Since at the most the normal data with Squad A are limited to those obtained in one session, and therefore it will be necessary to refer to other data in discussing the results with these men, it will be most convenient to present the results for both squads simultaneously. The data for any measurements will be tabulated for each squad separately, but when plotted the same figure will commonly contain curves for both squads. These curves may thus be compared directly according to the number of the experiment for each squad. For example, the data for the third experiment for each squad will be plotted on the same ordinate but as Squad A had 10 sessions and Squad B only 8 the curve for the latter will be 2 points short. An effort has been made to keep the tabular presentations and the figures uniform for all the measurements so that the detailed descriptions will not have to be repeated.

Because of circumstances over which we had no control, there were certain changes in the personnel of the two groups of men. This is regrettable, particularly for the psychological measurements. Individual variations in metabolism and allied physiological processes are not so large as in these more complex functions, involving considerable learning, with which we here have to deal. All things considered, it must be counted as rare good fortune that the changes in personnel were not more frequent, and, indeed, that it was possible to hold together for so long a time 10 men in each of the two groups of 12. Statistically, 10 men is not a very large group, but practically, and considering the complexity of the research, the sacrifices that it entailed and its long duration, together with the unstable conditions in all college-student bodies in 1917-18, it is remarkable that the study was finished with so many of the original subjects still serving. A group of 10 cooperative trustworthy subjects must, all things considered, be regarded as an exceedingly satisfactory number on which to base a result in such observations. Since this is the case, it has been deemed best not to complicate the average results with the more fragmentary records from those subjects who did not serve throughout the period. It would be very difficult to equate the results for individual differences when one subject dropped out and another came into the group. Therefore, unless otherwise specified, the average for any measurement with Squad A will be the average of the individual results for the following ten subjects: Bro, Can, Gar, Gul, Mon, Moy, Pea, Pec, Tom, and Vea. The values for Kon, Spe, and Fre will be given when they exist and may be compared, but are not included in the group average. The average result for Squad B will be the average of the individual records for the following 10 subjects: Fis, Har, How, Ham, Lon, Liv, Sne, Tho, Van, and Wil. The individual values, when available, will be given for M.cM, Kim, Mac, and Sch. It will be of interest to note occasionally how far the fragmentary records of the other subjects conform to the general findings for the group.

Although the amount of time devoted to any one measurement on an individual in an evening or morning session was rather limited, the apparatus and procedure had been arranged with a view to taking several observations of that particular kind, so that their average would be a fairly good sample of the individual's performance in that process or function. In illustration of this it will be seen as explained in the section on program and technique (p. 137) that in the measurements of the eye reactions, the word reactions, the eye movements, and the electrical threshold, particular attention had been paid to this matter. It was desired that the number of observations of a particular kind on one subject should be large enough so that some statement statistically significant regarding their consistency and uniformity might be given. When practicable, the standard deviation and coefficient of variability1 for the individual subject and for each experiment are included in the tables, and these individual variability measures are averaged as are the other results.

We are glad to take this opportunity of acknowledging the faithful services of Miss Anna Berlin and Mr. Edward S. Mills in reading the records and in many of the computations for the large amount of psychological data.

The date on which each neuro-muscular measurement was given for the first time to the members of the two squads can be found in the chronological record of the whole experiment (p. 60). It will be further observed that for Squad B the first and second sessions (October 6 and November 3) and the third and fourth (November 17 and December 15) were in each case separated by 4 weeks. Between the fourth and fifth sessions (December 15 and January 5) there was an interval of 3 weeks. The others were separated by 2 weeks or less. All of the night observations were on Saturdays except those on January 13 and 27, which were on Sundays. January 5 was an unfavorable date experi-mentally, as the men came to the Nutrition Laboratory immediately from their Christmas vacations. Starting with January 8, the diet of each man in Squad B was reduced to 1,375 net calories per day, and the men had had a total of this amount in the three meals preceding all evening psychological sessions.1

1 The standard deviation was computed by the usual formula: - S.D. or a = In order to compare the variability in the results for one measurement with that in a series of observations of another process, it is desirable to know the relative sue of the two variability measures in terms of per cent of their respective averages or other central tendency measures. In the following tables the coefficient of variability, C - -jp is employed, in which S.D. is the standard deviation and M the arithmetical average. (See Whipple, Manual of Mental and Physical Tests, Part I, Baltimore, 1914, p. 24).