PSYCHOLOGICAL MEASUREMENTS. 569 



Can} Some of the desired number combinations were always passed 

 over in the material which was covered. The best records in this 

 respect are those for Gar, September 29; Spe, October 27; and Bra, 

 February 2, when in each case there were two errors of this sort. Lon, 

 Sne, Tho, and Wil, all of Squad B, made individual records with as 

 small a number as 3 errors of omission. The low-diet averages for 

 Squad A show in number of combinations correctly marked a range 

 from 52.9^ in the case of Moy to 98.3 for Gul. The averages for combi- 

 nations missed range from 6.0 for Bro to 44.2 for Gul. The latter 

 usually missed many of the combinations; on January 12 his omissions 

 were 82. The low-diet average for the group of 10 men is 73.2 combina- 

 tions checked, 17.5 combinations missed; the number wrongly checked 

 is negligible. 



Squad B (see table 152) for the average of normal experiments show 

 a range for combinations checked of 31.8 to 64.8 with Mac and Van, 

 respectively. For combinations missed they show an average of 5.5 

 to 30.6^, with Tho and How, respectively. Their normal group averages 

 are 52.3 and 14.0. The low-diet group averages of the same squad 

 are 67.4 and 12.7, indicating improvement during the reduced diet. 



For purposes of comparison, it is desirable to state the results in one 

 figure which will stand as a combination of speed and accuracy. The 

 subjects worked with a time limit, and equal stress was laid on check- 

 ing the largest number of combinations, and also on going over the 

 material without missing any combinations. The latter was fre- 

 quently called to the subject's attention first to give it prominence; 

 still it is likely that the matter of speed was more prominent in the 

 thoughts of the subject. In computing the combination value, which 

 is given in tables 153 and 154, a credit of -fl was allowed for each correct 

 number combination found and checked. A demerit of —0.5 was 

 subtracted for every error of omission and for every wrong combina- 

 tion checked. On this basis the scores range from 6.5 {Spe, September 

 29) to 113 {Vea, February 2) with Squad A and from —6.5 (Sch, Jan- 

 uary 19) to 99 {Van, January 27) in the case of Squad B. In Squad 

 A of the 10 men for whom there are complete records the best number- 

 cancellation scores were made by Vea, Pea, Gul, and Bro. These 

 low-diet averages, as seen in the lower line of table 153, are respec- 

 tively: 88.8, 78.7, 76.2 and 73.8. With the exception of Gar, all of 

 the subjects improved on their first low-diet date, although Moy and 

 Tom did but very little better on October 13 than they had on Septem- 

 ber 29. Throughout the whole experiment each individual improved 



* Can frequently complained of his eyea during these tests. Part of the time he wore an eye- 

 shield during the group experiment. He complained that the light was too strong and 

 he frequently said in the evening that he had a headache. He was asked repeatedly 

 to have his eyes re-examined by a good specialist. 



' Spe, who did not have so much practice as the others, had a lower average, i. e., 36.2. 



' The data for Sch are too fragmentary for comparison. 



