When the subjects tried to recall and record as many as possible of the 25 monosyllable English words which were read for the memory-span test, nothing was said about what would be done in the case of words entered which were not pronounced. In a few instances, several words having nearly the same sound were entered, showing that the subject evidently remembered the general sound complex of the word, but was not absolutely sure as to its identity. In these special cases the extra words entered were not counted as errors. In tables 157 and 158, opposite the different dates and under the different subjects, will be found the number of words correctly entered and also the number of wrong words given. From the general average figure in the lower right-hand corner of tables 157 and 158 it will be seen that the usual score is a fraction over 8 words correctly remembered and recorded from the list of 25 pronounced, and on the average 1 wrong word is entered. The average performance is very nearly the same with both squads. With these figures in mind, the comparison of the individual subjects in their memory span shows for Squad A that the average number of correct words ranged from 5.9 to 12.1 (Mon and Tom) and the average number of errors from 3.0 to 0.4 (Mon and Bro). There were many individual records which ranged about 8 or above with no errors. The highest of these was Tom of January 26, with 14 correct words and no errors. The largest number of errors was made by Mon on November 24; there were 4 correct words and 7 wrong ones entered at that time. There is a fair amount of uniformity in the records of the individual subjects. A subject may show a variation of 2 on either side of his average number of correct words.

1It should be noted that this increase in the percentage of error does not occur on January 5, when there is a drop in the curve above, and is not due to the absence of any one man, since the whole personnel of the squad was present on this and on the other low-diet dates.

Table 157. - Squad A - Memory Span For 4-Letter English Words

Table 158. - Squad B - Memory Span For 4-Letter English Words

Squad B gives about the same range of memory span as Squad A. The averages of the 5 normal sessions range from 5.8 to 11.2 for words correctly remembered. In these 5 experiments the subjects had not received so much practice in the test as had Squad A, and the fact that the average is slightly below that for the latter group can not be regarded as at all significant. The number of errors was somewhat less with Squad B. Two of the subjects, Tho and Mac, made no error in any trial. This was a very good record, considering that in the case of Mac the number of words correctly remembered was above the average of these squads. Lon made a very good memory record; his average for correct words for the normal period is 11.0 and for the reduced-diet period 12.7. Only 2 errors were made in the 7 sessions in which he served - one on October 6 and one on January 27. Another subject, Ham, should be mentioned in this connection as having made only 2 errors, both during the normal period, 1 each on October 6 and December 15.

The data seem homogeneous, and no violence appears to be done by converting a subject's record, as shown in the number of words correctly recalled and the number of errors made into one score. This has been done very arbitrarily by deducting the number of errors from the number of words correctly remembered. The scores thus computed are given in tables 159 and 160. This simplification makes no especial change in the standing of the subjects in the groups. The average low-diet scores range from 2.9 to 11.3 for Mon and Tom, respectively. Bro and Moy also have large averages, viz, 9.3 and 8.9. The average score for all subjects in all sessions during the low diet is 7.3. Squad B shows a range of variability from 3.0 to 12.3. This group is led by Lon with a score for normal of 10.8 and a low-diet average of 12.3. The general averages for the group of 10 individuals in all sessions is 7.2 for the normal and 7.6 for the low diet. With the exception of Liv, Wil, and Fis, all the 10 regular subjects show better scores for the low-diet period than during the normal sessions, that is, the improvement continued during the reduction in diet.

The relation between the scores of the two groups and the food reduction period is brought out best in figure 106. Both curves show a tendency to improvement; Squad B is somewhat above A at the beginning and in four of the other sessions. In the fourth session they were at about the same level; in the seventh and eighth sessions Squad B dropped below. The fluctuations are rather large with both squads. Squad B shows a falling off on November 17 and December 15, but good results on January 5, i. c, just after the vacation, a date that shows poorly in several measurements. On January 13, their first low-diet session, they have the best average of the whole experiment in marked contrast to low-diet dates January 19 and January 27. Squad A demonstrates a large amount of variability in the early part of the investigation. The second session, as with B, showed great improvement over the first and the two squads came nearer to the same level. The average for the next three sessions is low. From December 8 to the end of the experiment, A showed consistent improvement, with relatively small changes from session to session.

Table 159. - Squad A - Memory Score

Table 160. - Squad B - Memory Score

The reduced diet can not be considered as favoring better scores than normal in the memory-span experiment. The decrease in the last two sessions with B and the poor scores in the third and fifth sessions for A seem in the direction of a low-diet effect. On the other hand, B, with no assigned reason, except that they did not receive so frequent practice as A, showed a marked decrease in the third and fourth sessions. Had they at these points kept up their level of November 3 and January 5, Squad A would by contrast have made a poor record in the early part of the experiment, and B would have appeared to decline in the second and third experiments with food reduction. Our conclusions from the memory-span results can not be positive on account of the apparently large normal fluctuations in practice. It seems safest to believe that the reduced diet produced no certain change.

Fig. 106. - Average scores for memory span.