E. B. TITCHENER— ETHNOLOGICAL TESTS OF SENSATION. 215 



that the 8 series of (A) are of the order of test-series. If we 

 ascribe them hypothetically to 8 different subjects, we may fix the 

 8-in-io Hmen of the last five as 21, 18, 15, 12 and 24 mm. respectively 

 (range 12-24) I ^o limen can be calculated for the first three. Even, 

 then, if the basis of the test is enlarged, and we take 50 judgments 

 from every one of our supposed 8 subjects; and even if the basis is 

 regularized, and we test all the subjects by the same stimuli; even so, 

 the limen shows a wide range and proves in certain cases to be in- 

 determinable. It is plainly not enough, in this test, to secure eight 

 out of ten right answers by a rapid procedure applied " under the 

 same conditions" to a number of unpractised subjects. For the 

 conditions during such a test are only by chance the same for dif- 

 ferent persons ; the probability is that they are diverse ; and the repe- 

 tition of the test on the same subject may change the original finding. 

 From this point of view, also, McDougall's results are open to grave 

 suspicion. 



It is fatally easy for the field-worker and the laboratory-worker 

 to misunderstand each other. So I had better say at once, and em- 

 phatically, that I do not want to see the refinements of the home- 

 laboratory carried into the field. When Galton suggested that " in 

 testing the delicacy of the various senses ... we should do wrong if 

 we pursued the strict methods appropriate to psychophysical investi- 

 gations, "^"^ I take him to have been heartily in the right. I criticize 

 McDougall's combination-method on the formal ground that it is 

 not a method, whether fine or coarse ; it is, so far as I can see, essen- 

 tially the rough equivalent of a section or fragment of the method 

 of constant stimuli; and as a mere fragment of a method it can lead 

 us nowhere. We are fortunate in having the method of constant 

 stimuli in the background, to give us numerical norms ; but we can- 

 not use a piece of a method as if it were the whole. 



II. Color Vision 



I have chosen, as a second example, Rivers's work upon color 

 vision. This is a very painstaking bit of research, and its conclusion 



I*' F. Galton, "On the Anthropometrical Laboratory," etc., Jonrn. Anthrop. 

 Inst., February, 1885, p. 4 of ofifprint. 



