464 MEMOIRS NATIONAL ACADEMY OF SCIENCES. [v<k xv, 



As a possible aid to interpretation the fact should be noted that the two "poorest" groups 

 under consideration are both inferior to the standard group and the two "best" groups are both 

 superior to the standard. The fact, then, that tests 1 and 2 differentiate "poorest" Infantry 

 privates from unselected recruits most sharply means that "poorest" Infantry privates are 

 relatively lower in these two tests than in the other tests of alpha. Conversely, the fact that 1 

 and 2 differentiate "poorest" Field Artillery privates least sharply from unselected recruits 

 indicates that "poorest," Field Artillery, are relatively higher in these two tests than in the other 

 tests of alpha. Moreover, if a selection of "poorest" Field Artillery privates results in a group 

 differing insignificantly from unselected recruits, as is the case when considered as a group, the 

 conclusion is obvious and necessary that the entire group of Field Artillery is superior to the group 

 of unselected recruits, and is, therefore, a selected group. Hence, we are dealing with a second- 

 ary selection 1 when considering these groups of "poorest" and "best" privates. The shift of 

 emphasis from tests 6 and 7 to 1 and to a lesser degree to 2 in passing from "best," Infantry, v. 

 standard to "best," Field Artillery, v. standard, confirms this interpretation. Further, either 

 the primary selection has produced qualitatively different groups, Infantry and Field Artillery, 

 or the secondary selection, giving the two sets of "best" and "poorest" groups, has been made 

 according to standards of military value varying from Infantry qualitatively to Field Artillery. 

 In the latter case the qualitative difference due to secondary selection follows a primary selection 

 that may be qualitatively the same for Infantry and Field Artillery — i. e., merely quantitative. 



The results for test 8 are peculiar. In the differentiations for Infantry of tables 129, 130, 

 131, and 132 it retains the same rank, but in the Field Artillery it differentiates the "poorest" 

 from the standard and the "best" from the "poorest" most sharply, and the "best" from the 

 standard only to a moderate degree. These facts suggest that the primary selection of the sampled 

 Artillery regiments has resulted in a group not measuring up in test 8 to the standard set in other 

 tests, but that general information, which test 8 tests, has counted heavily in the secondary 

 selections. It would be interesting to know more of the variations in the distribution of scores 

 for test 8 for different parts of the country, since a number of the items of this test are probably 

 much better known in some parts of the country than in others. It may happen, therefore, that 

 the men in the Artillery regiments in Camp Travis have come largely from parts of the country 

 where the information required for scoring high on test 8 is rather unusual. 



The patterns of differentiation for the noncommissioned officers of the Infantry and Field 

 Artillery regiments tend to confirm some of our previous conclusions and also to complicate the 

 situation further. We note again the evidence of primary selection of artillerymen of a sort that 

 correlates with test 1. When the differentiation pattern for noncommissioned officers, Field 

 Artillery, v. standard is compared with noncommissioned officers, Infantry, v. standard, the 

 contrast between these two types of differentiation occurs in the ranks of tests 3 and 8. The 

 inference made above with respect to test 8 that the primary selection in the Artillery was not 

 highly correlated with knowledge of miscellaneous facts, is supported here. When Field Artil- 

 lery noncommissioned officers are compared with "poorest," Field Artillery, this inference 

 (that general information was an important factor in the secondary selection) is further confirmed. 

 Further evidence of the qualitative differences in the selection types under discussion is given by 

 the consistently high rank of test 3 in all the intra-Inf antry comparisons and its medium or low 

 rank in the intra-Artillery comparison. 



It is perhaps worthy of notice that tests 3 and 8 differentiate Infantry noncommissioned 

 officers most sharply from the standard group, and Field Artillery noncommissioned officers 

 least sharply (excepting only test 5). Yet paradoxically these tests differentiate least clearly 

 Infantry noncommissioned officers from Artillery noncommissioned officers. This apparently 

 again is evidence of considerable difference in the standards to which men in the two arms of 

 service have had to measure up in order to be granted warrants as noncommissioned officers. 

 The fact that the two groups of noncommissioned officers are most alike as regards tests 3 

 and 8 is probably not significant of any special relationship between performance in these 



1 The fundamental selection of men by principal groups (Infantry, Artillery, standard group) may be referred to as a primary selection; a 

 selection within one of these groups of the "best" or "poorest" men, say, is then a secondary selection. 



