46 ROBERT CHEN AU LT GIVLER sudden descent. This also holds true in all the groups but the last, as appeared from the long Tiren graph. A comparison of the next two, Unrin and Unreen, showed less difference in the general motor discharge aroused by these two graphs than was evident in the case of the former two. This might be due to a number of things; first, the difference in the structure of the unaccented syllables in these pairs : -un may determine the motor supply as much as the -rin or the -reen. But the differences in the accented vowels are also to be taken into account, for in the one pair, short “e” and long “e” alternated, while in the other, short “i” alternates with long “e.” Hence we have two variables, and not one to deal with. Rinaz and Thira produced the strongest effects of the Rinaz- Rinad-Thira group and they were nearly equal in height and very similar in form; Rinad and Tira were exactly identical in form, but not so close together as were the other two. Evidently “th” and “z” gave the impetus to the responses, and the open “a.” was in each case provocative of restraint in the tapping, for the open “a” experiments took longer time to utter than those which closed with “z” or “d.” Rinad was found by the subjects to be a rather poor stimulus, while they attributed to Rinaz a sort of hypnotic or lulling character ; yet the graphs show that the latter of these sound-combinations was more arousing than the other. But as a general thing indifferent states were correlated Mean- wise with a greater motor output than were the pleasant. But Rinaz was the more pleasant of these two. Comparing Niral and Nidal with one another it appeared that the “r” as an initial accented consonant has a greater motor effect than does initial “d.” And yet the “d” can be given a much more explosive vocal character than the “r.” But the long “i” in Niral must not be forgotten. Nemal showed very well, espe¬ cially in the fourth group, the insistent character which was attributed to it in the introspection. Comparing the lengths of these graphs does not seem to throw any light on the matter of correlation, for while the long “i” in Niral might be construed as that factor which gives the length to this graph, yet Nerol is