TV] STATISTICAL STUDY 43 



were transmitted, while Pearson, in his later writings 

 at least, regards the law simply as a statistical 

 description of what is found when large numbers 

 are observed in mass. 



It has been mentioned that the characters which 

 especially lend themselves to statistical treatment 

 are those which vary continuously and which can be 

 accurately measured, but Prof. Pearson has applied 

 similar methods to discontinuous characters, which 

 can be classified into groups but not measured, for 

 example coat-colour in horses. He finds as the 

 results of his enquiries that the inheritance of such 

 characters can be stated in terms similar to those 

 obtained with measurable characters, so that the 

 principle of ancestral correlation leading up to the 

 law of ancestral heredity may be applied to these 

 characters also. But whatever may be the case with 

 characters which vary continuously, it will be seen 

 below that discontinuous characters are commonly 

 alternative in their inheritance, i.e. there is no 

 blending, but the ofispring exhibit one or other only ; 

 and in some at least of these cases, the character of 

 the offspring cannot be calculated with any more 

 exactness if the ancestry is known than if it is not. 

 Such instances show clearly that although the law 

 may be statistically true when applied to considerable 

 populations, it gives us no clue to the physiological 



