TRANSACTIONS OF SUB-SECTION I. 685 
of an Argument (modified from Ebbinghaus) involved the comprehension of a 
series of inferences. 
Further tests were added, dealing with non-reasoning elements in the reason- 
ing tests, such as reading, writing, and free association. 
In all twenty-five tests were used for the main group and were marked vari- 
ously for speed, accuracy, and amount. Introspections were obtained from 
children and trained adults to determine how far the tests actually involved the 
mental processes they were designed to represent; and the conclusions were 
further checked by experimental variations in the procedure and material. 
A larger number of children were tested than in the former investigation, 
viz., in the main group, sixty-five boys and thirty-one girls. The children 
belonged to the same form, and the age range was one year instead of two. 
The calculations for the boys and girls were kept apart. With the help of other 
investigators the more important tests were further applied to groups of 
different social status and of various ages. 
The results obtained from the various groups are consistent. In the main 
research the reliability coefficients, with few exceptions, range from -70 to -88. 
The correlations with intelligence for the better attested tests of reasoning 
(Opposites, Analogies, Syllogisms, Argument) provide the highest correlations, 
ranging from ‘47 to ‘74. Considering the number and size of the groups and 
the reliability of the tests and of masters’ estimates of intelligence, these 
correlations seem to be as satisfactory as any yet obtained. 
On drawing up hierarchies and employing the method of multiple correlation 
and other criteria it seems clear that the correlations are determined chiefly by 
the presence of elements essential to reasoning. On eliminating the influence of 
reading and writing by means of the appropriate formule, the correlations with 
intelligence remain, in the case of the more complex reasoning tests, but little 
changed. On correlating the size of the intelligence coefficients for the several 
tests with the degree to which (according to the children’s own introspections) 
reasoning entered into the performances, the coefficient is found to be ‘89. 
3. Additional Reasoning Tests suitable for the Mental Diagnosis of 
School Children. By W. H. Wincu, M.A. 
The following set of tests is offered as additional to those presented to the 
British Psychological Society on May 6, 1911 :— 
I.—Value and Uses of the Tests. 
They are believed to be of service, because,— 
(1) They help us to measure what are usually called the ‘higher mental 
functions.’ 
(2) Psychological tests of these functions are at present very few in number, 
and have not been ‘ standardised’ by usage. 
(3) They, with other similar tests, help Psychology to meet the complaint 
of the educationist that our science measures only the trivial units of mental 
work involved in elementary sensory and motor functions. 
(4) They are completely extra-scolaire, that is, are not taught in schools, and 
are therefore a good measure of the natural ability of the pupil as distinguished 
from the pedagogical proficiency of the school. 
(5) Whilst they are too difficult to adequately measure ‘ Mental Deficiency,’ 
they serve excellently well to indicate ‘Subnormality,’ or what is usually known 
as ‘ Backwardness.’ 
(6) They will help teachers to grade their pupils on a basis of natural 
ability. 
(7) With other similar tests, they will enable us to measure the ‘transfer’ 
effects (if any) of so-called rational studies, like Euclid (Demonstrative Geome- 
try), Grammar, and Problematic Arithmetic. 
(8) They may enable us to discover the mental differences (if any) between 
boys and girls of school age in an important mental function. 
(9) They will help us to select ‘Scholarship Children’ without our selection 
