52 W. G. SMITH : transmitting and registering apparatus was determined by a special set of experiments and that the figures given in the table are the corrected figures. It will be observed that with two of the subjects, B and C, the chronoscopic results are very similar to those gained by adding together the data in the two columns containing the graphic results. The difference in the case of A may probably be explained to some extent by the fact that in the experi- ments carried out by the graphic method the drum was driven at a high speed by an electric motor : this caused a considerable amount of noise, which no doubt distracted the attention : it is to be noted also that on the different days when I was able to secure reactions the subject was suffering from a certain amount of fatigue. In each case the mean variation of the interval between stimulus and rise of curve, st-r, is highest absolutely, though relatively the mean varia- tion of the antagonistic rise, r-ap, is the most pronounced. The percentage of instances in which the variation from the average bears the same algebraic sign for both these magni- tudes, st-r, and r-ap, is for each subject somewhat over 50 per cent, (on the average 55 per cent.). This suggests that there is a tendency in these magnitudes to rise and fall con- currently, but the tendency is in any case not a decided one. In the second table are presented the results gained from three individuals in whom the tendency to react in the an- tagonistic form is decidedly intermittent. The number of experiments recorded is somewhat small and there is no concurrent determination by the chronoscope. The columns have the same meaning as those in the former table, with the exception of that headed st-f : in this column is given the average length of the interval elapsing between stimulus and fall of the curve in those reactions which exemplify the ordinary form. 1 In this group sensorial and muscular re- actions are taken together. The one subject, D, whose results are sufficiently numerous to allow of a comparison of the two types, showed no material difference in length in the two types when they were separately estimated. 1 In the experiments of this group, in which the stimulus was the sound of an electric bell, the time registration was begun by the current which actuated the electro-magnet of the bell : the reaction times in the columns st-f and st-r are consequently longer than they should be by the latent time of movement of the bell hammer. Owing to accidental circumstances I was unable to determine the length of this latent period. Inasmuch as the problems discussed above do not involve an exact determination of the whole length of reaction time this error is of relatively little importance.
