The results are presented in tables 1 to 5. The dates in the first column indicate the day on which each individual set of ten reactions was taken. The second column presents the averages of each ten reactions in sigmas. The third column gives the mean variations of the various single reaction times from the average. The fourth column gives the variations of the day’s average, recorded in column 2, from the general average of that individual for all his experiments. These general averages are presented in table 6. At the bottom of each complete column of averages will be found the group average. This quantity is obtained by averaging the results from all the members of the squad for periods of, approximately, two weeks. The mean variations in the third column make it clear that the subjects did not settle down into what could be regarded as trained subjects. Trained subjects are expected to give mean variations which are consistently within the limits of 10 per cent of the average. While there are, of course, instances in which the mean variation falls within this limit, there is no consistent exhibition of the regularity indicative of thorough practice. This fact is further confirmed by a comparison of the results of March 30, 31, and April 1 with those of the earlier months. At the end of March, all effects of practice in November, December, and January, except the most general, may be regarded as having disappeared; and yet the averages and variations for the March and April dates resemble closely those of the month immediately preceding and also those of November. The effects of practice may, accordingly, be regarded as insignificant. This lack of special training accounts for the large variations which appear in some cases. As is usual in reaction experiments, the signal to which the reactors were to respond was in each case preceded by about two seconds by a bell signal to arouse attention. Conditions were thus rendered as nearly uniform as possible, but the variations indicate in three or four cases exceptional lapses of attention. Such exceptional cases can be eliminated without prejudicing the final validity of the results by substituting the median for the averages. In table 7, the medians are grouped together and show even more than the tables of average the absence of any general variation during the period of the tests. The obvious conclusion from these tests is that the quickness of the members of the squad underwent no general change during the whole period covered by the test. Individual members showed variations from time to time, but these variations are clearly accidental in character, for they show no regular tendencies and are in no way related to the changes in the character and amount of the diet.