SENSORY FARADIC THRESHOLD. 135 



grade of attention is of vital importance to satisfactory measurements 

 of the human threshold by the Martin method in untrained subjects. 



Of scarcely secondary importance seems to be the precise technique 

 by which the operator satisfies his scientific conscience that any given 

 position of the secondary coil of the inductorium corresponds to the 

 probable true threshold of the subject. As Grabfield and Martin 1 state, 

 "The threshold position found by moving the coil in, is often several 

 millimeters away from the threshold position moving it out." To 

 make their experimental conditions uniform, Grabfield and Martin dis- 

 carded the threshold readings which were found when the coil was 

 moving out 1 (p. 304). It is not impossible, however, that this very 

 discrepancy may serve as an indicator of the grade of attention, since 

 variations of attention seem to produce it, and there seems to be no 

 other ground for its existence. In our experience we have always found 

 successive threshold positions, even in the same direction, to vary more 

 or less. Our earliest measurements were based on the standard psycho- 

 physical method of averaging the threshold values found by increasing 

 and decreasing the stimulus respectively. It is commonly assumed in 

 psycho-physics that the true threshold lies between the apparent thresh- 

 old, which is found when a subthreshold stimulus is increased, and 

 that found by decreasing a suprathreshold stimulus. The difficulty 

 with this procedure in the present instance is probably due to a fatigue 

 of attention. A somewhat later procedure was to take the highest 

 value that was found three times out of five. This obviously produced 

 fatigue effects, since the first values were regularly higher than the 

 later ones. All values reported in this paper under dates subsequent 

 to January 1, 1914, were found by averaging the first three ingoing 

 threshold positions of the coil. Professor Martin kindly informed us 

 that his present procedure is to repeat the ingoing movements of the coil 

 until two thresholds agree. While this seems statistically somewhat 

 arbitrary, his results are much more regular than ours. 



Our variation of procedure should affect materially only the level of 

 measurements on different days. Differences between the successive 

 series on one day should still be comparable with the differences between 

 successive series on another day, even though the actual values are 

 somewhat higher or lower on the different days. Since our whole 

 statistical treatment is based on these serial differences rather than on 

 average levels, our variations in procedure, as regrettable as it was un- 

 avoidable in the present stage of experience with the Martin threshold, 

 are not vital to our main problem. 



A further difficulty connected with the use of the sensory threshold 

 for electrical stimulation is the nature of the sensation whose threshold 

 is measured. Probably it may safely be said that threshold induction 

 shocks are never felt as simple touch or pressure sensations. Martin, 

 Porter, and Nice 2 report an apparent difference in the sensations, and 



'Grabfield and Martin, Am. Journ. Physiol., 1912-13, 31, p. 300. 

 2 Martin, Porter and Nice, Psychol. Review, 1913, 20, p. 194. 



