But we are beginning to realise with increasing clearness that affective tone is the dominant factor in all mental activity; complexes owe their power and their very existence solely to its operation; its distribution, so to speak, is the all-important determinant of the mental state of the individual. Consequently any sound scheme of classification must, in the last analysis, be based upon the affective tone concomitant to the reactions concerned and affective considerations must over-ride all others of a formal and academic nature.

Before proceeding to the interpretation of these results I wish to enlarge for a moment upon the concept of 'positive' and 'negative' affective tone which I have introduced into these studies. I do not propose to discuss them exhaustively here but I feel that it will be wise to consolidate my position and to guard against possible criticism by recalling the terms in which I defined the words1.

It is important that the distinction drawn between the two kinds of affective tone should be a valid distinction and truly relevant to mental processes as they actually occur; also that the criterion chosen for establishing the presence of each kind of tone should be of a nature to effect such a valid discrimination.

It will be remembered that I defined negative tone as that variety which tends to repel attention, or to impede the accession to consciousness of the ideas to which it is concomitant; positive tone was defined as the opposite to this.

I think it will be conceded that the operation of negative tone, so defined, is clearly identical, in nature though not in intensity, with the process commonly known as 'repression.' The operation of positive tone is, of course, simply the reverse of this.

I identified the concomitance of these two varieties of tone as characteristic of certain classes of reaction by measuring quantitatively the effects of their operation; that is to say I actually measured the tendency for the stimulus words of the reactions concerned to have their accession to consciousness impeded-i.e. to be 'forgotten.'

Consequently, although I admitted that 'positive' and 'negative' tone might be considered as corresponding in many cases with 'pleasant' and 'unpleasant' respectively, the latter vague and unsatisfactory concepts play no part whatever in the application of my methods which are based solely on an empirical observation of the tendency for certain stimulus words to be driven from consciousness and of others to be attracted thereto.

1 Cf. "Some Properties of Complex Indicators," p. 281.

I submit that this purely empirical procedure yields results which are strictly relevant to mental processes as actually met with and, notably, to those varieties of them which are particularly studied by psychopathologists.

After this digression we may return to the consideration of the results recorded in Tables I-V.

I may as well say at the outset that I have doubts as to whether the study of 'reaction types' based upon any system of classifying reactions is likely to prove of great practical value apart from research work. But Jung and other authorities appear to consider it important and potentially valuable and it may prove to be so for certain purposes-e.g. diagnosis-but only in so far as we properly understand the significance of the different forms of reaction.

Inspection of Table II shows that the reaction classes may be divided into two main groups:

(i) Those which favour 'toned' reactions at the expense of 'untoned.' The principal numbers of this group are classes I, II (b), II (e), III, VII and (VIII).

(ii) Those which favour 'untoned' reactions at the expense of

'toned.' The chief examples here are II (a), II (d), IV and VI (a). Class V is rather indeterminate and conforms so closely to the probable figures that I shall not consider it further; classes II (c), VI (b), and VI (c) are too small to afford a reliable basis for discussion.

Of the classes comprising the first group all are incontestably 'inner' associations; in the second group classes IV and VI (a) are equally undoubtedly 'outer' associations and I have given reasons for holding that classes II (a) and II (d) should also be reckoned as 'outer.'

All this is in accordance with expectation; outer reactions are obviously of a more superficial type than inner, the stimulus word does not penetrate so deeply into the mind, so to speak, because a suitable reaction is easily found. This is rather a loose way of speaking; it would perhaps be more accurate to say that the subject follows the line of least resistance and gives the reaction which combines the maximum of accessibility with the minimum of negative tone. The more accessible, the more familiar, the more superficial an idea associated with the stimulus word is, the greater the chance of 'dodging' negative tone. Or, better, the accessible and familiar associated words are just those which, by virtue of the association having been formed in countless varying contexts, possess no specific tone.

This is well borne out by the figures for the principal classes. The

most superficial class of all is class VI (a), consisting of reactions conditioned by common phrases, antitheses, etc. Such reactions can take place with the minimum of attention to the true 'inwardness' of the stimulus word, they are as nearly as possible purely automatic. The actual number of neutral reactions in this class (160) is 132% of the probable number (122), while the negatively toned reactions only amount to 58 % of the probable number.

Class IV (Co-existence) is less superficial, the formation of such reactions requires rather more attention although no contemplation of the attributes of the object1 suggested by the stimulus word is necessary. The corresponding figures to those quoted above are: neutral reactions 122 %, negatively toned reactions 64 %.

Class II (a) (Simple Predicates) again demand for their formation somewhat closer attention still to the stimulus word, for the simple predicate is essentially an apprehended and named attribute. The figures here are 105% and 91 % respectively.

Class II (d) does not conform unless it be reckoned more superficial than II (a) which I think would be wrong-but it is so small compared to the other three that this observation is not surprising.

Similarly with the inner reactions we find that the tendency for negative tone to show itself is substantially proportional to the extent to which the reaction is personal and peculiar to the subject, or in other words to the degree of its 'innerness.'

Thus class I (Co-ordination) is comparatively superficial. The corresponding figures are, neutral reactions 89% of the probable number, negatively toned reactions 111%·

Class II (b) is clearly much more personal, consisting as it does of reactions containing an expression of personal opinion. The figures are 50 % and 174 % respectively.

Class VII is by definition the most intimately personal of all (cf. section 4 (c)) and accordingly we find that the figures are 30% and 268%

Class II (e) is rather small and not very easy to place: in my judgment it should probably be located between I and II (b).

Class VIII is much too small to give reliable figures in this connection; inasmuch as it contains a number of reactions taken from class VII and a few 'stereotypes,' it is highly personal, but is diluted to some extent by polyverbal reactions, which although significant are

1 N.B. Stimulus words giving rise to co-existence reactions are, of necessity, almost invariably concretes.

not quite so obviously peculiar to the individual subject as are the numbers of class VII; the corresponding figures are 36% and 200 %.

Thus we find, as we progress from class VI (a), the most superficial of all, to class VII the most peculiar, the most personal, the most truly inner, a steady increase in the numbers of negatively toned reactions and a steady decrease in the number of neutrally toned. These figures are shown in Table VI.

I contend that this alone is sufficient justification for the system of classification which I have adopted and if it be considered with the other evidence I have adduced the soundness of this system will, I think, be unmistakably apparent.

Anyone who has done any practical surveying will know what is meant by 'closing a traverse.' I start, let us say, from point A, I take observations and calculate the position of point B, thence I work to C, from C to D, from D to E and finally back again to A. If the position of A thus computed coincides with its known position from which I started I conclude that the intermediate measurements and calculations have been correctly made; it is an extraordinarily delicate check, as anyone who has tried it will admit.

A somewhat similar check can be applied to the investigations embodied in this paper and the two which preceded it1.

I started by showing that affective tone, as detected and measured by certain indicators, exerted an influence on the remembering of the words to which it was concomitant; I next used this fact as a means of differentiation between different combinations of these indicators and for determining their affective properties; finally I applied the results of this process of differentiation to the study of the relation between. 1 "Memory and Affective Tone," This Journal (General Section), Jan. 1921, and "Some Properties of Complex Indicators" on p. 281 of this issue of the Medical Section.

the various types of associations and the affective tone concomitant to them.

If these methods are valid and if the tendency I have found for negatively toned reactions to predominate among inner associations and neutral reactions among outer is a real tendency, we should expect the mean 'memory value' of learned words belonging to the former class to be smaller than that of words belonging to the latter. This would constitute a 'check back' on to my starting point.

I have accordingly computed the mean memory value for inner and outer associations (classified according to the system I have been advocating); the values are 6-37 and 6.86 respectively.

Thus we see that starting from Memory, proceeding to Complex Indicators, working from these to the Forms of Associations and finally back again to Memory, the results are uniformly concordant.

When we remember the many fortuitous causes which conspire to make the memory test insensitive and the considerable scope for error which there is in classifying the forms of associations, this 'closing of the traverse' can, I think, fairly be claimed as remarkably satisfactory evidence of the reliability of the methods used and the validity of the conclusions obtained.

Finally it is necessary to consider the bearing of these conclusions as to the form of association on the practical use of the association test. I have already said that I do not think it is likely to be very great, but it cannot be doubted that the more thoroughly the test is understood by those who use it and the more perfectly the relations between its various features are appreciated, the better the results obtained are likely to be. And I believe that it may well prove a very valuable weapon for purely research purposes.

The test is sometimes used as a preliminary to psycho-analysis; the physician applies it in order to gain some idea of the general mental type of the patient and some guide to his principal complexes. He is essentially on the look out for pointers which shall tell him where he may most profitably begin the detailed exploration of the patient's mind; he wishes to shorten his labours by selecting the most promising point de départ for the analysis. It may be doubted whether the method is as yet fully appreciated, but some psycho-analysts value it highly.

The success of the test and the amount of information to be gained from it must necessarily depend to a large extent on the experience of the physician. It is hardly a matter which can be reduced to a rigid formula; the conclusions drawn must rather result from a gradual