160 H. H. NEWMAN 



their mothers than males are. In view of the fact, however, that 

 the difference between the coefficients of uniparental correlation 

 for males and females is considerably less than the probable 

 error of the determination it may well be that the difference is 

 unreal. 



4. EVIDENCES OF ALTERNATIVE INHERITANCE OF AGGREGATES OF 



MERISTIC VARIATES 



The statistical method, if applied to the exclusion of the method 

 of dealing with individual cases, is apt to cover up some of the 

 most significant facts and to reveal only partial truths. A study 

 of the individual cases given in table 1 indicates strongly that sets 

 of quadruplets or individuals of a set tend to resemble one or 

 other parent to the exclusion of the other in the matter of the total 

 numbers of scutes in the banded region. Complete identity be- 

 tween individual offspring and the mother are common, too com- 

 mon to be mere matters of coincidence. Many cases exist in 

 which the four quadruplets fluctuate only very slightly about the 

 condition of the mother; while an equally large number of cases 

 are found in which none of the individuals of a set are at all close 

 to the mother in scute counts. These sets presumably fluctuate 

 about the scute count of the fathers. A third class of cases exists 

 in which one, two or three fetuses may be said to inherit the 

 maternal scute number, while the remainder presumably inherit 

 that of the fathers. From this it may be concluded that the germ 

 plasm of the mother is in some cases prepotent or dominant for the 

 entire set with respect to the factors that determine the numbers 

 of scutes, in other cases that of the father dominates in the same 

 fashion. It is not difficult either to understand that, since every 

 cell of the body is a hybrid product, maternal influences might 

 gain more or less complete ascendency in some fetuses of a set, 

 and might be subservient to the paternal influences in others, so 

 that there might be a very real segregation within sets of quad- 

 ruplets of maternal and paternal characteristics. An attempt 

 might be made to classify the 115 sets of fetuses into three classes 

 according to whether they show pure maternal, pure paternal or 



