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



All of these defects are characteristic of an imperfect, incom- 

 plete, partial method. It may be said, I think, without qualification 

 that numerical results of the same kind as those obtained by McDou- 

 gall and by the performers of Whipple's 1910-test may be got by 

 fractionating the results of the method of constant stimuli. Here is 

 an illustration from my own laboratory. 



Determination of the Two-point Limen by the Method of Constant 



Stimuli. 



80 series, 400 observations. Stimuli 0, 6, 12, 18, 24 mm. 

 (A) Results Arranged in Groups of 10 Series. 



(B) Results Arranged in Groups of 20 Series. 



(C) Results Arranged in Groups of 40 Series. 



(D) Result of 80 Series. 



The cases of inversion (4) are printed in italics. 



We notice that, as the size of the group increases, the range de- 

 creases and the distribution of judgments grows increasingly con- 

 stant ; that is what we should expect. We notice also, however, 



