Classics in the History of Psychology

An internet educational resource developed by

Christopher D. Green
York University, Toronto, Ontario

(Return to index)

The Association Method[1]
Carl G. Jung (1910)

First published in American Journal of Psychology, 31, 219-269.

Part 1 of 3


Ladies and Gentlemen: When I was honored with the invitation from Clark University to lecture before this esteemed assemblage, a wish was at the same time expressed that I should speak about my methods of work, and especially about the psychology of childhood. I hope to accomplish this task in the following manner:

In my first lecture I shall try to present to you the view points of my association methods; in my second lecture I shall discuss the significance of the familiar constellations; while in my third lecture I shall enter more fully into the psychology of the child.

I might easily confine myself exclusively to my theoretical views, but I believe that it will be better to illustrate my lectures with as many practical examples as possible. We shall therefore occupy ourselves first with the method of association, a method which has been of valuable assistance to me both practically and theoretically. The association method in vogue in psychology, as well as its history, is of course, so familiar to you that there is no need to speak of it. For practical purposes I make use of the following formulary: 

[p. 220]

This formulary has been constructed after many years of experience. The words are chosen and partially arranged in such a manner as to strike easily almost all complexes of practical occurrence. As shown by the above formulary there is a regular mixing of the grammatical qualities of the words. This, too, has its definite reasons.[2]

Before the experiment begins the test person receives the following instruction: "Answer as quickly as possible the first word that occurs to your mind." This instruction is so simple that it can easily be followed by anybody. The work itself, moreover, appears extremely easy, so that is might be expected that any one could accomplish it with the greatest facility and promptitude. But contrary to expectation the behavior is quite different. 

[p. 221]
[p. 222]

The following curves illustrate the course of the reaction time in an association experiment in four normal test persons. The length of each column denotes the length of the reaction time. 

[p. 223]

The illustrations below (pp. 224 ff.) show the course of the reaction time in hysterical individuals. The light cross-hatched columns denote the locations where the test person was unable to react (so-called failures).

The first thing that strikes us is the fact that many test persons show a marked prolongation of the reaction time. This would make us think at first of intellectual difficulties, - wrongly, however, as we are often dealing with very intelligent persons of fluent speech. The explanation lies rather in the emotions. In order to understand the matter comprehensively we must bear in mind that the association experiments cannot deal with a separated psychic function, for any psychic occurrence is never a thing in itself, but is always the resultant of the entire psychological past. The association experiment, too, is not merely a method for the reproduction of separated word couplets, but it is a kind of pastime, a conversation between experimenter and test person. In a certain sense it is even still more than that. Words are really something like condensed actions, situations, and things. When I present a word to the test person which denotes an action it is the same as if I should present to him the action itself, and ask him, "How do you behave towards it? What do you think of it? What do you do in this situation?" If I were a magician I should cause the situation corresponding to the stimulus word to appear in reality and placing the test person in its midst, I should then study his manner of reaction. The result of my stimulus words would thus undoubtedly approach infinitely nearer perfection. But as we are not magicians we must be contented with the linguistic substitutes for reality; at the same time we [p. 224] must not forget that the stimulus word will as a rule always conjure up its corresponding situation. It all depends on how the test person reacts to this situation. The situation "bride" or "bridegroom" will not evoke a simple reaction in a young lady; but the reaction will be deeply influenced by the provoked strong feeling tones, the more so if the experimenter be a man. 

It thus happens that the test person is often unable to react [p. 225] quickly and smoothly to all stimulus words. In reality, too, there are certain stimulus words which denote actions, situations, or things, about which the test person cannot think quickly and surely, and this fact is shown in the association experiments. The example which I have just presented shows an abundance of long reaction times and other disturbances. 

[p. 226] In this case the reaction to the stimulus word is in some way impeded, that is, the adaptation to the stimulus word is disturbed. The stimulus words are therefore merely a part of reality acting upon us; indeed, a person who shows such disturbances to the stimulus words, is in a certain sense really but imperfectly adapted to reality. 

Disease is an imperfect adaptation; hence in this case we are dealing with something morbid in the psyche, - with something which is either temporary or persistently pathological, that is, we are dealing with a psychoneurosis, with a functional disturbance of the mind. This rule, however, as we shall see later, is not without its exceptions.

Let us in the first place continue the discussion concerning the prolonged reaction time. It often happens that the test person actually does not know what to answer to the stimulus [p. 227] word. The test person waives any reaction; for the moment he totally fails to obey the original instructions, and shows himself incapable of adapting himself to the experimenter. If this phenomenon occurs frequently in an experiment it signifies a higher degree of disturbance in adjustment. I call attention to the fact that it is quite indifferent what reason the test person gives for the refusal. Some find that too many ideas suddenly occur to them, others, that not enough ideas come to their minds. In most cases, however, the difficulties first perceived are so deterrent that they actually give up the whole reaction. The following example shows a case of hysteria with many failures of reaction: 

In example 3 we find a characteristic phenomenon. The test person is not content with the requirements of the instruction, that is, she is not satisfied with one word but reacts with many words. She apparently does more and better than the instruction requires, but in so doing she does not fulfill the requirements of the instruction. Thus she reacts: - custom - good - barbaric; foolish - narrow minded - restricted; family - big - small - everything possible.

[p. 228] These examples show in the first place that many other words connect themselves with the reaction word. The test person is unable to suppress the ideas which subsequently occur to her. In doing this she also pursues a certain tendency which perhaps is more distinctly expressed in the following reaction: new - old - as an opposite. The addition of "as an opposite" denotes that the test person has the desire to add something explanatory or supplementary. This tendency is also shown in the following reaction: finger - not only hand, also foot - a limb - member - extremity.

Here we have a whole series of supplements. It seems as if the reaction were not sufficient for the test person, as if something else must always be added, as if what has been already said were incorrect or in some way imperfect. This feeling we may with Janet designate as the 'sentiment d'incomplêtude,' which by no means explains everything. I enter somewhat deeper into this phenomenon because it is quite frequently encountered in neurotic individuals. Indeed it is not merely a small and unimportant subsidiary manifestation in an insignificant experiment, but rather an elemental and universal manifestation which otherwise plays a rôle in the psychic life of neurotics.

With his desire to supplement the test person betrays a tendency to give the experimenter more than he wants, he even exerts the greatest efforts to seek further mental occurrences in order finally to discover something quite satisfactory. If we translate this elementary observation into the psychology of everyday life, it signifies that the test person has a tendency constantly to give to others more feeling than is required and expected. According to Freud, this is a sign of a reinforced object-libido, that is, it is a compensation for an inner unsatisfaction and voidness of feeling. In this elementary observation we therefore see one of the main qualities of hysterics, namely, the tendency to allow themselves to be carried away by everything, to attach themselves enthusiastically to everything, and to always promise too much and hence do little. Patients having this symptom, in my experience, are always hard to deal with; at first they are enthusiastically enraptured with the physician, for a time going so far as to accept everything blindly; but they soon merge into just as blind a resistance against the physician , thus rendering any educative influence absolutely impossible.

We see therefore in this phenomenon the expression of a tendency to give more than the instruction demands and expects. This tendency betrays itself also in other failures to follow the instruction:

to quarrel - angry - different things - I always quarrel at home;
to marry - how can you marry? - reunion - union;
[p. 229] plum - to eat - to pluck - what do you mean by it? - is it symbolic?
to sin - this idea is quite strange to me, I do not recognize it.

These reactions show that the test person gets away altogether from the situation of the experiment. For the instruction demands that he should answer only the word which next occurs to him. Here we find that the stimulus words apparently act with excessive strength, that they are taken as if they were direct personal questions. The test person entirely forgets that we deal with mere words which stand in print before us, and seeks in them a personal meaning; he tries to divine them and defend himself against them, thus altogether forgetting the instructions.

This elementary observation depicts another common peculiarity of hysterics, namely, that of taking everything personally, of never being able to remain objective, and of allowing themselves to be carried away by momentary impressions; this again shows the characteristics of the enhanced object-libido.

Another sign of impeded adaptation is the often occurring repetitions of the stimulus words. The test persons repeat the stimulus word as if they had not heard or understood it distinctly. They repeat it just as we repeat a difficult question in order better to grasp it before answering. This same tendency is shown in the experiment. The questions are repeated because the stimulus words act on hysterical individuals almost like difficult and personal questions. In principle it is the same phenomenon as the subsequent completion of the reaction.

In many experiments we observe that the same reaction constantly reappears to the most varied stimulus words. These words seem to possess a special reproduction tendency, and it is very interesting to examine their relationship to the test person. For example, I have observed a case in which the patient repeated the word "short" a great many times and often in places where it had no meaning. The test person could not directly state the reason for the repetition of the word "short." From experience I knew that such predicates always relate either to the test person himself or to the person nearest to him. I assumed that in this word "short" he designated himself, and that in this way he helped to express something very painful to him. The test person is of very small stature. He is the youngest of four brothers, who in contrast to him are all tall. He was always the "child" in the family, he was nicknamed "Short" and was treated by all as the "little one." This resulted in a total loss of self-confidence. Although he was intelligent, and despite long [p. 230] study, he could not decide to present himself for examination; he finally became impotent, and merged into a psychosis in which, whenever he was alone, he took delight in walking about in his room on his toes in order to appear taller. The word "short," therefore, signified to him a great many painful experiences. This is usually the case with the repeated words; they always contain something very important for the individual psychology of the test person.

The signs thus far depicted are not found arbitrarily spread throughout the whole experiment, but only in very definite locations; namely, in those stimulus words which strike against special emotionally accentuated complexes. This fact is the foundation of the so-called "diagnosis of facts" (Tatbestandsdiagnostik); that is, of the method employed to discover by means of an association experiment, the culprit among a number of persons suspected of a crime. That this is possible I should like to demonstrate briefly in a concrete case.

On the 6th of February, 1908, our supervisor reported to me that a nurse complained to her of having been robbed during the forenoon of the previous day. The facts were as follows: The nurse kept her money, amounting to 70 francs, in a pocketbook which she had placed in her cupboard where she also kept her clothes. The cupboard contained two compartments, of which one belonged to the nurse who was robbed, and the other to the head nurse. These two nurses and a third one, who was an intimate friend of the head nurse, slept in the same room where the cupboard was. The room was in a section which was regularly occupied in common by six nurses who had free access to this room. Given such a state of affairs it is not to be wondered that the supervisor shrugged her shoulders when I asked her whom she most suspected.

Further investigation showed that on the morning of the theft the above-mentioned friend of the head nurse was slightly indisposed and remained in bed in the room the whole morning. Hence, following the indications of the plaintiff, the theft could have taken place only in the afternoon. Of the other four nurses upon whom suspicion could fall, there was only one who regularly attended to the cleaning of the room in question, while the remaining three had nothing to do in this room, nor was it shown that any of them had spent any time there on the previous day.

It was therefore natural that these last three nurses should be regarded for the time being as less implicated, and I therefore began by subjecting the first three to the experiment.

From the particulars of the case, I also knew that the cupboard was locked but that the key was kept not far away in a
[p. 231] very conspicuous place, that on opening the cupboard the first thing to be seen was a fur ornament (boa), and, moreover, that the pocketbook was between the linen in an inconspicuous place. The pocketbook was of dark reddish leather, and contained the following objects: one 50 franc banknote, one 20 franc piece, some centimes, one small silver watch chain, one stencil used in the insane asylum to mark the kitchen utensils, and one small receipt from Dosenbach's shoeshop in Zürich.

Besides the plaintiff and the guilty one, only the head nurse knew the exact particulars of the deed, for as soon as the former missed her money she immediately asked the head nurse to help her find it, thus the head nurse had been able to learn the smallest details, which naturally rendered the experiment still more difficult, for she was precisely the one most suspected. The conditions for the experiment were better for the others, since they knew nothing concerning the particulars of the deed, and some not even that a crime had been committed. As critical stimulus words I selected the name of the robbed nurse, plus the following words: cupboard, door, open, key, yesterday, banknote, gold, 70, 50, 20, money, watch, pocketbook, chain, silver, to hide, fur, dark reddish, leather, centimes, stencil, receipt, Dosenbach. Besides these words which referred directly to the deed, I took also the following, which had a special affective value: theft, to take, to steal, suspicion, blame, court, police, to lie, to fear, to discover, to arrest, innocent.

The objection is often made to the last species of words that they may produce a strong affective resentment even in innocent persons, and for that reason one cannot attribute to them any comparative value. Nevertheless, it may always be questioned whether the affective resentment of an innocent person will have the same effect on the association as that of a guilty one, and that question can only be authoritatively answered by experience. Until the contrary shall be demonstrated, I maintain that even words of the above mentioned type may profitably be used.

I then distributed these critical words among twice as many indifferent stimulus words in such a manner that each critical word was followed by two indifferent ones. As a rule it is well to follow up the critical words by indifferent words in order that the action of the first may be clearly distinguished. But one may also follow up one critical word by another, especially if one wishes to bring into relief the action of the second. Thus I placed together "darkish red" and "leather," and "chain" and "silver."

After this preparatory work I undertook the experiment [p. 232] with the three above mentioned nurses. As examinations of this kind can be rendered into a foreign tongue only with the greatest difficulty, I will content myself with presenting the general results, and with giving some examples. I first undertook the experiment with the friend of the head nurse, and judging by the circumstances she appeared only slightly moved. The head nurse was next examined; she showed marked excitement, her pulse being 120 per minute immediately after the experiment. The last to be examined was the nurse who attended to the cleaning of the room in which the theft occurred. She was the most tranquil of the three; she displayed but little embarrassment, and only in the course of the experiment did it occur to her that she was suspected of stealing, a fact which manifestly disturbed her towards the end of the experiment.

The general impression from the examination spoke strongly against the head nurse. It seemed to me that she evinced a very "suspicious," or I might almost say, "impudent" countenance. With the definite idea of finding in her the guilty one I set about adding up the results.

One can make use of many special methods of computing, but they are not all equally good and equally exact. (One must always resort to calculation, as appearances are enormously deceptive.) The method which is most to be recommended is that of the probable average of the reaction time. It shows at a glance the difficulties which the person in the experiment had to overcome in the reaction.

The technique of this calculation is very simple. The probable average is the middle number of the various reaction times arranged in a series. The reaction times are, for example[3] placed in the following manner: 5, 5, 5, 7, 7, 7, 7, 8, 9, 9, 9, 12, 13, 14. The number found in the middle (8) is the probable average of this series. Following the order of the experiment, I shall denote the friend of the head nurse by the letter A, the head nurse by B, and the third nurse by C.

The probable averages of the reaction are:

A               B               C
10.0          12.0            13.5

No conclusions can be drawn from this result. But the average reaction times calculated separately for the indifferent reactions, for the critical, and for those immediately following the critical (post-critical) are more interesting. 

>From this example we see that whereas A has the shortest reaction time for the indifferent reactions, she shows in com-
[p. 233] parison to the other two persons of the experiment, the longest time for the critical reactions.

The difference between the reaction times, let us say between the indifferent and the critical, is 6 for A, 2 for B, and 3 for C, that is, it is more than double for A when compared with the other two persons.

In the same way we can calculate how many complex indicators there are on an average for the indifferent, critical, etc., reactions.

The difference between the indifferent and critical reactions for A = 0.7, for B = 0, for C = 0.4. A is again the highest.

Another question to consider is, in what special way do the imperfect reactions behave?

The result for A = 34%, for B = 28%, and for C = 30%.

Here, too, A reaches the highest value, and in this, I believe, we see the characteristic moment of the guilt-complex in A. I am, however, unable to explain here circumstantially the reasons why I maintain that memory errors are related to an emotional complex, as this would lead me beyond the limits of the present work. I therefore refer the reader to my work "Ueber die Reproductionsstörrungen im Associationsexperiment" (IX Beitrag der Diegnost. Associat. Studien).

As it often happens that an association of strong feeling tone produces in the experiment a perseveration, with the result that not only the critical association, but also two or three successive [p. 234] associations are imperfectly reproduced, it will be very interesting for our cases to see how many imperfect reproductions are so arranged in the series. The result of computation shows that the imperfect reproductions thus arranged in series are for A 64.7%, for B 55.5%, and for C 30.0%.

Again we find that A has the greatest percentage. To be sure this may partially depend on the fact that A also possesses the greatest number of imperfect reproductions. Given a small quantity of reactions it is usual that the greater the total number of the same the more imperfect reactions will occur in groups. But in order that this should be probable it could not occur in so great a measure as in our case, where on the other hand B and C have not a much smaller number of imperfect reactions when compared to A. It is significant that C with her slight emotions during the experiment shows the minimum of imperfect reproductions arranged in series.

As imperfect reproductions are also complex indicators, it is necessary to see how they distribute themselves in respect to the indifferent, critical, etc., reactions. 

It is hardly necessary to bring into prominence the differences between the indifferent and the critical reactions of the various subjects as shown by the resulting numbers of the table. In this respect, too, A occupies first place.

Naturally, here, too, there is a probability that the greater the quantity of the imperfect reproductions the greater is their number in the critical reactions. If we suppose that the imperfect reproductions are distributed regularly and without choice among all the reactions there will be a greater number of them for A (in comparison to B and C) even as reactions to critical words, since A has the greater number of imperfect reproductions. Admitting such a uniform distribution of the imperfect reproductions, it is easy to calculate how many we ought to expect to belong to each individual kind of reaction.
From this calculation it appears that the disturbances of reproductions which concern the critical reactions for A surpass by far the expected, for C they are 0.9 higher than the ex-[p. 235] pected, while for B the real number is less than the one expected. 

All this points to the fact than in the subject A the critical stimulus words acted with the greatest intensity, and hence the greatest suspicion falls on A. Practically one may venture to designate such a subject as probably guilty. The same evening A made a complete confession of the theft, and thus the success of the experiment was confirmed.

I maintain that such a result should be of scientific interest and worthy of consideration. There is much in experimental psychology which is less useful than the material treated in this work. Putting aside altogether the theoretical interest, we have in this case something that is not to be despised from a practical point of view, to wit, we have brought to light the culpable affair in a much easier and shorter way than is customary. What has been possible once or twice ought to be possible again in other cases, and it is well worth while to investigate the means of rendering the method increasingly capable of rapid and sure results.

This applicability of the experiment shows it possible to strike a concealed (indeed and unconscious) complex by means of a stimulus word; and conversely we may assume with great certainty that behind a reaction which shows a complex indicator there is a hidden complex, even though the test person strongly denies it. One must get rid of the idea that educated and intelligent test persons are able to see and admit their own complexes. Every human mind contains much that is unacknowledged and hence unconscious as such; and no one can boast that he stands completely above his complexes. Those who persist in maintaining it do not see the spectacles which they wear on their noses.

It has long been thought that the association experiment [p. 236] enables one to distinguish certain intellectual types. That is not the case. The experiment does not give us any particular insight into the purely intellectual, but rather only into the emotional processes. To be sure we can erect certain types of reaction; they are not, however, based on intellectual peculiarities, but depend entirely on the proportionate emotional state. Educated test persons usually show superficial and linguistically deep rooted associations, whereas the uneducated form more valuable associations and often of ingenious significance. This behavior would be paradoxical from an intellectual viewpoint. The meaningful associations of the uneducated are not really the product of intellectual thinking, but are simply the results of a special emotional state. The whole thing is more important to the uneducated, his emotion is greater and for that reason he pays more attention to the experiment than the educated person, and that is why his associations are more significant. Aside from the types determined by education we have to consider three principal individual types:

1. An objective type with undisturbed reactions.
2. A so-called complex type with many disturbances in the experiment occasioned by the constellation of a complex.
3. A so-called definition-type. This type consists in the fact that the reaction always gives an explanation or a definition of the content of the stimulus word; e. g.:
                                                                            apple, - a tree-fruit;

                 table,- a piece of household furniture;
   to promenade, - an activity;
  father, - chief of the family.

This type is chiefly found in stupid persons, and it is therefore quite usual in imbecility. But it can also be found in persons who are not really stupid, but who do not wish to be taken as stupid. Thus a young student from whom associations were taken by an older intelligent woman student reacted altogether with definitions. The test person was of the opinion that it was an examination in intelligence, and therefore directed most of his attention to the significance of the stimulus words; his associations, therefore, looked like those of an idiot. Not all idiots, however, react with definitions; probably only those so react who would like to appear smarter than they are, that is, those whom their stupidity is painful. I designate this widespread complex as "intelligence-complex." A normal test person reacts in a most overdrawn manner as follows:

                                                                              anxiety - heart anguish;

  to kiss - love's unfolding;
               to kiss - perception of friendship.

This type gives a constrained and unnatural impression. The test persons wish to be more than they are, they wish to
[p. 237] exert more influence than they really have. Hence we see that persons with an intelligence complex are usually not natural and unconstrained; that they are always somewhat unnatural and flowery; they show a predilection for complicated foreign words, high sounding quotations, and other intellectual ornaments. In this way they wish to influence their fellow beings, they wish to impress others with their apparent education and intelligence, and thus to compensate for the painful feeling of stupidity. The definition type is closely related to the predicate type, or to express it more precisely, to the predicate type expressing personal judgment (Wertprädikattypus). For example:

                                                                                flower - pretty;
                                                                                money - convenient;
                                                                                animal - ugly;
                                                                                knife - dangerous;
                                                                                death - ghastly.

In the definition type the intellectual significance of the stimulus word is rendered prominent, while in the predicate type it is its emotional significance. There are predicate types which are altogether overdrawn where there appear reactions like the following:
                                                                                piano - horrible;
                                                                                to sing - heavenly;

     mother - ardently loved;
                      father - something good, nice, holy.

In the definition type an absolute intellectual make-up is manifested or rather simulated, but here there is a very emotional one. Yet, just as the definition type really conceals a lack of intelligence so the excessive emotional expression conceals or overcompensates an emotional deficiency. This conclusion is very interestingly illustrated by the following discovery:

- On investigating the influence of the familiar milieus on the association type it was found that young individuals seldom possess a predicate type, but that on the other hand, the predicate type increases in frequency with the advancing age. In women the increase of the predicate type begins a little after the 40th year, and in men after the 60th. That is the precise time when, owing to the deficiency of sexuality, there actually occurs considerable emotional loss. If a test person evinces a distinct predicate type it may always be inferred that a marked internal emotional deficiency is thereby compensated. Still one cannot reason conversely, namely that an inner emotional deficiency must produce a predicate type, no more than that idiocy directly produces a definition type. A predicate type can also betray itself through the external behavior, as, for example, through a particular affectation, enthusiastic exclamations, an embellished behavior, and the constrained sounding language so often observed in society.

[p. 238] The complex type shows no particular tendency except the concealment of a complex, whereas the definition and predicate types betray a positive tendency to exert in some way a definite influence on the experimenter. But whereas the definition type tends to bring to light its intelligence, the predicate type displays its emotion. I need hardly add of what importance such determinations are for the diagnosis of character.

After finishing an association experiment I usually add another experiment of a different kind, the so-called reproduction. I repeat the same stimulus words and ask the test persons whether they still remember their former reactions. In many instances the memory fails, and as experience shows, these locations are stimulus words which touched an emotionally accentuated complex, or stimulus words immediately following such critical words.

This phenomenon has been designated as paradoxical and contrary to all experience. For it is known that emotionally accentuated things are better retained in memory than indifferent things. This is quite true, but it does not hold for the linguistic expression of an emotionally accentuated content. On the contrary one very easily forgets what he has said under emotion, one is even apt to contradict himself about it. Indeed the efficacy of cross-examinations in court depends on this fact. The reproduction method therefore serves to render still more prominent the complex stimulus. In normal persons we usually find a limited number of false reproductions, seldom more than 10-20%, while in abnormal persons, especially in hysterics, we often find from 20-40% of false reproductions. The reproduction certainty is therefore in certain cases a measure for the emotivity of the test person.

By far the larger number of neurotics show a pronounced tendency to cover up their intimate affairs in impenetrable darkness, even from the doctor, so that the doctor finds it very difficult to form a proper picture of the patient's psychology. In such cases I am greatly assisted by the association experiment. When the experiment is finished I first look over the general course of the reaction times. I see a great many very prolonged times which in itself means that the patient can only adjust himself with difficulty, that his psychological functions proceed with marked internal frictions, with resistances. By far the greater number of neurotics react only under great and hence very distinct resistances, there are, however, cases in which the average reaction times are as short as in the normal and in whom the other complex indicators are lacking, but who, despite that fact, undoubtedly present neurotic symptoms. These rare cases are especially found among very intelligent [p. 239] and educated chronic patients who after many years of practice have learned to control their outward behavior and therefore outwardly display very little if anything of their neuroses. On superficial observation they can be taken as normal, yet in some places they show disturbances which betray the repressed complex. 

After examining the reaction times I turn my attention to the type of the association to ascertain with what type I am dealing. If it is a predicate type I draw the conclusions which I have detailed above; if it is a complex type I try to ascertain the nature of the complex. With the necessary experience one can readily emancipate himself from the test person's statements and almost without any previous knowledge of the test persons it is possible under certain circumstances to read the most intimate complexes from the results of the experiment. I at first look for the reproduction words and put them together, and I then look for the stimulus words which show the greatest disturbances. In many cases a mere assortment of these words suffices to show the complex. In some cases it is necessary to put a question here and there. It will be best to illustrate this with a concrete example:

It concerns an educated woman of 30 years who has been married for three years. After her marriage she suffers from episodic excitements in which she is violently jealous of her husband. The marriage is a happy one in every other respect and it should be noted that the husband gives no cause for the jealousy. The patient is sure that she loves him and that her excited states are groundless. She cannot imagine whence these excited states originate, and feels quite perplexed over them. It is to be noted that the patient is a catholic and has been brought up religiously, while her husband is a protestant. This difference of religion did not admittedly play any part. A more thorough anamnesis showed the existence of an extreme prudishness. Thus, for example, no one was allowed to talk in the patient's presence about her sister's childbirth, because the sexual moment suggested therein caused her the greatest excitement. She always undressed in the adjoining room and never in her husband's presence, etc. At the age of 27 she was supposed to have had no idea how children were born. The associations gave the results shown in the accompanying chart.

The blue columns represent failures of reproductions, the green ones represent repetitions of stimulus words, and the yellow columns show those associations in which the patient either laughed or made mistakes, using many instead of one word. The height of the columns represent the length of the reaction time.

[p. 240] The stimulus words characterized by marked disturbances are the following: yellow, to pray, to separate, to marry, to quarrel, old, family, happiness, false, fear, to kiss, bride, to choose, contented. The strongest disturbances are found in the following stimulus words: to pray, to marry, happiness, false, fear, and contented. These words, therefore, seemingly strike the complex above all. The conclusion that can be drawn from this is that she is not indifferent to the fact that her husband is a protestant, for she again thinks of praying, that there is something wrong with marriage, that she is false, that is, she entertains fancies of faithlessness, she is afraid (of the husband? of the future?), she is not contented with her choice (to choose) and she thinks of separation. The patient therefore has a separation complex for she is very discontented with her married life. When I told her this result she was affected and at first attempted to deny it, then to mince over it, but finally she admitted everything I said and added still more. She reproduced a large number of fancies of faithlessness, reproaches against her husband, etc. Her prudishness and jealousy were merely a projection of her own sexual wishes on her husband. Because she was faithless in her fancies and did not admit it to herself she was jealous of her husband.

It is impossible in a lecture to give a review of all the possible uses of the association experiment. I must content myself with having demonstrated to you at least some of its chief uses.