Excerpts from Must Read Books & Articles on Mental Health Topics
Articles- Part XX

The Nature of the Therapeutic Action of Psycho-Analysis
James Strachey (1934)

It was as a therapeutic procedure that psycho-analysis originated. It is in the main as a therapeutic agency that it exists to-day. We may well be surprised, therefore, at the relatively small proportion of psycho-analytical literature which has been concerned with the mechanisms by which its therapeutic effects are achieved. A very considerable quantity of data have been accumulated in the course of the last thirty or forty years which throw light upon the nature and workings of the human mind; perceptible progress has been made in the task of classifying and subsuming such data into a body of generalized hypotheses or scientific laws. But there has been a remarkable hesitation in applying these findings in any great detail to the therapeutic process itself. I cannot help feeling that this hesitation has been responsible for the fact that so many discussions upon the practical details of analytic technique seem to leave us at cross-purposes and at an inconclusive end. How, for instance, can we expect to agree upon the vexed question of whether and when we should give a `deep interpretation', while we have no clear idea of what we mean by a `deep interpretation', while, indeed, we have no exactly formulated view of the concept of `interpretation' itself, no precise knowledge of what `interpretation' is and what effect it has upon our patients? We should gain much, I think, from a clearer grasp of problems such as this. If we could arrive at a more detailed understanding of the workings of the therapeutic process we should be less prone to those occasional feelings of utter disorientation which few analysts are fortunate enough to escape; and the analytic movement itself might be less at the mercy of proposals for abrupt alterations in the ordinary technical procedure-proposals which derive much of their strength from the prevailing uncertainty as to the exact nature of the analytic therapy. My present paper is a tentative attack upon this problem; and even though it should turn out that its very doubtful conclusions cannot be maintained, I shall be satisfied if I have drawn attention to the urgency of the problem itself. I am most anxious, however, to make it clear that what follows is not a practical discussion upon psychoanalytic technique. Its immediate bearings are merely theoretical. I have taken as my raw material the various sorts of procedures which (in spite of very considerable individual deviations) would be generally regarded as within the limits of `orthodox' psycho-analysis and the various sorts of effects which observation shows that the application of such procedures tends to bring about; I have set up a hypothesis which endeavours to explain more or less coherently why these particular procedures bring about these particular effects; and I have tried to show that, if my hypothesis about the nature of the therapeutic action of psycho-analysis is valid, certain implications follow from it which might perhaps serve as criteria in forming a judgment of the probable effectiveness of any particular type of procedure.

It will be objected, no doubt, that I have exaggerated the novelty of my topic. (1) `After all', it will be said, `we (to understand and have long understood the main principles that govern the therapeutic action of analysis'. And to this, of course, I entirely agree; indeed I propose to begin what I have to say by summarizing as shortly as possible the accepted views upon the subject. For this purpose I must go back to the period between the years 1912 and 1917 during which Freud gave us the greater part of what he has written directly on the therapeutic side of psycho-analysis, namely the series of papers on technique (2) and the twenty-seventh and twenty-eighth chapters of the Introductory Lectures.

This period was characterized by the systematic application of the method known as `resistance analysis'. The method in question was by no means a new one even at that time, and it was based upon ideas which had long been implicit in analytical theory, and in particular upon one of the earliest of Freud's views of the function of neurotic symptoms. According to that view (which was derived essentially from the study of hysteria) the function of the neurotic symptom was to defend the patient's personality against an unconscious trend of thought that was unacceptable to it, while at the same time gratifying the trend up to a certain point. It seemed to follow, therefore, that if the analyst were to investigate and discover the unconscious trend and make the patient aware of it--if he were to make what was unconscious conscious--the whole raison d'etre of the symptom would cease and it must automatically disappear. Two difficulties arose, however. In the first place some part of the patient's mind was found to raise obstacles to the process, to offer resistance to the analyst when he tried to discover the unconscious trend; and it was easy to conclude that this was the same part of the patient's mind as had originally repudiated the unconscious trend and had thus necessitated the creation of the symptom. But, in the second place, even when this obstacle seemed to be surmounted, even when the analyst had succeeded in guessing or deducing the nature of the unconscious trend, had drawn the patient's attention to it and had apparently made him fully aware of it--even then it would often happen that the symptom persisted unshaken. The realization of these difficulties led to important results both theoretically and practically. Theoretically, it became evident that there were two senses in which a patient could become conscious of an unconscious trend; he could be made aware of it by the analyst in some intellectual sense without becoming `really' conscious of it. To make this state of things more intelligible, Freud devised a kind of pictorial allegory. He imagined the mind as a kind of map. The original objectionable trend was pictured as being located in one region of this map and the newly discovered information about it, communicated to the patient by the analyst, in another. It was only if these two impressions could be `brought together' (whatever exactly that might mean) that the unconscious trend would be `really' made conscious. What prevented this from happening was a force within the patient, a barrier once again, evidently, the same `resistance' which had opposed the analyst's attempts at investigating the unconscious trend and which had contributed to the original production of the symptom. The removal of this resistance was the essential preliminary to the patient's becoming `really' conscious of the unconscious trend. And it was at this point that the practical lesson emerged: as analysts our main task is not so much to investigate the objectionable unconscious trend as to get rid of the patient's resistance to it.
     But how are we to set about this task of demolishing the resistance? Once again by the same process of investigation and explanation which we have already applied to the unconscious trend. But this time we are not faced by such difficulties as before, for the forces that are keeping up the repression, although they are to some extent unconscious, do not belong to the unconscious in the systematic sense; they are a part of the patient's ego, which is co-operating with us, and are thus more accessible. Nevertheless the existing state of equilibrium will not be upset, the ego will not be induced to do the work of re-adjustment that is required of it, unless we are able by our analytic procedure to mobilize some fresh force upon our side.
     What forces can we count upon? The patient's will to recovery, in the first place, which led him to embark upon the analysis. And, again, a number of intellectual considerations which we can bring to his notice. We can make him understand the structure of his symptom and the motives for his repudiation of the objectionable trend. We can point out the fact that these motives are out-of-date and no longer valid; that they may have been reasonable when he was a baby, but are no longer so now that he is grown up. And finally we can insist that his original solution of the difficulty has only led to illness, while the new one that we propose holds out a prospect of health. Such motives as these may play a part in inducing the patient to abandon his resistances; nevertheless it is from an entirely different quarter that the decisive factor emerges. This factor, I need hardly say, is the transference. And I must now recall, very briefly, the main ideas held by Freud on that subject during the period with which I am dealing.

I should like to remark first that, although from very early times Freud had called attention to the fact that transference manifested itself in two ways--negatively as well as positively, a good deal less was said or known about the negative transference than about the positive. This of course corresponds to the circumstance that interest in the destructive and aggressive impulses in general is only a comparatively recent development. Transference was regarded predominantly as a libidinal phenomenon. It was suggested that in everyone there existed a certain number of unsatisfied libidinal impulses, and that whenever some new person came upon the scene these impulses were ready to attach themselves to him. This was the account of transference as a universal phenomenon. In neurotics, owing to the abnormally large quantities of unattached libido present in them, the tendency to transference would be correspondingly greater; and the peculiar circumstances of the analytic situation would further increase it. It was evidently the existence of these feelings of love, thrown by the patient upon the analyst, that provided the necessary extra force to induce his ego to give up its resistances, undo the repressions and adopt a fresh solution of its ancient problems. This instrument, without which no therapeutic result could be obtained, was at once seen to be no stranger; it was in fact the familiar power of suggestion, which had ostensibly been abandoned long before. Now, however, it was being employed in a very different way, in fact in a contrary direction. In pre-analytic days it had aimed at bringing about an increase in the degree of repression; now it was used to overcome the resistance of the ego, that is to say, to allow the repression to be removed.
     But the situation became more and more complicated as more facts about transference came to light. In the first place, the feelings transferred turned out to be of various sorts; besides the loving ones there were the hostile ones, which were naturally far from assisting the analyst's efforts. But, even apart from the hostile transference, the libidinal feelings themselves fell into two groups: friendly and affectionate feelings which were capable of being conscious, and purely erotic ones which had usually to remain unconscious. And these latter feelings, when they became too powerful, stirred up the repressive forces of the ego and thus increased its resistances instead of diminishing them, and in fact produced a state of things that was not easily distinguishable from a negative transference. And beyond all this there arose the whole question of the lack of permanence of all suggestive treatments. Did not the existence of the transference threaten to leave the analytic patient in the same unending dependence upon the analyst?
     All of these difficulties were got over by the discovery that the transference itself could be analysed. Its analysis, indeed, was soon found to be the most important part of the whole treatment. It was possible to make conscious its roots in the repressed unconscious just as it was possible to make conscious any other repressed material--that is, by inducing the ego to abandon its resistances--and there was nothing self-contradictory in the fact that the force used for resolving the transference was the transference itself. And once it had been made conscious, its unmanageable, infantile, permanent characteristics disappeared; what was left was like any other `real' human relationship. But the necessity for constantly analysing the transference became still more apparent from another discovery. It was found that as work proceeded the transference tended, as it were, to eat up the entire analysis. More and more of the patient's libido became concentrated upon his relation to the analyst, the patient's original symptoms were drained of their cathexis, and there appeared instead an artificial neurosis to which Freud gave the name of the `transference neurosis'. The original conflicts, which had led to the onset of neurosis, began to be re-enacted in the relation to the analyst. Now this unexpected event is far from being the misfortune that at first sight it might seem to be. In fact it gives us our great opportunity. Instead of having to deal as best we may with conflicts of the remote past, which are concerned with dead circumstances and mummified personalities, and whose outcome is already determined, we find ourselves involved in an actual and immediate situation, in which we and the patient are the principal characters and the development of which is to some extent at least under our control. But if we bring it about that in this revivified transference conflict the patient chooses a new solution instead of the old one, a solution in which the primitive and unadaptable method of repression is replaced by behaviour more in contact with reality, then, even after his detachment from the analysis, he will never be able to fall back into his former neurosis. The solution of the transference conflict implies the simultaneous solution of the infantile conflict of which it is a new edition. `The change', says Freud in his Introductory Lectures, `is made possible by alterations in the ego occurring as a consequence of the analyst's suggestions. At the expense of the unconscious, the ego becomes wider by the work of interpretation which brings the unconscious material into consciousness; through education it becomes reconciled to the libido and is made willing to grant it a certain degree of satisfaction; and its horror of the claims of its libido is lessened by the new capacity it acquires to expend a certain amount of the libido in sublimation. The more nearly the course of the treatment corresponds with this ideal description the greater will be the success of the psycho-analytic therapy'. (3)I quote these words of Freud's to make it quite clear that at the time he wrote them he held that the ultimate factor in the therapeutic action of psychoanalysis was suggestion on the part of the analyst acting upon the patient's ego in such a way as to make it more tolerant of the libidinal trends.

In the years that have passed since he wrote this passage Freud has produced extremely little that bears directly on the subject; and that little goes to show that he has not altered his views of the main principles involved. Indeed, in the additional lectures which were published last year, he explicitly states that he has nothing to add to the theoretical discussion upon therapy given in the original lectures fifteen years earlier. (4) At the same time there has in the interval been a considerable further development of his theoretical opinions, and especially in the region of ego-psychology. He has, in particular, formulated the concept of the super-ego. The re-statement in super-ego terms of the principles of therapeutics which he laid down in the period of resistance analysis may not involve many changes. But it is reasonable to expect that information about the super-ego will be of special interest from our point of view; and in two ways. In the first place, it would at first sight seem highly probable that the super-ego should play an important part, direct or indirect, in the setting-up and maintaining of the repressions and resistances the demolition of which has been the chief aim of analysis. And this is confirmed by an examination of the classification of the various kinds of resistance made by Freud in Hemmung Symptom and Angst (1926). (5) Of the five sorts of resistance there mentioned it is true that only one is attributed to the direct intervention of the super-ego, but two of the ego-resistances--the repressionresistance and the transference-resistance--although actually originating from the ego, are as a rule set up by it out of fear of the super-ego. It seems likely enough therefore that when Freud wrote the words which I have just quoted, to the effect that the favourable change in the patient 'is made possible by alterations in the ego' he was thinking, in part at all events, of that portion of the ego which he subsequently separated off into the super-ego. Quite apart from this, moreover, in another of Freud's more recent works, the Group Psychology (1921), there are passages which suggest a different point--namely, that it may be largely through the patient's super-ego that the analyst is able to influence him. These passages occur in the course of his discussion on the nature of hypnosis and suggestion. (6) He definitely rejects Bernheim's view that all hypnotic phenomena are traceable to the factor of suggestion, and adopts the alternative theory that suggestion is a partial manifestation of the state of hypnosis. The state of hypnosis, again, is found in certain respects to resemble the state of being in love. There is 'the same humble subjection, the same compliance, the same absence of criticism towards the hypnotist as towards the loved object'; in particular, there can be no doubt that the hypnotist, like the loved object, 'has stepped into the place of the subject's ego-ideal'. Now since suggestion is a partial form of hypnosis and since the analyst brings about his changes in the patient's attitude by means of suggestion, it seems to follow that the analyst owes his effectiveness, at all events in some respects, to his having stepped into the place of the patient's super-ego. Thus there are two convergent lines of argument which point to the patient's super-ego as occupying a key position in analytic therapy: it is a part of the patient's mind in which a favourable alteration would be likely to lead to general improvement, and it is a part of the patient's mind which is especially subject to the analyst's influence.
     Such plausible notions as these were followed up almost immediately after the super-ego made its first debut. (7) They were developed by Ernest Jones, for instance, in his paper on `The Nature of Auto-Suggestion'. (8) Soon afterwards (9) Alexander launched his theory that the principal aim of all psycho-analytic therapy must be the complete demolition of the super-ego and the assumption of its functions by the ego. According to his account, the treatment falls into two phases. In the first phase the functions of the patient's super-ego are handed over to the analyst, and in the second phase they are passed back again to the patient, but this time to his ego. The super-ego, according to this view of Alexander's (though he explicitly limits his use of the word to the unconscious parts of the ego-ideal), is a portion of the fundamental apparatus which is essentially primitive, out of date and out of touch with reality, which is incapable of adapting itself, and which operates automatically, with the monotonous uniformity of a reflex. Any useful functions that it performs can be carried out by the ego, and there is therefore nothing to be done with it but to scrap it. This wholesale attack upon the super-ego seems to be of questionable validity. [t seems probable that its abolition, even if that were practical politics, would involve the abolition of a large number of highly desirable mental activities. But the idea that the analyst temporarily takes over the functions of the patient's super-ego during the treatment and by so doing in some way alters it agrees with the tentative remarks which I have already made.
     So, too, do some passages in a paper by Rado upon `The Economic Principle in Psycho-Analytic Technique'. (10) The second part of this paper, which was to have dealt with psycho-analysis, has unfortunately never been published; but the first one, on hypnotism and catharsis, (11) contains much that is of interest. It includes a theory that the hypnotic subject introjects the hypnotist in the form of what Rado calls a `parasitic super-ego', which draws off the energy and takes over the functions of the subject's original superego. One feature of the situation brought out by Rado is the unstable and temporary nature of this whole arrangement. If, for instance, the hypnotist gives a command which is too much in opposition to the subject's original super-ego, the parasite is promptly extruded. And, in any case, when the state of hypnosis comes to an end, the sway of the parasite super-ego also terminates and the original super-ego resumes its functions.
     However debatable may be the details of Rado's description, it not only emphasizes once again the notion of the super-ego as the fulcrum of psychotherapy, but it draws attention to the important distinction between the effects of hypnosis and analysis in the matter of permanence. Hypnosis acts essentially in a temporary way, and Rado's theory of the parasitic super-ego, which does not really replace the original one but merely throws it out of action, gives a very good picture of its apparent workings. Analysis, on the other hand, in so so far as it seeks to affect the patient's super-ego, aims at something much more far-reaching and permanent--namely, at an integral change in the nature of the patient's super-ego itself. (12) Some even more recent developments in psycho-analytic theory give a hint, so it seems to me, of the kind of lines along which a clearer understanding of the question may perhaps be reached.

This latest growth of theory has been very much occupied with the destructive impulses and has brought them for the first time into the centre of interest; and attention has at the same time been concentrated on the correlated problems of guilt and anxiety. What I have in mind especially are the ideas upon the formation of the super-ego recently developed by Melanie Klein and the importance which she attributes to the processes of introjection and projection in the development of the personality. I will re-state what I believe to be her views in an exceedingly schematic outline. (13) The individual, she holds, is perpetually introjecting and projecting the objects of its idimpulses, and the character of the introjected objects depends on the character of the id-impulses directed towards the external objects. Thus, for instance, during the stage of a child's libidinal development in which it is dominated by feelings of oral aggression, its feelings towards its external object will be orally aggressive; it will then introject the object, and the introjected object will now act (in the manner of a super-ego) in an orally aggressive way towards the child's ego. The next event will be the projection of this orally aggressive introjected object back on to the external object, which will now in its turn appear to be orally aggressive. The fact of the external object being thus felt as dangerous and destructive once more causes the id-impulses to adopt an even more aggressive and destructive attitude towards the object in self-defence. A vicious circle is thus established. This process seeks to account for the extreme severity of the super-ego in small children, as well as for their unreasonable fear of outside objects. In the course of the development of the normal individual, his libido eventually reaches the genital stage, at which the positive impulses predominate. His attitude towards his external objects will thus become more friendly, and accordingly his introjected object (or super-ego) will become less severe and his ego's contact with reality will be less distorted. In the case of the neurotic, however, for various reasons-whether on account of frustration or of an incapacity of the ego to tolerate id-impulses, or of an inherent excess of the destructive components-development to the genital stage does not occur, but the individual remains fixated at a pre-genital level. His ego is thus left exposed to the pressure of a savage id on the one hand and a correspondingly savage super-ego on the other, and the vicious circle I have just described is perpetuated.

I should like to suggest that the hypothesis which I have stated in this bald fashion may be useful in helping us to form a picture not only of the mechanism of a neurosis but also of the mechanism of its cure. There is. after all, nothing new in regarding a neurosis as essentially an obstacle or deflecting force in the path of normal development; nor is there anything new in the belief that psycho-analysis (owing to the peculiarities of the analytic situation) is able to remove the obstacle and so allow the normal development to proceed. I am only trying to make our conceptions a little more precise by supposing that the pathological obstacle to the neurotic individual's further growth is in the nature of a vicious circle of the kind I have described. If a breach could somehow or other be made in the vicious circle, the processes of development would proceed upon their normal course. If, for instance, the patient could be made less frightened of his super-ego or introjected object, he would project less terrifying imagos on to the outer object and would therefore have less need to feel hostility towards it; the object which he then introjected would in turn be less savage in its pressure upon the id-impulses. which would be able to lose something of their primitive ferocity. In short, a benign circle would be set up instead of the vicious one, and ultimately the patient's libidinal development would proceed to the genital level, when, as in the case of a normal adult, his super-ego will be comparatively mild and his ego will have a relatively undistorted contact with reality. (14)
     But at what point in the vicious circle is the breach to be made and how is it actually to be effected`? It is obvious that to alter the character of a person's super-ego is easier said than done. Nevertheless, the quotations that I have already made from earlier discussions of the subject strongly suggest that the super-ego will be found to play an important part in the solution of our problem. Before we go further, however, it will be necessary to consider a little more closely the nature of what is described as the analytic situation. The relation between the two persons concerned in it is a highly complex one, and for our present purposes I am going to isolate two elements in it. In the first place, the patient in analysis tends to centre the whole of his idimpulses upon the analyst. I shall not comment further upon this fact or its implications, since they are so immensely familiar. I will only emphasize their vital importance to all that follows and proceed at once to the second element of the analytic situation which I wish to isolate. The patient in analysis tends to accept the analyst in some way or other as a substitute for his own super-ego. I propose at this point to imitate with a slight difference the convenient phrase which was used by Rado in his account of hypnosis and to say that in analysis the patient tends to make the analyst into an `auxiliary super-ego'. This phrase and the relation described by it evidently require some explanation.

When a neurotic patient meets a new object in ordinary life, according to our underlying hypothesis he will tend to project on to it his introjected archaic objects and the new object will become to that extent a phantasy object. It is to be presumed that his introjected objects are more or less separated out into two groups, which function as a `good' introjected object (or mild super-ego) and a `bad' introjected object (or harsh super-ego). According to the degree to which his ego maintains contacts with reality, the `good' introjected object will be projected on to benevolent real outside objects and the `bad' one on to malignant real outside objects. Since, however, he is by hypothesis neurotic, the `bad' introjected object will predominate, and will tend to be projected more than the `good' one; and there will further be a tendency, even where to begin with the `good' object was projected, for the `bad' one after a time to take its place. Consequently, it will be true to say that in general the neurotic's phantasy objects in the outer world will be predominantly dangerous and hostile. Moreover, since even his `good' introjected objects will be `good' according to an archaic and infantile standard, and will be to some extent maintained simply for the purpose of counteracting the `bad' objects, even his `good' phantasy objects in the outer world will be very much out of touch with reality. Going back now to the moment when our neurotic patient meets a new object in real life and supposing (as will be the more usual case) that he projects his `bad' introjected object on to it--the phantasy external object will then seem to him to be dangerous; he will be frightened of it and, to defend himself against it, will become more angry. Thus when he introjects this new object in turn, it will merely be adding one more terrifying imago to those he has already introjected. The new introjected imago will in fact simply be a duplicate of the original archaic ones, and his super-ego will remain almost exactly as it was. The same will be also true mutatis mutandis where he begins by projecting his `good' introjected object on to the new external object he has met with. No doubt, as a result, there will be a slight strengthening of his kind super-ego at the expense of his harsh one, and to that extent his condition will he improved. But there will be no qualitative change in his super-ego, for the new `good' object introjected will only be a duplicate of an archaic original and will only reinforce the archaic `good' super-ego already present.
     The effect when this neurotic patient comes in contact with a new object in analysis is from the first moment to create a different situation. His superego is in any case neither homogeneous nor well-organised; the account we have given of it hitherto has been over-simplified and schematic. Actually the introjected imagos which go to make it up are derived from a variety of different stages of his history and function to some extent independently. Now, owing to the peculiarities of the analytic circumstances and of the analyst's behaviour, the introjected imago of the analyst tends in part to be rather definitely separated off from the rest of the patient's super-ego. (This, of course, presupposes a certain degree of contact with reality on his part. Here we have one of the fundamental criteria of accessibility to analytic treatment; another, which we have already implicitly noticed, is the patient's ability to attach his id-impulses to the analyst.) This separation between the imago of the introjected analyst and the rest of the patient's super-ego becomes evident at quite an early stage of the treatment; for instance in connection with the fundamental rule of free association. The new bit of super-ego tells the patient that he is allowed to say anything that may come into his head. This works satisfactorily for a little; but soon there comes a conflict between the new bit and the rest, for the original super-ego says: `You must not say this, for, if you do, you will be using an obscene word or betraying so-and-so's confidences'. The separation off of the new bit--what I have called the `auxiliary' super-ego--tends to persist for the very reason that it usually operates in a different direction from the rest of the super-ego. And this is true not only of the `harsh' super-ego but also of the `mild' one. For, though the auxiliary super-ego is in fact kindly, it is not kindly in the same archaic way as the patient's introjected `good' imagos. The most important characteristic of the auxiliary super-ego is that its advice to the ego is consistently based upon real and contemporary considerations and this in itself serves to differentiate it from the greater part of the original super-ego.
     In spite of this, however, the situation is extremely insecure. There is a constant tendency for the whole distinction to break down. The patient is liable at any moment to project his terrifying imago on to the analyst just as though he were anyone else he might have met in the course of his life. If this happens, the introjected imago of the analyst will be wholly incorporated into the rest of the patient's harsh super-ego, and the auxiliary super-ego will disappear. And even when the content of the auxiliary super-ego's advice is realised as being different from or contrary to that of the original super-ego, very often its quality will be felt as being the same. For instance, the patient may feel that the analyst has said to him: `If you don't say whatever comes into your head, I shall give you a good hiding', or, `If you don't become conscious of this piece of the unconscious I shall turn you out of the room'. Nevertheless, labile though it is, and limited as is its authority, this peculiar relation between the analyst and the patient's ego seems to put into the analyst's grasp his main instrument in assisting the development of the therapeutic process. What is this main weapon in the analyst's armoury? Its name springs at once to our lips. The weapon is, of course, interpretation. And here we reach the core of the problem that I want to discuss in the present paper.

What, then, is interpretation? and how does it work? Extremely little seems to be known about it, but this does not prevent an almost universal belief in its remarkable efficacy as a weapon: interpretation has, it must be confessed, many of the qualities of a magic weapon. It is, of course, felt as such by many patients. Some of them spend hours at a time in providing interpretations of their own--often ingenious, illuminating, correct. Others, again, derive a direct libidinal gratification from being given interpretations and may even develop something parallel to a drug-addiction to them. In nonanalytical circles interpretation is usually either scoffed at as something ludicrous, or dreaded as a frightful danger. This last attitude is shared, I think, more than is often realized, by a certain number of analysts. This was particularly revealed by the reactions shown in many quarters when the idea of giving interpretations to small children was first mooted by Melanie Klein. But I believe it would be true in general to say that analysts are inclined to feel interpretation as something extremely powerful whether for good or ill. I am speaking now of our feelings about interpretation as distinguished from our reasoned beliefs. And there might seem to be a good many grounds for thinking that our feelings on the subject tend to distort our beliefs. At all events, many of these beliefs seem superficially to be contradictory; and the contradictions do not always spring from different schools of thought, but are apparently sometimes held simultaneously by one individual. Thus, we are told that if we interpret too soon or too rashly, we run the risk of losing a patient: that unless we interpret promptly and deeply we run the risk of losing a patient; that interpretation may give rise to intolerable and unmanageable outbreaks of anxiety by `liberating' it: that interpretation is the only way of enabling a patient to cope with an unmanageable outbreak of anxiety by 'resolving' it; that interpretations must always refer to material on the very point of emerging into consciousness: that the most useful interpretations are really deep ones; `Be cautious with your interpretations!' says one voice; `When in doubt, interpret!' says another. Nevertheless, although there is evidently a good deal of confusion in all of this, I do not think these views are necessarily incompatible; the various pieces of advice may turn out to refer to different circumstances and different cases and to imply different uses of the word `interpretation'.
     For the word is evidently used in more than one sense. It is, after all, perhaps only a synonym for the old phrase we have already come across--making what is unconscious conscious', and it shares all of that phrase's ambiguities. For in one sense, if you give a German-English dictionary to someone who knows no German, you will be giving him a collection of interpretations, and this, I think, is the kind of sense in which the nature of interpretation has been discussed in a recent paper by Bernfeld. (15) Such descriptive interpretations have evidently no relevance to our present topic, and I shall proceed without more ado to define as clearly as I can one particular sort of interpretation, which seems to me to be actually the ultimate instrument of psycho-analytic therapy and to which for convenience I shall give the name of `mutative' interpretation.
     I shall first of all give a schematized outline of what I understand by a mutative interpretation, leaving the details to be filled in afterwards; and, with a view to clarity of exposition, I shall take as an instance the interpretation of a hostile impulse. By virtue of his power (his strictly limited power) as auxiliary super-ego, the analyst gives permission for a certain small quantity of the patient's id-energy (in our instance, in the form of an aggressive impulse) to become conscious. (16) Since the analyst is also, from the nature of things, the object of the patient's id-impulses, the quantity of these impulses which is now released into consciousness will become consciously directed towards the analyst. This is the critical point. If all goes well, the patient's ego will become aware of the contrast between the aggressive character of his feelings and the real nature of the analyst, who does not behave like the patient's `good' or `bad' archaic objects. The patient, that is to say, will become aware of a distinction between his archaic phantasy object and the real external object. The interpretation has now become a mutative one, since it has produced a breach in the neurotic vicious circle. For the patient, having become aware of the lack of aggressiveness in the real external object, will be able to diminish his own aggressiveness; the new object which he introjects will be less aggressive, and consequently the aggressiveness of his super-ego will also be diminished. As a further corollary to these events, and simultaneously with them, the patient will obtain access to the infantile material which is being re-experienced by him in his relation to the analyst.
     Such is the general scheme of the mutative interpretation. You will notice that in my account the process appears to fall into two phases. I am anxious not to pre-judge the question of whether these two phases are in temporal sequence or whether they may not really be two simultaneous aspects of a single event. But for descriptive purposes it is easier to deal with them as though they were successive. First, then, there is the phase in which the patient becomes conscious of a particular quantity of id-energy as being directed towards the analyst; and secondly there is the phase in which the patient becomes aware that this id-energy is directed towards an archaic phantasy object and not towards a real one.

The first phase of a mutative interpretation--that in which a portion of the patient's id-relation to the analyst is made conscious in virtue of the latter's position as auxiliary super-ego--is in itself complex. In the classical model of an interpretation, the patient will first be made aware of a state of tension in his ego, will next be made aware that there is a repressive factor at work (that his super-ego is threatening him with punishment), and will only then be made aware of the id-impulse which has stirred up the protests of his super-ego and so given rise to the anxiety in his ego. This is the classical scheme. In actual practice, the analyst finds himself working from all three sides at once, or in irregular succession. At one moment a small portion of the patient's super-ego may be revealed to him in all its savagery, at another the shrinking defencelessness of his ego, at yet another his attention may be directed to the attempts which he is making at restitution-at compensating for his hostility; on some occasions a fraction of id-energy may even be directly encouraged to break its way through the last remains of an already weakened resistance. There is, however, one characteristic which all of these various operations have in common; they are essentially upon a small scale. For the mutative interpretation is inevitably governed by the principle of minimal doses. It is, I think, a commonly agreed clinical fact that alterations in a patient under analysis appear almost always to be extremely gradual: we are inclined to suspect sudden and large changes as an indication that suggestive rather than psycho-analytic processes are at work. The gradual nature of the changes brought about in psycho-analysis will be explained if, as I am suggesting, those changes are the result of the summation of an immense number of minute steps, each of which corresponds to a mutative interpretation. And the smallness of each step is in turn imposed by the very nature of the analytic situation. For each interpretation involves the release of a certain quantity of id-energy, and, as we shall see in a moment, if the quantity released is too large, the highly unstable state of equilibrium which enables the analyst to function as the patient's auxiliary super-ego is bound to be upset. The whole analytic situation will thus be imperilled, since it is only in virtue of the analyst's acting as auxiliary super-ego that these releases of idenergy can occur at all.
     Let us examine in greater detail the effects which follow from the analyst attempting to bring too great a quantity of id-energy into the patient's consciousness all at once. (17) On the one hand, nothing whatever may happen, or on the other hand there may be an unmanageable result; but in neither event will a mutative interpretation have been effected. In the former case (in which there is apparently no effect) the analyst's power as auxiliary superego will not have boon strong enough for the job ho has sot himself. But this again may be for two very different reasons. It may be that the id-impulses ho was trying to bring out were not in fact sufficiently urgent at the moment: for, after all, the emergence of an id-impulse depends on two factors-not only on the permission of the super-ego, but also on the urgency (the degree of cathexis) of the id-impulse itself. This, then, may be ono cause of an apparently negative response to an interpretation, and evidently a fairly harmless ono. But the same apparent result may also be duo to something else; in spite of the id-impulse being really urgent, the strength of the patient's own repressive forces (the degree of repression) may have been too great to allow his ego to listen to the persuasive voice of the auxiliary superego. Now here we have a situation dynamically identical with the next ono we have to consider, though economically different. This next situation is one in which the patient accepts the interpretation, that is, allows the idimpulse into his consciousness, but is immediately overwhelmed with anxiety. This may show itself in a number of ways: for instance, the patient may produce a manifest anxiety-attack, or he may exhibit signs of `real' anger with the analyst with complete lack of insight, or he may break off the analysis. In any of these cases the analytic situation will, for the moment at least, have broken down. The patient will be behaving just as the hypnotic subject behaves when, having been ordered by the hypnotist to perform an action too much at variance with his own conscience, ho breaks off the hypnotic relation and wakes up from his trance. This state of things, which is manifest where the patient responds to an interpretation with an actual outbreak of anxiety or ono of its equivalents, may be latent where the patient shows no response. And this latter case may be the more awkward of the two, since it is masked, and it may sometimes, I think, be the effect of a greater overdose of interpretation than where manifest anxiety arises (though obviously other factors will be of determining importance here and in particular the nature of the patient's neurosis). I have ascribed this threatened collapse of the analytic situation to an overdose of interpretation: but it might be more accurate in some ways to ascribe it to an insufficient dose. For what has happened is that the second phase of the interpretative process has not occurred: the phase in which the patient becomes aware that his impulse is directed towards an archaic phantasy object and not towards a real one.

In the second phase of a complete interpretation, therefore, a crucial part is played by the patient's sense of reality: for the successful outcome of that phase depends upon his ability, at the critical moment of the emergence into consciousness of the released quantity of id-energy, to distinguish between his phantasy object and the real analyst. The problem hero is closely related to one that I have already discussed, namely that of the extreme liability of the analyst's position as auxiliary super-ego. The analytic situation is all the time threatening to degenerate into a 'real' situation. But this actually means the opposite of what it appears to. It moans that the patient is all the time on the brink of turning the real external object (the analyst) into the archaic one; that is to say, he is on the brink of projecting his primitive introjected imagos on to him. In so far as the patient actually does this, the analyst becomes like anyone else that he meets in real life--a phantasy object. The analyst then ceases to possess the peculiar advantages derived from the analytic situation; ho will be introjected like all other phantasy objects into the patient's superego, and will no longer be able to function in the peculiar ways which are essential to the effecting of a mutative interpretation. In this difficulty the patient's sense of reality is an essential but a very feeble ally; indeed, an improvement in it is ono of the things that we hope the analysis will bring about. It is important, therefore, not to submit it to any unnecessary strain: and that is the fundamental reason why the analyst must avoid any real behaviour that is likely to confirm the patient's view of him as a 'bad' or a good' phantasy object. This is perhaps more obvious as regards the 'bad' object. If, for instance, the analyst wore to show that he was really shocked or frightened by ono of the patient's id-impulses, the patient would immediately treat him in that respect as a dangerous object and introject him into his archaic severe super-ego. Thereafter, on the one hand, there would be a diminution in the analyst's power to function as an auxiliary super-ego and to allow the patient's ego to become conscious of his id-impulses-that is to say, in his power to bring about the first phase of a mutative interpretation; and, on the other hand, ho would, as a real object, become sensibly less distinguishable from the patient's 'bad' phantasy object and to that extent the carrying through of the second phase of a mutative interpretation would also be made more difficult. Or again, there is another case. Supposing the analyst behaves in an opposite way and actively urges the patient to give free rein to his id-impulses. There is then a possibility of the patient confusing the analyst with the imago of a treacherous parent who first encourages him to seek gratification, and then suddenly turns and punishes him. In such a case, the patient's ego may look for defence by itself suddenly turning upon the analyst as though he were his own id, and treating him with all the severity of which his super-ego is capable. Here again, the analyst is running a risk of losing his privileged position. But it may be equally unwise for the analyst to act really in such a way as to encourage the patient to project his `good' introjected object on to him. For the patient will then tend to regard him as a good object in an archaic sense and will incorporate him with his archaic `good' imagos and will use him as a protection against his `bad' ones. In that way, his infantile positive impulses as well as his negative ones may escape analysis, for there may no longer be a possibility for his ego to make a comparison between the phantasy external object and the real one. It will perhaps be argued that, with the best will in the world, the analyst, however careful he may be, will be unable to prevent the patient from projecting these various imagos on to him. This is of course indisputable, and, indeed, the whole effectiveness of analysis depends upon its being so. The lesson of these difficulties is merely to remind us that the patient's sense of reality has the narrowest limits. It is a paradoxical fact that the best way of ensuring that his ego shall be able to distinguish between phantasy and reality is to withhold reality from him as much as possible. But it is true. His ego is so weak--so much at the mercy of his id and super-ego-that he can only cope with reality if it is administered in minimal doses. And these doses are in fact what the analyst gives him, in the form of interpretations.

It seems to me possible that an approach to the twin practical problems of interpretation and reassurance may be facilitated by this distinction between the two phases of interpretation. Both procedures may, it would appear, be useful or even essential in certain circumstances and inadvisable or even dangerous in others. In the case of interpretation (18) the first of our hypothetical phases may be said to `liberate' anxiety, and the second to `resolve' it. Where a quantity of anxiety is already present or on the point of breaking out, an interpretation, owing to the efficacy of its second phase, may enable the patient to recognize the unreality of his terrifying phantasy object and so to reduce his own hostility and consequently his anxiety. On the other hand, to induce the ego to allow a quantity of id-energy into consciousness is obviously to court an outbreak of anxiety in a personality with a harsh superego. And this is precisely what the analyst does in the first phase of an interpretation. As regards `reassurance', I can only allude briefly here to some of the problems it raises. (19) I believe, incidentally, that the term needs to be defined almost as urgently as `interpretation', and that it covers a number of different mechanisms. But in the present connection reassurance may be regarded as behaviour on the part of the analyst calculated to make the patient regard him as a `good' phantasy object rather than as a real one. I have already given some reasons for doubting the expediency of this, though it seems to be generally felt that the procedure may sometimes be of great value, especially in psychotic cases. It might, moreover, be supposed at first sight that the adoption of such an attitude by the analyst might actually directly favour the prospect of making a mutative interpretation. But I believe that it will be seen on reflection that this is not in fact the case: for precisely in so far as the patient regards the analyst as his phantasy object, the second phase of the interpretation does not occur--since it is of the essence of that phase that in it the patient should make a distinction between his phantasy object and the real one. It is true that his anxiety may be reduced; but this result will not have been achieved by a method that involves a permanent qualitative change in his super-ego. Thus, whatever tactical importance reassurance may possess, it cannot, I think, claim to be regarded as an ultimate operative factor in psycho-analytic therapy.
     It must here be noticed that certain other sorts of behaviour on the part of the analyst may be dynamically equivalent to the giving of a mutative interpretation, or to one or other of the two phases of that process. For instance, an `active' injunction of the kind contemplated by Ferenczi may amount to an example of the first phase of an interpretation; the analyst is making use of his peculiar position in order to induce the patient to become conscious in a particularly vigorous fashion of certain of his id-impulses. One of the objections to this form of procedure may be expressed by saying that the analyst has very little control over the dosage of the id-energy that is thus released, and very little guarantee that the second phase of the interpretation will follow. He may therefore be unwittingly precipitating one of those critical situations which are always liable to arise in the case of an incomplete interpretation. Incidentally, the same dynamic pattern may arise when the analyst requires the patient to produce a `forced' phantasy or even (especially at an early stage in an analysis) when the analyst asks the patient a question; here again, the analyst is in effect giving a blindfold interpretation, which it may prove impossible to carry beyond its first phase. On the other hand, situations are fairly constantly arising in the course of an analysis in which the patient becomes conscious of small quantities of id-energy without any direct provocation on the part of the analyst. An anxiety situation might then develop, if it were not that the analyst, by his behaviour or, one might say, absence of behaviour, enables the patient to mobilize his sense of reality and make the necessary distinction between an archaic object and a real one. What the analyst is doing here is equivalent to bringing about the second phase of an interpretation, and the whole episode may amount to the making of a mutative interpretation. It is difficult to estimate what proportion of the therapeutic changes which occur during analysis may not be due to implicit mutative interpretations of this kind. Incidentally, this type of situation seems sometimes to be regarded, incorrectly as I think, as an example of reassurance.

But it is now time to turn to two other characteristics which appear to be essential properties of every mutative interpretation. There is in the first place one already touched upon in considering the apparent or real absence of effect which sometimes follows upon the giving of an interpretation. A mutative interpretation can only be applied to an id-impulse which is actually in a state of cathexis. This seems self-evident; for the dynamic changes in the patient's mind implied by a mutative interpretation can only be brought about by the operation of a charge of energy originating in the patient himself: the function of the analyst is merely to ensure that the energy shall flow along one channel rather than along another. It follows from this that the purely informative `dictionary' type of interpretation will be non-mutative, however useful it may be as a prelude to mutative interpretations. And this leads to a number of practical inferences. Every mutative interpretation must be emotionally `immediate'; the patient must experience it as something actual. This requirment, that the interpretation must be `immediate', may be expressed in another way by saying that interpretations must always be directed to the `point of urgency'. At any given moment some particular id-impulse will be in activity; this is the impulse that is susceptible of mutative interpretation at that time, and no other one. It is, no doubt, neither possible nor desirable to be giving mutative interpretations all the time; but, as Melanie Klein has pointed out, it is a most precious quality in an analyst to be able at any moment to pick out the point of urgency. (20)

But the fact that every mutative interpretation must deal with an `urgent' impulse takes us back one more to the commonly felt fear of the explosive possibilities of interpretation, and particularly of what is vaguely referred to as `deep' interpretation. The ambiguity of the term, however, need not bother us. It describes, no doubt, the interpretation of material which is either genetically early and historically distant from the patient's actual experience or which is under an especially heavy weight of repression--material, in any case, which is in the normal course of things exceedingly inaccessible to his ego and remote from it. There seems reason to believe, moreover, that the anxiety which is liable to be aroused by the approach of such material to consciousness may be of peculiar severity. (21) The question whether it is `safe' to interpret such material will, as usual, mainly depend upon whether the second phase of the interpretation can be carried through. In the ordinary run of case the material which is urgent during the earlier stages of the analysis is not deep. We have to deal at first only with more or less far-going displacements of the deep impulses, and the deep material itself is only reached later and by degrees, so that no sudden appearance of unmanageable quantities of anxiety is to be anticipated. In exceptional cases, however, owing to some peculiarity in the structure of the neurosis, deep impulses may be urgent at a very early stage of the analysis. We are then faced by a dilemma. If we give an interpretation of this deep material, the amount of anxiety produced in the patient may be so great that his sense of reality may not be sufficient to permit of the second phase being accomplished, and the whole analysis may be jeopardised. But it must not be thought that, in such critical cases as we are now considering, the difficulty can necessarily be avoided simply by not giving any interpretation or by giving more superficial interpretations of non-urgent material or by attempting reassurances. It seems probable, in fact, that these alternative procedures may do little or nothing to obviate the trouble; on the contrary, they may even exacerbate the tension created by the urgency of the deep impulses which are the actual cause of the threatening anxiety. Thus the anxiety may break out in spite of these palliative efforts and, if so, it will be doing so under the most unfavourable conditions, that is to say, outside the mitigating influences afforded by the mechanism of interpretation. It is possible, therefore, that, of the two alternative procedures which are open to the analyst faced by such a difficulty, the interpretation of the urgent id-impulses, deep though they may be, will actually be the safer.

I shall have occasion to return to this point for a moment later on, but I must now proceed to the mention of one further quality which it seems necessary for an interpretation to possess before it can be mutative, a quality which is perhaps only another aspect of the one we have been describing. A mutative interpretation must be `specific': that is to say, detailed and concrete. This is, in practice, a matter of degree. When the analyst embarks upon a given theme, his interpretations cannot always avoid being vague and general to begin with; but it will be necessary eventually to work out and interpret all the details of the patient's phantasy system. In proportion as this is done the interpretations will be mutative, and much of the necessity for apparent repetitions of interpretations already made is really to be explained by the need for filling in the details. I think it possible that some of the delays which despairing analysts attribute to the patient's id-resistance could be traced to this source. It seems as though vagueness in interpretation gives the defensive forces of the patient's ego the opportunity, for which they are always on the lookout, of baffling the analyst's attempt at coaxing an urgent id-impulse into consciousness. A similarly blunting effect can be produced by certain forms of reassurance, such as the tacking on to an interpretation of an ethnological parallel or of a theoretical explanation: a procedure which may at the last moment turn a mutative interpretation into a non-mutative one. The apparent effect may be highly gratifying to the analyst; but later experience may show that nothing of permanent use has been achieved or even that the patient has been given an opportunity for increasing the strength of his defences. Here we have evidently reached a topic discussed not long ago by Edward Glover in one of the very few papers in the whole literature which seriously attacks the problem of interpretation. (22) Glover argues that, whereas a blatantly inexact interpretation is likely to have no effect at all, a slightly inexact one may have a therapeutic effect of a non-analytic, or rather anti-analytic, kind by enabling the patient to make a deeper and more efficient repression. He uses this as a possible explanation of a fact that has always seemed mysterious, namely, that in the earlier days of analysis, when much that we now know of the characteristics of the unconscious was still undiscovered, and when interpretation must therefore often have been inexact, therapeutic results were nevertheless obtained.

The possibility which Glover here discusses serves to remind us more generally of the difficulty of being certain that the effects that follow any given interpretation are genuinely the effects of interpretation and not transference phenomena of one kind or another. I have already remarked that many patients derive direct libidinal gratification from interpretation as such; and I think that some of the striking signs of abreaction which occasionally follow an interpretation ought not necessarily to be accepted by the analyst as evidence of anything more than that the interpretation has gone home in a libidinal sense.
     The whole problem, however, of the relation of abreaction to psychoanalysis is a disputed one. Its therapeutic results seem, up to a point, undeniable. It was from them, indeed, that analysis was born; and even today there are psycho-therapists who rely on it almost exclusively. During the War [World War 1], in particular, its effectiveness was widely confirmed in cases of 'shell-shock'. It has also been argued often enough that it plays a leading part in bringing about the results of psycho-analysis. Rank and Ferenczi, for instance, declared that in spite of all advances in our knowledge abreaction remained the essential agent in analytic therapy. (23) More recently, Reik has supported a somewhat similar view in maintaining that `the element of surprise is the most important part of analytic technique'. (24) A much less extreme attitude is taken by Nunberg in the chapter upon therapeutics in his text-book of psycho-analysis . (25) But he, too, regards abreaction as one of the component factors in analysis, and in two ways. In the first place, he mentions the improvement brought about by abreaction in the usual sense of the word, which he plausibly attributes to a relief of endo-psychic tension due to a discharge of accumulated affect. And in the second place, he points to a similar relief of tension upon a small scale arising from the actual process of becoming conscious of something hitherto unconscious, basing himself upon a statement of Freud's that the act of becoming conscious involves a discharge of energy. (26) On the other hand, Rado appears to regard abreaction as opposed in its function to analysis. He asserts that the therapeutic effect of catharsis is to be attributed to the fact that (together with other forms of non-analytic psycho-therapy) it offers the patient an artificial neurosis in exchange for his original one, and that the phenomena observable when abreaction occurs are akin to those of an hysterical attack. (27) A consideration of the views of these various authorities suggests that what we describe as 'abreaction' may cover two different processes: one a discharge of affect and the other a libidinal gratification. If so, the first of these might be regarded (like various other procedures) as an occasional adjunct to analysis, sometimes, no doubt, a useful one, and possibly even as an inevitable accompaniment of mutative interpretations; whereas the second process might be viewed with more suspicion, as an event likely to impede analysis--especially if its true nature were unrecognised. But with either form there would seem good reason to believe that the effects of abreaction are permanent only in cases in which the predominant aetiological factor is an external event: that is to say, that it does not in itself bring about any radical qualitative alteration in the patient's mind. Whatever part it may play in analysis is thus unlikely to be of anything more than an ancillary nature.

If we now turn back and consider for a little the picture I have given of a mutative interpretation with its various characteristics, we shall notice that my description appears to exclude every kind of interpretation except those of a single class-the class, namely, of transference interpretations. Is it to be understood that no extra-transference interpretation can set in motion the chain of events which I have suggested as being the essence of psychoanalytical therapy? That is indeed my opinion, and it is one of my main objects in writing this paper to throw into relief-what has, of course, already been observed, but never, I believe, with enough explicitness-the dynamic distinctions between transference and extra-transference interpretations. These distinctions may be grouped under two heads. In the first place, extra-transference interpretations are far less likely to be given at the point of urgency. This must necessarily be so, since in the case of an extra-transference interpretation the object of the id-impulse which is brought into consciousness is not the analyst and is not immediately present, whereas, apart from the earliest stages of an analysis and other exceptional circumstances, the point of urgency is nearly always to be found in the transference. It follows that extra-transference interpretations tend to be concerned with impulses which are distant both in time and space and are thus likely to be devoid of immediate energy. In extreme instances, indeed, they may approach very closely to what I have already described as the handing-over to the patient of a German-English dictionary. But in the second place, once more owing to the fact that the object of the id-impulse is not actually present, it is less easy for the patient, in the case of an extra-transference interpretation, to become directly aware of the distinction between the real object and the phantasy object. Thus it would appear that, with extratransference interpretations, on the one hand what I have described as the first phase of a mutative interpretation is less likely to occur, and on the other hand, if the first phase does occur, the second phase is less likely to follow. In other words, an extra-transference interpretation is liable to be both less effective and more risky than a transference one. (28) Each of these points deserves a few words of separate examination.
     It is, of course, a matter of common experience among analysts that it is possible with certain patients to continue indefinitely giving interpretations without producing any apparent effect whatever. There is an amusing criticism of this kind of 'interpretation-fanaticism' in the excellent historical chapter of Rank and Ferenczi. (29) But it is clear from their words that what they have in mind are essentially extra-transference interpretations, for the burden of their criticism is that such a procedure implies neglect of the analytic situation. This is the simplest case, where a waste of time and energy is the main result. But there are other occasions, on which a policy of giving strings of extra-transference interpretations is apt to lead the analyst into more positive difficulties. Attention was drawn by Reich (30) a few years ago in the course of some technical discussions in Vienna to a tendency among inexperienced analysts to get into trouble by eliciting from the patient great quantities of material in a disordered and unrelated fashion: this may, he maintained, be carried to such lengths that the analysis is brought to an irremediable state of chaos. He pointed out very truly that the material we have to deal with is stratified and that it is highly important in digging it out not to interfere more than we can help with the arrangement of the strata. He had in mind, of course, the analogy of an incompetent archaeologist, whose clumsiness may obliterate for all time the possibility of reconstructing the history of an important site. I do not myself feel so pessimistic about the results in the case of a clumsy analysis, since there is the essential difference that our material is alive and will, as it were, re-stratify itself of its own accord if it is given the opportunity: that is to say, in the analytic situation. At the same time, I agree as to the presence of the risk, and it seems to me to be particularly likely to occur where extra-transference interpretation is excessively or exclusively resorted to. The means of preventing it, and the remedy if it has occurred, lie in returning to transference interpretation at the point of urgency. For if we can discover which of the material is `immediate' in the sense I have described, the problem of stratification is automatically solved; and it is a characteristic of most extra-transference material that it has no immediacy and that consequently its stratification is far more difficult to decipher. The measures suggested by Reich himself for preventing the occurrence of this state of chaos are not inconsistent with mine; for he stresses the importance of interpreting resistances as opposed to the primary id-impulses themselves--and this, indeed, was a policy that was laid down at an early stage in the history of analysis. But it is, of course, one of the characteristics of a resistance that it arises in relation to the analyst; and thus the interpretation of a resistance will almost inevitably be a transference interpretation.
     But the most serious risks that arise from the making of extra-transference interpretations are due to the inherent difficulty in completing their second phase or in knowing whether their second phase has been completed or not. They are from their nature unpredictable in their effects. There seems, indeed, to be a special risk of the patient not carrying through the second phase of the interpretation but of projecting the id-impulse that has been made conscious on to the analyst. This risk, no doubt, applies to some extent also to transference interpretations. But the situation is less likely to arise when the object of the id-impulse is actually present and is moreover the same person as the maker of the interpretation. (31) (We may here once more recall the problem of `deep' interpretation, and point out that its dangers, even in the most unfavorable circumstances, seem to be greatly diminished if the interpretation in question is a transference interpretation.) Moreover, there appears to be more chance of this whole process occurring silently and so being overlooked in the case of an extra-transference interpretation, particularly in the earlier stages of an analysis. For this reason, it would seem to be important after giving an extra-transference interpretation to be specially on the qui vive for transference complications. This last peculiarity of extratransference interpretations is actually one of their most important from a practical point of view. For on account of it they can be made to act as `feeders' for the transference situation, and so to pave the way for mutative interpretations. In other words, by giving an extra-transference interpretation, the analyst can often provoke a situation in the transference of which he can then give a mutative interpretation.
     It must not be supposed that because I am attributing these special qualities to transference interpretations, I am therefore maintaining that no others should be made. On the contrary, it is probable that a large majority of our interpretations are outside the transference--though it should be added that it often happens that when one is ostensibly giving an extra-transference interpretation one is implicitly giving a transference one. A cake cannot be made of nothing but currants; and, though it is true that extra-transference interpretations are not for the most part mutative, and do not themselves bring about the crucial results that involve a permanent change in the patient's mind, they are none the less essential. If I may take an analogy from trench warfare, the acceptance of a transference interpretation corresponds to the capture of a key position, while the extra-transference interpretations correspond to the general advance and to the consolidation of a fresh line which are made possible by the capture of the key position. But when this general advance goes beyond a certain point, there will be another check, and the capture of a further key position will be necessary before progress can be resumed. An oscillation of this kind between transference and extratransference interpretations will represent the normal course of events in an analysis.

Although the giving of mutative interpretations may thus only occupy a small portion of psycho-analytic treatment, it will, upon my hypothesis, be the most important part from the point of view of deeply influencing the patient's mind. It may be of interest to consider in conclusion how a moment which is of such importance to the patient affects the analyst himself. Mrs. Klein has suggested to me that there must be some quite special internal difficulty to be overcome by the analyst in giving interpretations. And this, I am sure, applies particularly to the giving of mutative interpretations. This is shown in their avoidance by psycho-therapists of non-analytic schools; but many psycho-analysts will be aware of traces of the same tendency in themselves. It may be rationalized into the difficulty of deciding whether or not the particular moment has come for making an interpretation. But behind this there is sometimes a lurking difficulty in the actual giving of the interpretation, for there seems to be a constant temptation for the analyst to do something else instead. He may ask questions, or he may give reassurances or advice or discourses upon theory, or he may give interpretations--but interpretations that are not mutative, extra-transference interpretations, interpretations that are non-immediate, or ambiguous, or inexact--or he may give two or more alternative interpretations simultaneously, or he may give interpretations and at the same time show his own scepticism about them. All of this strongly suggests that the giving of a mutative interpretation is a crucial act for the analyst as well as for the patient, and that he is exposing himself to some great danger in doing so. And this in turn will become intelligible when we reflect that at the moment of interpretation the analyst is in fact deliberately evoking a quantity of the patient's id-energy while it is alive and actual and unambiguous and aimed directly at himself. Such a moment must above all others put to the test his relations with his own unconscious impulses.

I will end by summarizing the four main points of the hypothesis I have put forward:
(1) The final result of psycho-analytic therapy is to enable the neurotic patient's whole mental organization, which is held in check at an infantile stage of development, to continue its progress towards a normal adult state.
(2) The principle effective alteration consists in a profound qualitative modification of the patient's super-ego, from which the other alterations follow in the main automatically.
(3) This modification of the patient's super-ego is brought about in a series of innumerable small steps by the agency of mutative interpretations, which are effected by the analyst in virtue of his position as object of the patient's id-impulses and as auxiliary super-ego.
(4) The fact that the mutative interpretation is the ultimate operative factor in the therapeutic action of psycho-analysis does not imply the exclusion of many other procedures (such as suggestion, reassurance, abreaction, etc.) as elements in the treatment of any particular patient.

1. I have not attempted to compile a full bibliography of the subject, though a number of the more important contributions to it are referred to in the following pages.
2. Collected Papers, Vol. II.
3. P. 381.
4. New Introductory Lectures (1933). p. 194.
5. Pp. 117-118.
6. P. 77.
7. In Freud's paper at the Berlin Congress in 1922, subsequently expanded into The Ego and the Id (1923).
8. Int. J. Psycho-Anal., Vol. IV, 1923.
9. At the Salzburg Congress in 1924: 'A Metapsychological Description of the Process of Cure'. Int. J. Psycho-Anal., Vol. V1, 1925.
10. Also first read at Salzburg in 1924.
11. Int. J. Psycho-Anal., Vol. VI, 1925: in a revised form in German, Zeitschrift, Bd. XII, 1926.
12. This hypothesis seems to imply a contradiction of some authoritative pronouncements, according to which the structure of the super-ego is finally laid down and fixed at a very early age. Thus Freud appears in several passages to hold that the super-ego (or at all events its central core) is formed once and for all at the period at which the child emerges from its Oedipus complex. (See, for instance, The Ego and the Id, pp. 68-69.) So, too, Melanie Klein speaks of the development of the super-ego 'ceasing' and of its formation 'having reached completion' at the onset of the latency period (The Psycho-Analysis of Children, pp. 250 and 252), though in many other passages (e.g., p. 369) she implies that the superego can be altered at a later age under analysis. I do not know how far the contradiction is a real one. My theory does not in the least dispute the fact that in the normal course of events the super-ego becomes fixed at an early age and subsequently remains essentially unaltered. Indeed, it is a part of my view that in practice nothing except the process of psycho-analysis can alter it. It is of course a familiar fact that in many respects the analytic situation reconstitutes an infantile condition in the patient, so that the fact of being analysed may, as it were, throw the patient's super-ego once more into the melting pot. Or, again, perhaps it is another mark of the non-adult nature of the neurotic that his super-ego remains in a malleable state.
13. See The Psycho-Analysis of Children (1932), passim, especially Chapters VIII and IX.
14. A similar view has often been suggested by Melanie Klein. See, for instance, The PsychoAnalysis of Children, p. 369. It has been developed more explicitly and at greater length by Melitta Schmideberg: 'Zur Psychoanalyse asozialer Kinder and Jugendlicher' (Zeitschrift, Bd. XVIII, 1932).
15. 'Der Begriff der Deutung in der Psychoanalyse', Zeitschrift fur angewandte Psvchologie, Bd. 42, 1932. A critical summary of this by Gero will be found in Imago, Bd. XIX, 1933.
16. I am making no attempt at describing the process in correct meta-psychological terms. For instance, in Freud's view, the antithesis between conscious and unconscious is not, strictly speaking, applicable to instinctual impulses themselves, but only to the ideas which represent them in the mind. ('The Unconscious', Collected Papers, Vol. IV, p. 109.) Nevertheless, for the sake of simplicity, I speak throughout this paper of 'making id-impulses conscious'.
17. Incidentally, it seems as though a qualitative factor may be concerned as well: that is, some kinds of id-impulses may be more repugnant to the ego than others.
18. For the necessity for 'continuous and deep-going interpretations' in order to diminish or prevent anxiety-attacks, see Melanie Klein's Psycho-Analysis of Children, pp. 58-59. On the other hand: `The anxiety belonging to the deep levels is far greater, both in amount and intensity, and it is therefore imperative that its liberation should be duly regulated'. (Ibid., p. 139.)
19. Its uses were discussed by Melitta Schmideberg in a paper read to the British Psycho-Analytical Society on February 7, 1934.
20. The Psycho-Analysis of Children, pp. 58-59.
21. Ibid.. p. 139.
22. 'The Therapeutic Effect of Inexact Interpretation', J. In. Psycho-Anal., Vol. XII, 1931.
23. Entwicklungsziele tier Psychoanalyse (1924), p. 27.
24. 'New Ways in Psycho-Analytic Technique', In. J. Psycho-Anal.. Vol. XIV, 1933.
25. Allgemeine Neurosenlehre auf psychoanalytischer Grundlage (1932), pp. 303-304. This chapter appears in English in an abbreviated version as a contribution to Lorand's Psycho-Analysis Today (1933). There is very little, I think, in Nunberg's comprehensive catalogue of the factors at work in analytic therapy that conflicts with the views expressed in the present paper, though I have given a different account of the inter-relation between those factors.
26. Beyond the Pleasure Principle, p. 28.
27. 'The Economic Principle in Psycho-Analytic Technique', Int. J. Psycho-Anal., Vol. VI. 1925.
28. This corresponds to the fact that the pseudo-analysts and 'wild' analysts limit themselves as a rule to extra-transference interpretations. It will he remembered that this was true of Freud's original 'wild' analyst ('Observations on "Wild" Psycho-Analysis' (1910). Collected Papers, Vol. 11).
29. Entsricklungsziele tier Psychoanalyse, p. 31 .
30. 'Bericht fiber das "Seminar fur psychoanalytische Therapie" in Wien', Zeitschrift, Bd. XIII. 1927. This has recently been re-published as a chapter in Reich's volume upon Charakteranalyse (1933), which contains a quantity of other material with an interesting hearing on the subject of the present paper.
31. It even seems likely that the whole possibility of effecting mutative interpretations may depend upon this fact that in the analytic situation the giver of the interpretation and the object of the id-impulse interpreted are one and the same person. I am not thinking here of the argument mentioned above--that it is easier under that condition for the patient to distinguish between his phantasy object and the real object--but of a deeper consideration. The patient's original super-ego is, as I have argued, a product of the introjection of his archaic objects distorted by the projection of his infantile id-impulses. I have also suggested that our only means of altering the character of this harsh original super-ego is through the mediation of an auxiliary super-ego which is the product of the patient's introjection of the analyst as an object. The process of analysis may from this point of view be regarded as an infiltration of the rigid and unadaptable original super-ego by the auxiliary super-ego with its greater contact with the ego and with reality. This infiltration is the work of the mutative interpretations: and it consists in a repeated process of introjection of imagos of the analyst--imagos, that is to say, of a real figure and not of an archaic and distorted projection--so that the quality of the original super-ego becomes gradually changed. And since the aim of the mutative interpretations is thus to cause the introjection of the analyst, it follows that the id-impulses which they interpret must have the analyst as their object. If this is so, the views expressed in the present paper will require some emendation. For in that case, the first criterion of a mutative interpretation would be that it must be a transference interpretation. Nevertheless, the quality of urgency would still remain important; for, of all the possible transference interpretations which could be made at any particular moment, only the one which dealt with an urgent id-impulse would he mutative. On the other hand, an extra-transference interpretation even of an extremely urgent id-impulse could never be mutative-though it might, of course, produce temporary relief along the lines of abreaction or reassurance.