808 TRANSACTIONS OF SECTION L. 



fifteen children eight years old who had to go back, excluding one in whose case 

 the tests were not completed, only three in the final reckoning are to be adjudged 

 backward. 



I have found cases in which girls could satisfy the tests of a superior level, 

 but were unable to satisfy the tests of their own age or of the age immediately 

 preceding it. M. Binet was prepared for such cases, and the child who presents 

 this characteristic is estimated as of the intellectual level of the year, at which 

 she fulfils the required number of tests. 



10. On Testing Intelligence in Children. By Professor Dr. Ernst 



Meumann. 



The outcome of the author's reflections can be shortly expressed in the 

 following sentences : — 



(1) The test methods cannot be dispensed with because of their great practical 

 value. 



(2) To be excluded are all tests of acquired knowledge, especially the 

 examination of school knowledge and use of school work, because some of them 

 may be lacking in any normal child. Because of this it becomes impossible to 

 determine the normal standard of intelligence which would be of general validity 

 for each life year and for all children from any Milieu. But this is the principal 

 demand for a general comparing research into the normal child and its develop- 

 ment. 



(3) Therefore the tests have to be reduced to merely functional examinations. 



(4) It is impossible to measure the general intelligence with one or a few 

 psycho-analytical tests and very difficult and uncertain with one or only a few 

 practical tests. It is therefore better to give up the thought of being able to 

 determine the general intelligence by tests methods ; instead of these we seek to 

 examine the higher intelligence, which apart from the testing of a few elementary 

 functions, principally sensibility and sensibility of difference for some senses, 

 is to be considered as the working-up of impressions or representations by 

 synthetical and combinatory thinking. 



(5) This is best obtained by all the methods which examine the working with 

 abstract elements, the working with leading representations, the working with 

 the solution of combinations which are accustomed, easy and rich in content, and 

 the combining of them with combinations of new representations and the connec- 

 tion of these to the leading representations. 



11. The Pitfalls of 'Mental Tests.' By Charles S. Myers, M.A., M.D. 



A protest was entered against the collection of vast quantities of psycho- 

 logical data, especially by an army of untrained observers. Nothing can be more 

 dangerous than the supposition that the consequent errors cancel one another 

 'in the long run.' 



In physical anthropometry, despite the standardisation of measurements, 

 considerable differences occur when practised observers measure the same 

 individual. What must be the degree of divergence when mental characters 

 are _ measured by untrained observers, who not only improperly use and read 

 their 'instrument,' but affect in different ways the attitude of their subjects 

 towards the test ! 



Within any given community the individual variations in physical, and no 

 doubt also in mental, characters are so wide that the average of any measurement 

 must differ very widely from the average of that measurement in another com- 

 munity, for the difference between the averages to be with certainty significant. 

 Indeed, it must be large enough to be apparent to the unprejudiced eye. Thus 

 the statistical treatment of racial mental characters does not discover, so much 

 as measure, racial differences. Accuracy is therefore essential. 



The statistician who aims at collecting pyschological data in large numbers 

 is apt to neglect the various influences which, in different degrees, affect different 

 subjects in the tests, and to pour all data from whatever source into the statistical 



