750 TRANSACTIONS OF SECTION L. 
19—much lower than was expected. The pupils were given a week’s practice 
in this work, and again tested. The coefficient of correlation proved to be °54. 
After still another week’s practice the coefficient rose to ‘69. Such results 
prompt us to ask whether tests of intelligence should not give prominence to 
proficiency in the acquirement of a habit rather than to ability manifested in 
giving immediate answers to a questionnaire. 
12. The Mental Differences between the Sexes. By Cyriu Burr. 
Problems.—Attempts to summarise secondary sex-differences commonly reduce 
them to terms of a single principle—e.g. natural and sexual selection during the 
evolution of the race (Darwin), abridgment or prolongation of immaturity during 
the development of the individual (Spencer), environmental influences accumulat- 
ing through inheritance (Lombroso), environmental influences reimposed but not 
inherited (Mill), or a fundamental antithesis in physiological constitution 
(Geddes and Thomson). The implications of such theories suggest the follow- 
ing problems as crucial issues between them: Are the inborn differences between 
the sexes as large as the acquired? Are the mental differences as large as the 
physical? Are the differences on the higher levels of mental life as large as 
those on the lower? 
Methods.—These problems require for their solution familiar devices of 
experimental and statistical procedure. 
In collecting data it is necessary to separate inborn mental differences from 
those that are acquired. In the earlier of the experiments here reported,’ there- 
fore, the subjects chosen were children of the same age, school, and class, and 
the tests employed were tasks which depended upon previous knowledge or train- 
ing to a measurable or negligible extent; in later experiments the groups tested 
represented various ages from five to twenty-five, and differed in social status 
and school-training; and the experiments included tests of acquired knowledge 
and interests. 
In calculating results it is necessary to measure the sex-differences in the 
various tests in the same terms. They are, therefore, expressed in terms of the 
amount by which the percentage of males exceeding the central measure of the 
females deviates from 50 per cent. A positive sign indicates that the central 
measure of the females is surpassed by a majority of the males; a negative sign 
indicates an analogous superiority on the part of the females. 
Results.—It is found that where the two sexes have been taught in different 
schools, departments, or classes, the differences are much larger than when they 
have been taught in mixed classes. In the latter case the differences revealed 
persist throughout the various ages and social classes measured with but little 
increase or diminution. In the former case they tend to be larger in the case 
of older subjects and to vary considerably in different social strata. 
None of the inborn mental differences are as large as the physical (stature, 
weight, height, + 47 to + 50 per cent.). The largest inborn differences are 
found on the lowest levels, in simple processes of sense-perception and movement 
(e.g. skin discrimination, — 43 per cent.; tapping, + 38 per cent.). Instincts 
and emotions show differences which appear to be somewhat smaller, but are 
far-reaching in their consequences. On higher levels the differences become 
progressively still smaller (e.g., memory — 37 per cent., reading — 22 per cent., 
writing — 20 per cent., addition + 16 per cent., multiplication + 13 per cent.). 
On the highest levels of all, those concerned with reasoning, these sex-differences 
are insignificant (e.g., finding opposites — 8 per cent., argument + 7 per cent., 
syllogisms + 5 per cent.). In general, the higher the process tested and the 
more complex the capacity involved, the smaller the innate sex-differences tend 
to become. 
Conclusions.—With three of the above principles of explanation (develop- 
mental, transmissionist, constitutional), taken singly and alone, these results 
seem incompatible. _ The distribution of the differences suggests a composite 
hypothesis. The larger and more obvious differences seem to be acquired under 
1 A fuller account of the earlier experiments has been published in The Journal 
of Experimental Pedagogy, 1911, 1912. 
