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



" I believe," he writes, " that the smallness of the mean variation in 

 most of the quantitative investigations will convince those acquainted 

 with the procedure of experimental psychology of the trustworthi- 

 ness of the observations."^*'' But what is the procedure of experi- 

 mental psychology? A set of instructions, carefully formulated and 

 intelligently grasped ; an instrument of precision ; a large number of 

 observations, made in accordance with a prescribed method and suffi- 

 cient for mathematical treatment; a variation of conditions to throw 

 this or that aspect of the subject-matter into high relief; experiments 

 distributed regularly over months or years. Under such circum- 

 stances, truly, the m. v. may be an index of steadiness of attention 

 or of general attitude. Not by any means necessarily under other 

 circumstances! For a small m. v. may mean that the subject has 

 over-simplified his instructions, and is performing an easier task 

 than the task set him ; or that he has discovered some secondary 

 criterion of judgment, some shortcut to response, and is not perform- 

 ing his allotted task at all. Or a small m. v. may mean that the unit 

 of the instrument is too large, and that the performance of the task 

 is thus artificially regularized. In these cases uniformity of result 

 would spell laziness, or perverted ingenuity, or too gross a gradua- 

 tion of stimuli ; in other and more extreme cases it might be due to 

 fatigue or to motor habit. We cannot argue directly from labora- 

 tory-experiment to field-test. 



Secondly, however, we must make full use of the laboratory. I 

 suppose that most laboratories possess records of practice-work done 

 by undergraduate students according to the schemata of the principal 

 psychophysical methods ; I have quoted a record of this sort, a de- 

 termination of the two-point limen by the method of constant stimuli. 

 Such records are not worth publishing, but they are worth preserv- 

 ing.^" They furnish norms of the performance of comparatively un- 



56 R, 4. 



°^ It is important to preserve not only the numerical values of the limen 

 and of the measure of precision but also the rough data of the whole experi- 

 ment. If the students are supplied with two cross-section forms and a 

 carbon paper the duplicate may easily be obtained. 



The rough data of the field-tests should also be accessible ; all through 

 this paper I have felt the need of further detail. I should think that sales 

 to laboratories could be assured beforehand, enough to cover the cost of 

 mimeographing the complete records. 



