making psychological examinations of young children so that those who can do approximately the same amount of work can be grouped together.

The Relation between Morality and Intellect. CLARA F. CHASSELL, Teachers College, Columbia University.

Purpose of the Study:

To discover whether the findings of investigators as to the correlation between desirable qualities, studied particularly in the field of mental abilities, hold also when the relation between morality and intellect is investigated.

Method and Sources of Data:

1. A summary of previous statistical and experimental studies. 2. A study by the ranking method of approximately 1,000 students in 28 colleges and universities, judged by from one to six members of the faculty in the respective institutions, in the traits "morality in the broadest sense," "intellect as shown in studies," and "intellect as shown in activities other than studies."

3. A study by the same method of more than 200 of these students, judged by a number of their classmates, in the traits "unselfishness," "loyalty to school and friends," "justice to all," courage in support of convictions," "self-control," "activity for social welfare," "reliability," "intellect as shown in studies," and "intellect as shown in activities other than studies."

4. A supplementary study of the relation between faculty and student judgments thus obtained, and college grades.

5. A determination of the correlation between scores in conduct scales and general intelligence tests in the case of more than 100 elementary school children.

Results:

1. The range for the correlations between morality and intelligence as found by various investigators, with a single element as the measure of morality, extended from .23 to .79. The average correlation was .48.

2. The range of certain correlations found for these traits, with two or more elements as the measures of morality, extended from .21 to .83, with a median of .51.

Thus the correlations in the investigation point to a relationship between morality and intellect when the study is confined to restricted groups, of approximately .50, and justify the conclusion that, in this field also, correlation and not compensation is the rule.

The Relation of Intelligence to Age in Negro Children. ADA HART ARLITT, Bryn Mawr College.

This investigation was undertaken to determine whether such differences as have been shown to exist between negro and white intelligence are constant from age to age.

The subjects were 243 negro children, 180 of whom were selected at random from the Playground and the public and private schools of New Orleans and 63 of whom came from the schools in the neighborhood of Philadelphia. All of the New Orleans and 36 of the Pennsylvania negroes were tested by the writer. The remaining 27 were examined by four Graduate students. In the latter case the replies were recorded as far as possible verbatim and were graded in consultation with the writer. The tests used were the Stanford Revision of the Binet Tests.

Of the total group 129 were boys and 114 were girls. At ages five and six combined there were 28 boys and 26 girls, at age seven 25 boys and 25 girls, at age eight 25 girls and 25 boys, at age nine 25 boys and 24 girls and at ages ten to fifteen combined 26 boys and 14 girls. As to social status 33.7 per cent. were very inferior, 58 per cent. inferior, 6.5 per cent. were average and 1.6 per cent. very superior. There was an approximately equal proportion of children from all social groups, except the very superior, at each age.

The median intelligence quotient for negro children does not remain constant from age group to age group as it has been shown to do for native- born whites. Both the median I.Q. and the 25 and 75 percentiles show a progressive decrease with increasing age. At ages five and six combined the median I.Q. is 100, at age seven 90.9, at age eight 87.5, at age nine 83.9 and at ages ten to fifteen combined 78.9. Compared with Terman's I.Q. of 93 for nativeborn whites of inferior social status to which this group is most nearly comparable, the median I.Q. for negroes is above average at ages five and six combined and falls below average at every age beyond these, decreasing steadily with increasing age. Certain peculiarities in the tests account in part for this phenomenon.

As to sex differences girls are superior to boys by 6 points at ages five and six combined. Beyond these ages there is no significant difference as to median I.Q. There are more girls than boys with I.Q.'s above 110 and more boys than girls with I.Q.'s below 70.

Experimental Critique of the Foundation of Tests for Ethical Discrimination. AUGUSTA F. BRONNER, Judge Baker Foundation.

The recent considerable interest in ethical discrimination tests calls for a critical attitude concerning their foundations. On the basis of known conduct tendencies, can we find such tests diagnostic? A method for such critical attitude might be evolved by using well-known material, such as the fables test in the Stanford-Binet Series. Comparing responses of delinquents and non-delinquents, what does one find? There is a necessity, of course, for comparison according to mentality and chronological age-levels. There is opportunity for supplementing such a comparative study by noting important specific individual reactions and their value for prognosis.

Individual Differences: Studied by Means of a Practice Experiment. GEORGINA S. GATES, Barnard College.

Occasionally investigators find a negative or zero correlation between "desirable mental traits." In studies of the school subjects, certain tests have been found to have a high inter-correlation in one school or locality, and a low inter-correlation in another.

That the cause of apparent inconsistencies may be inequality of previous practice is suggested by an experiment in which twentythree individuals repeated five tests approximately twenty-five times each.

Among the results obtained were the following:

1. Inter-correlations between tests increased as long as improvement in the functions continued.

2. Correction for attenuation did not do away with the increase. 3. The very best records in the entire series showed in spite of their presumable unreliability a relatively high inter-correlation.

4. The relation between improvement made by the same individual in different tests or at different periods in the same test was only such as might occur by chance.

The conclusion suggested by the first three facts is that where measures of ultimate capacity only are considered, inter-correlations are relatively high and individual differences seem to be increased.

Intelligence Irregularity as Measured by Scattering in the Binet Scale. J. E. W. WALLIN, Miami University.

[The abstract is omitted through an error arising in the Secretary's Office.]

How Is a Science of Social Psychology Possible? J. R. KANTOR, Indiana University.

Although it is becoming increasingly recognized that the data derived from the observation of group activities ought to be brought under scientific control and evaluation, psychology is today poorly equipped to handle such data. Why is it that psychology lacks the technique to interpret social behavior, whether of the single individual or of a number of individuals considered as a particular group? This deficiency is rooted in the physiological origin of current psychology, according to which psychological phenomena must be interpreted exclusively in terms of properties of biological organisms. This physiological psychology has failed utterly in its aims to interpret social phenomena, a failure which we may sum up in three hopeless results.

(1) Because psychologists were working with physiological materials they could not handle such responses as language, and the attitudes and thoughts marking off individuals of different groups. (2) Psychologists, not really finding any facts in physiological organisms by which to interpret complex reactions, have resorted to putative powers called instincts, which are nothing but names and not factors in social phenomena. (3) Again, because physiological conceptions are so fruitless for the description of social phenomena, psychologists have been forced to develop noxious conceptions of a universal superconsciousness or group mind.

In view of this situation it appears imperative to bring social psychology back to a concrete stimulus-response basis, for only on such a foundation may we hope to achieve a scientific interpretation of social phenomena. Accordingly, in our study of social behavior we find that just as in individual psychology we study the reactions of definite individuals to stimuli in the form of persons, objects, acts, events and conditions. With this difference, however, that in social behavior the stimulus objects are conventional in the sense that all the members of the same group have acquired common cultural reactions to them.

Behavior and Group Psychology. A. A. ROBACK, Northeastern College.

This paper deals with the question whether behaviorism can be applied in the social field of disciplines as a substitute for the psychology of the group in a mentalistic sense. Many recent writers on group phenomena make use of behavioristic terms without really being aware of the issues between behaviorism and traditional psychology, and in consequence their terms are out of gear with the point of view they start from. Prima facie it would seem that it is more desirable to speak of crowd behavior than of a crowd consciousness, especially as a group cannot be said to possess an introspective consciousness. But a similar objection holds against the behavior approach, for it is impossible to tell what a crowd or in fact any group is doing at a given moment except in the roundest, hence unscientific, terms. The conception of a group mind does in fact unite the manifestations of a group under a unitary head, and in this way it affords a synthetic approach to a synthetic subject. The concept of group behavior, if treated in any significant sense of the word, can but give an analytic basis for our study, which, from the nature of the case, defeats its own end.

The suggestion to abandon the study of the group as such and to concentrate all the energy upon the observation of the individual as influenced by the group involves not only a circular mode of reasoning but is fatal to any collective psychology. Knowledge about the individual will give us no more information about the group than the knowledge of the properties of oxygen and hydrogen would acquaint us with the power and appearance of Niagara Falls.

Group psychology furthermore tries to reconstruct the Greek or Roman mind from the writings of representative Greeks and Romans and the traditions of the two nations as carried over from generation to generation. For the behaviorist, all that has been written by and about the Greeks comes down to innervations and contractions of certain muscles, and even that must be inferred, for in reality what he has before him is a series of marks occasioned by a number of agents, from the original writer down to the pressfeeder. In what respect is the copyist's work to be regarded as inferior to the author's, since the action of the musculature is not essentially different in the two cases? The whole branch of collective psychology, and with it most of the social sciences, must fall with the acceptance of an orthodox behaviorism.

Are There Any Sensations? R. M. OGDEN, Cornell University.

The question refers both to the existence and to the systematic utility of the so-called "element" of sensation. The difficulty of isolation makes the element less real to observation than other units of perception and a phenomenological analysis brings to light