods, and especially methods of measurement. 3 Some of these difficulties can be, and have been, overcome by the application of statistical methods, for example in demand analysis. And they have to be overcome if, for example, some of the equations of mathematical economics are to provide a basis even of merely qualitative applications; for without such measurement we should often not know whether or not some counteracting influences exceeded an effect calculated in merely qualitative terms. Thus merely qualitative considerations may well be deceptive at times; just as deceptive, to quote Professor Frisch, 'as to say that when a man tries to row a boat forward, the boat will be driven backward because of the pressure exerted by his feet'. But it cannot be doubted that there are some fundamental difficulties here. In physics, for example, the parameters of our equations can, in principle, be reduced to a small number of natural constants--a reduction which has been successfully carried out in many important cases. This is not so in economics; here our parameters are themselves in the most important cases quickly changing variables. This clearly reduces the significance, interpretability, and testability of measurements.


The thesis of the unity of scientific method, whose application to theoretical sciences I have just been defending, can be extended, with certain limitations, even to the field of the historical sciences. And this can be done without giving up the fundamental distinction between theoretical and historical sciences--for example, between sociology or economic theory or political theory on the one hand, and social, economic, and political history on the other--a distinction which has been so often and emphatically reaffirmed by the best historians. It is the distinction between the interest in universal laws and the interest in particular facts. I wish to defend the view, so often attacked as oldfashioned by historicists, that history is characterized by its interest in actual, singular, or specific events, rather than in laws or generalizations.

This view is perfectly compatible with the analysis of scientific method, and especially of causal explanation, given in the preceding sections. The situation is simply this: while the theoretical sciences are mainly interested in finding and testing universal laws, the historical sciences take all kinds of universal laws for granted and are mainly interested in finding and testing singular statements. For example, given a certain singular 'explicandum'--a singular event--they may look for singular initial conditions which (together with all kinds of universal laws which may be of little interest) explain that explicandum. Or they may test a given singular hypothesis by using it, along with other singular statements, as an initial condition, and by deducing from these initial conditions (again with the help of all kinds of universal laws of little interest) some new 'prognosis' which may describe an event which has happened in the distant past, and which can be confronted with the empirical evidence--perhaps with documents or inscriptions, etc.

In the sense of this analysis, all causal explanation of a singular event can be said to be historical in so far as the 'cause' is always described by singular initial conditions. And this agrees entirely with the popular idea that to explain a thing causally is to explain how and why it happened, that is to say, to tell its 'story'. But it is only in history that we are really interested in the causal explanation of a singular event. In the theoretical sciences, such causal explanations are mainly means to a different end--the testing of universal laws.

If these considerations are correct, then the burning interest in questions of origin shown by some evolutionists and historicists, who despise old-fashioned history and wish to reform it into a theoretical science, is somewhat misplaced. Questions of origin are 'how and why' questions. They are comparatively unimportant theoretically and usually have only a specific historical interest.

Against my analysis of historical explanation it may be argued that history does make use of universal laws contrary to the emphatic declaration of so many historians that history has no interest whatever in such laws. To this we may answer that a singular event is the cause of another singular event--which is its effect --only relative to some universal laws. (This has been seen by Max Weber. His remarks on p. 179 of his Ges. Schr. zur Wissenschaftslehre ( 1922) are the closest anticipation of which I know to the analysis offered here. But he is mistaken, I believe, when he suggests that the difference between theoretical and historical science lies in the degree of generality of the laws used.) But these laws may be so trivial, so much part of our common knowledge, that we need not mention them and rarely notice them. If we say that the cause of the death of Giordano Bruno was being burnt at the stake, we do not need to mention the universal law that all living things die when exposed to intense heat. But such a law was tacitly assumed in our causal explanation.

Among the theories which the political historian presupposes are, of course, certain theories of sociology --the sociology of power, for example. But the historian uses even these theories, as a rule, without being aware of them. He uses them in the main not as universal laws which help him to test his specific hypotheses but as implicit in his terminology. In speaking of governments, nations, armies, he uses, usually unconsciously, the 'models' provided by scientific or pre-scientific sociological analysis (see the foregoing section).

The historical sciences, it may be remarked, do not stand quite apart in their attitude towards universal laws. Whenever we encounter an actual application of science to a singular or specific problem we find a similar situation. The practical chemist, for example, who wishes to analyse a certain given compound--a piece of rock, say--hardly considers any universal law. Instead, he applies, possibly without much thought, certain routine techniques which, from the logical point of view, are tests of such singular hypotheses as 'this compound contains sulphur'. His interest is mainly a historical one--the description of one set of specific events, or of one individual physical body.

I believe that this analysis clarifies some well-known controversies between certain students of the method of history.1 One historicist group asserts that history, which does not merely enumerate facts but attempts to present them in some kind of causal connection, must be interested in the formulation of historical laws, since causality means, fundamentally, determination by law. Another group, which also includes historicists, argues that even 'unique' events, events which occur only once and have nothing 'general' about them, may be the cause of other events, and that it is this kind of causation that history is interested in. We can now see that both groups are partly right and partly wrong. Universal law and specific events are together necessary for any causal explanation, but outside the theoretical sciences, universal laws usually arouse little interest.

This leads us to the question of the uniqueness of historical events. In so far as we are concerned with the historical explanation of typical events they must necessarily be treated as typical, as belonging to kinds or classes of events. For only then is the deductive method of causal explanation applicable. History, however, is interested not only in the explanation of specific events but also in the description of a specific event as such. One of its most important tasks is undoubtedly to describe interesting happenings in their peculiarity or uniqueness; that is to say, to include aspects which it does not attempt to explain causally, such as the 'accidental' concurrence of causally unrelated events. These two tasks of history, the disentanglement of causal threads and the description of the 'accidental' manner in which these threads are interwoven, are both necessary, and they supplement each other; at one time an event may be considered as typical, i.e. from the standpoint of its causal explanation, and at another time as unique.

These considerations may be applied to the question of novelty, discussed in section 3. The distinction made there between 'novelty of arrangement' and 'intrinsic newness' corresponds to the present distinction between the standpoint of causal explanation and that of the appreciation of the unique. So far as newness can be rationally analyzed and predicted, it can never be 'intrinsic'. This dispels the historicist doctrine that social science should be applicable to the problem of predicting the emergence of intrinsically new events-a claim which may be said to rest ultimately on an insufficient analysis of prediction and of causal explanation.


But is this all? Is there nothing whatever in the historicist demand for a reform of history--for a sociology which plays the role of a theoretical history, or a theory of historical development? (See sections 12 and 16.) Is there nothing whatever in the historicist idea of 'periods'; of the 'spirit' or 'style' of an age; of irresistible historical tendencies; of movements which captivate the minds of individuals and which surge on like a flood, driving, rather than being driven by, individual men? Nobody who has read, for example, the speculations of Tolstoy in War and Peace--historicist, no doubt, but stating his motives with candour-on the movement of the men of the West towards the East and the counter movement of the Russians towards the West, 1 can deny that historicism answers a real need. We have to satisfy this need by offering something better before we can seriously hope to get rid of historicism.

Tolstoy's historicism is a reaction against a method of writing history which implicitly accepts the truth of the principle of leadership; a method which attributes much--too much, if Tolstoy is right, as he undoubtedly is--to the great man, the leader. Tolstoy tries to show, successfully I think, the small influence of the actions and decisions of Napoleon, Alexander, Kutuzov, and the other great leaders of 1812, in the face of what may be called the logic of events. Tolstoy points out, rightly, the neglected but very great importance of the decisions and actions of the countless unknown individuals who fought the battles, who burned Moscow, and who invented the partisan method of fighting. But he believes that he can see some kind of historical determination in these events-fate, historical laws, or a plan. In his version of historicism, he combines both methodological individualism and collectivism; that is to say, he represents a highly typical combination--typical of his time, and, I am afraid, of our own--of democratic-individualist and collectivist-nationalistic elements.

This example may remind us that there are some sound elements in historicism; it is a reaction against the naïve method of interpreting political history merely as the story of great tyrants and great generals.

Historicists rightly feel that there may be something better than this method. It is this feeling which makes their idea of 'spirits'--of an age, of a nation, of an army--so seductive.

Now I have not the slightest sympathy with these 'spirits'; neither with their idealistic prototype nor with their dialectical and materialistic incarnations, and I am in full sympathy with those who treat them with contempt. And yet I feel that they indicate, at least, the existence of a vacuum, of a place which it is the task of sociology to fill with something more sensible, such as an analysis of problems arising within a tradition. There is room for a more detailed analysis of the logic of situations. The best historians have often made use, more or less unconsciously, of this conception: Tolstoy, for example, when he describes how it was not decision but 'necessity' which made the Russian army yield Moscow without a fight and withdraw to places where it could find food. Beyond this logic of situations, or perhaps as a part of it, we need something like an analysis of social movements. We need studies, based on methodological individualism, of the social institutions through which ideas may spread and captivate individuals, of the way in which new traditions may be created, and of the way in which traditions work and break down. In other words, our individualistic and institutionalist models of such collective entities as nations, or governments, or markets, will have to be supplemented by models of political situations as well as of social movements such as scientific and industrial progress. (A sketch of such an analysis of progress will be found in the next section.) These models may then be used by historians, partly like the other models, and partly for the purpose of explanation, along with the other universal laws they use. But even this would not be enough; it would still not satisfy all those real needs which historicism attempts to satisfy.

If we consider the historical sciences in the light of our comparison between them and the theoretical sciences, then we can see that their lack of interest in universal laws puts them in a difficult position. For in theoretical science laws act, among other things, as centres of interest to which observations are related, or as points of view from which observations are made. In history the universal laws, which for the most part are trivial and used unconsciously, cannot possibly fulfil this function. It must be taken over by something else. For undoubtedly there can be no history without a point of view; like the natural sciences, history must be selective unless it is to be choked by a flood of poor and unrelated material. The attempt to follow causal chains into the remote past would not help in the least, for every concrete effect with which we might start has a great number of different partial causes; that is to say, initial conditions are very complex, and most of them have little interest for us.

The only way out of this difficulty is, I believe, consciously to introduce a preconceived selective point of view into one's history; that is, to write that history which interests us. This does not mean that we may twist the facts until they fit into a framework of preconceived ideas, or that we may neglect the facts that do not fit. On the contrary, all available evidence which has a bearing on our point of view should be considered carefully and objectively (in the sense of 'scientific objectivity', to be discussed in the next section). But it means that we need not worry about all those facts and aspects which have no bearing upon our point of view and which therefore do not interest us.

Such selective approaches fulfil functions in the study of history which are in some ways analogous to those of theories in science. It is therefore understandable that they have often been taken for theories. And indeed, those rare ideas inherent in these approaches which can be formulated in the form of testable hypotheses, whether singular or universal, may well be treated as scientific hypotheses. But as a rule, these historical 'approaches' or 'points of view' cannot be tested. They cannot be refuted, and apparent confirmations are therefore of no value, even if they are as numerous as the stars in the sky. We shall call such a selective point of view or focus of historical interest, if it cannot be formulated as a testable hypothesis, a historical interpretation.

Historicism mistakes these interpretations for theories. This is one of its cardinal errors. It is possible, for example, to interpret 'history' as the history of class struggle, or of the struggle of races for supremacy, or as the history of religious ideas, or as the history of the struggle between the 'open' and the 'closed' society, or as the history of scientific and industrial progress. All these are more or less interesting points of view, and as such perfectly unobjectionable. But historicists do not present them as such; they do not see that there is necessarily a plurality of interpretations which are fundamentally on the same level of both, suggestiveness and arbitrariness (even though some of them may be distinguished by their fertiliy--a point of some importance). Instead, they present them as doctrines or theories, asserting that 'all history is the history of class struggle', etc. And if they actually find that their point of view is fertile, and that many facts can be ordered and interpreted in its light, then they mistake this for a confirmation, or even for a proof, of their doctrine.

On the other hand, the classical historians who rightly oppose this procedure are liable to fall into a different error. Aiming at objectivity, they feel bound to avoid any selective point of view; but since this is impossible, they usually adopt points of view without being aware of them. This must defeat their efforts to be objective, for one cannot possibly be critical of one's own approach, and conscious of its limitations, without being aware of it.

The way out of this dilemma, of course, is to be clear about the necessity of adopting a point of view; to state this point of view plainly, and always to remain conscious that it is one among many, and that even if it should amount to a theory, it may not be testable.


In order to make our considerations less abstract, I shall try in this section to sketch, in very brief outline, a theory of scientific and industrial progress. I shall try to exemplify, in this way, the ideas developed in the last four sections; more especially the idea of situational logic, and of a methodical individualism which keeps clear of psychology. I choose the example of scientific and industrial progress because undoubtedly it was this phenomenon which inspired modern nineteenthcentury historicism, and because I have previously discussed some of Mill's views on this subject.

Comte and Mill, it will be remembered, held that progress was an unconditional or absolute trend, which is reducible to the laws of human nature. 'A law of succession,' writes Comte, 'even when indicated with all possible authority by the method of historical observation, ought not to be finally admitted before it has been rationally reduced to the positive theory of human nature . . .'  He believes that the law of progress is deducible from a tendency in human individuals which impels them to perfect their nature more and more. In all this, Mill follows him completely, trying to reduce his law of progress to what he calls the 'progressiveness of the human mind'  whose first 'impelling force . . . is the desire of increased material comforts'. According to both Comte and Mill the unconditional or absolute character of this trend or quasi-law enables us to deduce from it the first steps or phases of history, without requiring any initial historical conditions or observations or data.  In principle, the whole course of history should be thus deducible; the only difficulty being, as Mill puts it, that 'so long a series . . ., each successive term being composed of an even greater number and variety of parts, could not possibly be computed by human faculties.' 

The weakness of this 'reduction' of Mill's seems obvious. Even if we should grant Mill's premises and deductions, it still would not follow that the social or historical effect will be significant. Progress might be rendered negligible, say, by losses due to an unmanageable natural environment. Besides, the premises are based on only one side of 'human nature' without considering other sides such as forgetfulness or indolence. Thus where we observe the precise opposite of the progress described by Mill, there we can equally well 'reduce' these observations to 'human nature'. (Is it not, indeed, one of the most popular devices of so-called historical theories to explain the decline and fall of empires by such traits as idleness and a propensity to over-eat?) In fact we can conceive of very few events which could not be plausibly explained by an appeal to certain propensities of 'human nature'. But a method that can explain everything that might happen explains nothing.

If we wish to replace this surprisingly naïve theory by a more tenable one, we have to make two main alterations. First, we have to attempt to find conditions of progress, and to this end we must try to imagine, for example, conditions under which progress would be arrested. This immediately leads to the realization that a psychological propensity alone cannot be sufficient to explain progress, since conditions may be found on which it may depend. Thus we must, next, replace the theory of psychological propensities by something better; I suggest, by an institutional (and technological) analysis of the conditions of progress.

How could we arrest scientific and industrial progress? By closing down or controlling laboratories for research, by suppressing or controlling scientific periodicals and other means of discussion, by suppressing scientific congresses and conferences, by suppressing Universities and other schools, by suppressing books, the printing press, writing, and, in the end, speaking. All these things which indeed might be suppressed (or controlled) are social institutions. Language is a social institution without which scientific progress is unthinkable, since without it there can be neither science nor a growing and progressive tradition. Writing is a social institution, and so are the organizations for printing and publishing and all the other institutional instruments of scientific method. Scientific method itself has social aspects. Science, and more especially scientific progress, are the results not of isolated efforts but of the free competition of thought. For science needs ever more competition between hypotheses and ever more rigorous tests. And the competing hypotheses need personal representation, as it were: they need advocates, they need a jury, and even a public. This personal representation must be institutionally organized if we wish to ensure that it works. And these institutions have to be paid for, and protected by law. Ultimately, progress depends very largely on political factors; on political institutions that safeguard the freedom of thought: on democracy.

It is of some interest that what is usually called 'scientific objectivity' is based, to some extent, on social institutions. The naïve view that scientific objectivity rests on the mental or psychological attitude of the individual scientist, on his training, care, and scientific detachment, generates as a reaction the sceptical view that scientists can never be objective. On this view their lack of objectivity may be negligible in the natural sciences where their passions are not excited, but for the social sciences where social prejudices, class bias, and personal interests are involved, it may be fatal. This doctrine, developed in detail by the so-called "Sociology of Knowledge" (see sections 6 and 26) entirely overlooks the social or institutional character of scientific knowledge, because it is based on the naïve view that objectivity depends on the psychology of the individual scientist. It overlooks the fact that neither the dryness nor the remoteness of a topic of natural science prevent partiality and self-interest from interfering with the individual scientist's beliefs, and that if we had to depend on his detachment, science, even natural science, would be quite impossible. What the 'sociology of knowledge' overlooks is just the sociology of knowledge--the social or public character of science: it overlooks the fact that the inter-subjectivity of science and its institutions for the dissemination and discussion of new ideas, are the safeguards of scientific objectivity. These also impose a mental discipline upon the individual scientist.

In this connection, I may perhaps touch upon another of the doctrines presented in section 6 (Objectivity and Valuation). There it was argued that, since scientific research in social problems must itself influence social life, it is impossible for the social scientist who is aware of this influence to retain the proper scientific attitude of disinterested objectivity. But there is nothing peculiar to social science in this situation. A physicist or a physical engineer is in the same position. Without being a social scientist he can realize that the invention of a new aircraft may have a tremendous influence on society.

I have just sketched some of the institutional conditions on whose realization scientific and industrial progress depends. Now it is important to realize that most of these conditions cannot be called necessary, and that all of them taken together are not sufficient.

The conditions are not necessary, since without these institutions (language perhaps excepted) scientific progress would not be strictly impossible. 'Progress', after all, has been made from the spoken to the written word, and even further (although this early development was perhaps not, properly speaking, scientific progress).

On the other hand, and this is more important, we must realize that with the best institutional organization in the world, scientific progress may one day stop. There may, for example, be an epidemic of mysticism. This is certainly possible, for since some intellectuals do react to scientific progress (or to the demands of an open society) by withdrawing into mysticism, everyone might react in this way. Such a possibility may perhaps be counteracted by devising a further set of social institutions, such as educational institutions, to discourage uniformity of outlook and encourage diversity. Also, the idea of progress and its enthusiastic propagation may have some effect. But all this cannot make progress certain. For we cannot exclude the logical possibility, say, of a bacterium or virus that spreads a wish for Nirvana.

We thus find that even the best institutions can never be foolproof. As I have said before, 'Institutions are like fortresses. They must be well designed and properly manned'. But we can never make sure that the right man will be attracted by scientific research. Nor can we make sure that there will be men ofimagination who have the knack of inventing new hypotheses. And ultimately, much depends on sheer luck, in these matters. For truth is not manifest, and it is a mistake to believe-as did Comte and Mill--that once the 'obstacles' (the allusion is to the Church) are removed, truth will be visible to all who genuinely want to see it.

I believe that the result of this analysis can be generalized. The human or personal factor will remain the irrational element in most, or all, institutional social theories. The opposite doctrine which teaches the reduction of social theories to psychology, in the same way as we try to reduce chemistry to physics, is, I believe, based on a misunderstanding. It arises from the false belief that this 'methodological psychologism' is a necessary corollary of a methodological individualism--of the quite unassailable doctrine that we must try to understand all collective phenomena as due to the actions, interactions, aims, hopes, and thoughts of individual men, and as due to traditions created and preserved by individual men. But we can be individualists without accepting psychologism. The 'zero method' of constructing rational models is not a psychological but rather a logical method.

In fact, psychology cannot be the basis of social science. First, because it is itself just one of the social sciences: 'human nature' varies considerably with the social institutions, and its study therefore presupposes an understanding of these institutions. Secondly, because the social sciences are largely concerned with the unintended consequences, or repercussions, of human actions. And 'unintended' in this context does not perhaps mean 'not consciously intended'; rather it characterizes repercussions which may violate all interests of the actor, whether conscious or unconscious: although some people may claim that a liking for mountains and solitude may be explained psychologically, the fact that if many people like the mountains, they cannot enjoy solitude, is not a psychological fact; but this kind of problem is at the very root of social theory.

With this, we reach a result which contrasts startlingly with the still fashionable method of Comte and Mill. Instead of reducing sociological considerations to the apparently firm basis of the psychology of human nature, we might say that the human factor is the ultimately uncertain and wayward element in social life and in all social institutions. Indeed this is the element which ultimately cannot be completely controlled by institutions (as Spinoza first saw); for every attempt at controlling it completely must lead to tyranny; which means, to the omnipotence of the human factor--the whims of a few men, or even of one.

But is it not possible to control the human factor by science--the opposite of whim? No doubt, biology and psychology can solve, or will soon be able to solve, the 'problem of transforming man'. Yet those who attempt to do this are bound to destroy the objectivity of science, and so science itself, since these are both based upon free competition of thought; that is, upon freedom. If the growth of reason is to continue, and human rationality to survive, then the diversity of individuals and their opinions, aims, and purposes must never be interfered with (except in extreme cases where political freedom is endangered). Even the emotionally satisfying appeal for a common purpose, however excellent, is an appeal to abandon all rival moral opinions and the cross-criticisms and arguments to which they give rise. It is an appeal to abandon rational thought.

The evolutionist who demands the 'scientific' control of human nature does not realize how suicidal this demand is. The mainspring of evolution and progress is the variety of the material which may become subject to selection. So far as human evolution is concerned it is the 'freedom to be odd and unlike one's neighbour'--'to disagree with the majority, and go one's own way'. Holistic control, which must lead to the equalization not of human rights but of human minds, would mean the end of progress.

Index - Home PageFull Index of articles on the site