132 PSYCHOLOGICAL METHODS OF TESTING INTELLIGENCE 



to the special care and capacity for judging of the 

 teacher who did the estimating. 



I have made a further calculation for this Class 

 Vb by excluding those six pupils in whose cases there 

 existed, according to the teacher's notes, special con- 

 ditions. The correlation for the remaining 34 then 

 rose at once to 0.74 ; in other words, it approximated 

 very closely the lowest correlations computed for the 

 other classes. From this it follows for this particu- 

 lar class, and presumably as a general principle, too, 

 that the low correlation first secured is not due to 

 any distinct thorough-going discrepancy between de- 

 gree of intelligence and school efficiency, but rather 

 to an unusual discrepancy between endowment and 

 performance in a minority of the pupils. This small 

 group demands the special consideration of the 

 teacher and individual treatment, for it is with them 

 that the danger is greatest that the ordinary valua- 

 tion of the children in terms of their school work 

 may lead to an erroneous appraisement and han- 

 dling. 



Turning to the higher schools, I have at my dis- 

 posal now a single class only, but the estimation of 

 the intelligence of this class has special value on ac- 

 count of the great thoroughness and precautions of 

 method adopted, and on account of the fact that sev- 

 eral teachers joined in estimating the same children. 

 Table XVII exhibits the correlations that I have 

 computed. I have to thank the regular master of the 

 class for the material. 



The class was an Untertertia grade in a Gymna- 

 sium. 4 ' The regular teacher (Teacher A) was well 



4 This would correspond scholastically approximately to our first 

 high-school year. Translator. 



