RESEMBLANCE IN INTELLIGENCE 517 



tion would be — 1, which would represent complete 

 disagreement. A correlation of would mean no more 

 resemblance than we should get if we paired off any 

 father with anyone else's son at random. Correlations 

 between and 1 would therefore represent all degrees 

 of positive resemblance from very slight to practically 

 perfect. We are now ready to examine the correlation 

 coefficients of resemblance, which have actually been 

 found for related individuals in the case of intelUgence. 



Family Resemblance in Intelligence. — The co- 

 efficient of resemblance between parents and children in 

 intelligence seems to be in the neighborhood of .50, which 

 is about half way between perfect resemblance and no 

 resemblance at all. It is a striking fact that the parent- 

 child resemblance in many physical traits also turns out 

 to be rather uniformly in the neighborhood of .50. This 

 would suggest that mental and physical traits are in- 

 herited to much the same degree. 



The correlations between sibs (i.e., children of the 

 same parents) have also been found to approximate .50, 

 both for physical and mental traits. This holds for 

 spelling and arithmetical ability as measured by standard 

 tests, and for ability in other school subjects as measured 

 by school marks. Ratings of 2000 school children by 

 their teachers on such traits as intelligence, vivacity, 

 conscientiousness, popularity, temper, self-consciousness 

 and assertiveness gave correlations between sibs approxi- 

 mating .50 for all of these traits. - 



In Germany school marks earned by several hundred 

 children were compared with the school marks earned 

 by their parents and grandparents. Correlation coeffi- 

 cients were not computed, but the results shown in Table 

 1 are significant. The marks were given on a scale of 

 five, 1 being highest and 5 lowest. 



