664 THE AMERICAN NATURALIST [Vol. LI 



a particular lambing, or the basis of the best lambing she 

 shows. From a genetic standpoint the second criterion 

 would seem the better, but practical breeders would be 

 very likely to use the first. Unfortunately, records in 

 Southdowns on which a comparison can be based are few, 

 forty-three animals only being available. Table I pre- 

 sents the correlation of each individual record with the 

 average lambing record for each ewe, while Table II pre- 

 sents the correlation of the best record for each ewe with 

 her average. Nine of the ewes had four lambings to their 

 credit, nine had three, while twenty-five had only two. 

 The inadequacy of these data is recognized, since there is 

 a false agreement between a single number and its average 

 with another as compared to its agreement with its average 

 with several numbers. Since, however, the material is 

 suggestive from a comparative standpoint, it is presented, 

 as the same actual error exists in each table: 



TABLE I 



The coeffiei 

 + 0.02099. 



The coefficient of correlation for Table II was found to bo 

 0.92354 + 0.01513. 



