560 



Journal of Agricultural Research 



Vol. XX, No. 7 



The variety and complexity of the relations which may be back of a 

 correlation are well illustrated in this case. Thus, the weight at weaning 

 (33 days of age) should be correlated with the birth weight and with the 

 gain between birth and weaning simply because it is their sum. The 

 relations of birth weight with gestation period and the prenatal rate of 

 growth are also essentially mathematical rather than causal. Birth 

 weight is necessarily fully determined by the character of the prenatal 

 growth curve and the time at which this is interrupted by birth. 



In the relation between gestation period and size of litter we come to 

 a case in which there is no necessary mathematical relationship. We 

 naturally attempt to account for the high negative correlation by the 

 hypothesis that a large number in a litter in some way causes early 



&cr/n 

 O-JJdoys 



fVe/afyfat 



*i 



Rate, of 

 Growf/7 



(Sesfdf/on 

 Pe>r/od 



Fxternal 

 conditions 



Conc//'f/on 

 of Dasn 



//erecf/'fq 

 of Panr 



Fig. 1. — Diagram illustrating the interrelations among the factors which determine the weight of guinea 

 pigs at birth and at weaning (33 days). 



parturition. Similarly, a large number in a litter might be expected to 

 be a cause of slow growth in the foetuses. 



Birth weight and gain after birth are highly correlated. Here neither 

 variable can be spoken of as the cause of variation in the other, and the 

 relation is not mathematical. They are evidently influenced by common 

 causes, among which heredity, size of litter, and conditions which affect 

 the health of the dam up to the time of birth at once come to mind. 



Most of the variables are connected with each other through more than 

 one path. Thus, weight at birth is correlated with weight at weaning 

 both as a component of a sum and as the effect of common causes. 



There may be a conflict of the paths. Thus, a large number in a litter 

 has a fairly direct tendency to shorten the gestation period, but this is 

 probably balanced in part by its tendency to reduce the rate of growth 

 of the foetuses, slow growth permitting a longer gestation period. Large 

 litters tend to reduce gestation period and rate of growth before and 

 after birth. But large litters are themselves most apt to come when 



