222 



SCIENCE 



[N. S. Vol. XXXVIII. No. 972 



pear. The quicksand begins at the nest step; 

 that of constructing tests which shall have a 

 useful meaning. Individual differences in the 

 tapping test are exquisitely clear through 

 many aspects of the experiment; but what 

 these individual differences represent in the 

 personality of the subject we do not know. 

 The problem of mental tests is duplex; to con- 

 struct a test at once free from physical and 

 physiological inaccuracies, and one that shall 

 have a useful significance for the subject's 

 adaptation to life. Without the first the 

 second is unattainable ; without the second the 

 first is futile. 



The questions of interpretation must not be 

 taken too lightly. Psychological experiments 

 of the present class must consistently repre- 

 sent those mental properties of the individual 

 that it is desired to compare, properties such 

 as it is useful to know about. The value of 

 mental tests depends upon their correlation 

 with the personality of the subject; and the 

 essential task in the scientific development of 

 any mental test is to determine how well it 

 indicates some phase of the subject's per- 

 sonality. 



Because it is much easier to do, we have 

 been apt to develop handy psychological 

 methods and then try to make them mean 

 something, rather than to start from the things 

 that are important to know, and trying to 

 develop methods for determining them. But 

 to start with the tapping test as a measure of 

 voluntary motor ability, or with the A test as 

 a measure of rate of perception, is too obvi- 

 ously approaching the problem at the wrong end. 

 We must not be bound by the notion that one 

 test tests one thing, another test another; one 

 test usually tests several things, and it must 

 take several tests to test one thing well. 

 First must be known the direction our in- 

 quiries must take; a task whose extreme com- 

 plexity demands analytical and systematic 

 observation of human behavior, not to men- 

 tion insight. Then one may seek to develop 

 measures which shall be themselves reliable, 

 and shall show the most constant relation to 

 the elementary traits that are to be measured. 

 The test is never an ultima ratio. If we want 



to determine how good a test the average daily 

 wage is of the number of applicants for poor 

 relief, we must have other, most reliable infor- 

 mation of the number of such applicants. In 

 the same way, in order to know how far any 

 mental test is a reflection of personality, we 

 must have accurate knowledge, from other 

 experiential sources, of how this personality 

 compares with others in the phase we may be 

 testing. 



Mental faculties differ a great deal ia the 

 completeness with which they are experi- 

 mentally covered. Those mental tests are of 

 the best assured value where the use made a£ 

 the method is immediate so to speak in terms 

 of its own result. Thus we may interpret a 

 test of astigmatism in terms of its own result, 

 because it represents a nearly constant attri- 

 bute of the individual, unaffected by other 

 uncontrolled factors. A test of color-blindness 

 can be interpreted in terms of its own result, 

 to decide on the fitness of the subject for rail- 

 way or marine service. But this is not so 

 much the case with the strength tests, such as 

 are used in the gymnasiums. A persork may 

 test quite high on the dynamometers who can 

 not make nearly so efficient use of his strength, 

 or actually not be so strong, as one who can 

 not make so good a record with them. Prac- 

 tical life puts the eyes to the same test that 

 the Snellen types do, therefore they are a good 

 test; it does not put the muscles to the same 

 test that the dynamometers do, therefore they 

 are an inferior test. 



There is in our experimental literature a 

 happily growing tendency, as exemplified in 

 the work of Healy, G. G. Femald, Simpson, 

 and others, to submit the tests of the higher 

 mental processes to the test of concrete experi- 

 ence. The most prominent result of this Pra- 

 gestellung has been the series of graded tests. 

 We wish to be able to say that a child has in 

 certain ways the ability of an 8, 9 or 10 year 

 old. Therefore we determine what degree of 

 these abilities is actually characteristic of 8, 

 9 or 10 year old children. Just as, if we 

 wished for a test of honesty, we should try to 

 find some way in which persons known to be 

 honest differed from persons known to be dis- 



