26 



THE AMERICAN NATURALIST [Vol. XLIX 



mother, we find that it ranges from 2.15 in the case of 

 which had 5 English young, to 2.79 in the case of 

 $162), which had 14 English young. The average number 

 of English young to a mother is 14.4. 



After this series of matings had been completed, a 

 second series was begun in which the same 13 females 

 were mated with one of the darkest bucks produced in the 

 Series I matings (a son of $162?). The selected buck was 

 c?4CL4 (Fig. 6), grade 3.75, considerably darker than his 

 father (Fig. 5). This series of matings produced 189 

 English young, together with a like number of self (non- 

 English) young. The grade distribution of the English 

 young is shown in Table I, Series II. All of the 13 

 mothers except one (5162 7 ) produced darker offspring in 

 the Series II than in the Series I matings. The lowest 

 average grade was shown by the young of ?176r, viz., 2.44. 

 For Series I matings the lowest average was 2.15. The 

 highest average grade in the Series II matings was given 

 by the young of $16i?, viz., 3.50. For Series I matings 

 the highest average was 2.78. Consequently, both maxi- 

 mum and minimum averages were higher in the Series II 

 than in the Series I matings. The grand average of all 

 the 189 Series II offspring was 2.92 as compared with 2.43, 

 the average grade of the Series I young. Their modal 

 grade is 3.25. The modal grade for Series I was 2.00. 

 Since the mothers were identical in both series, the differ- 

 ence in the young can be attributed only to the difference 

 in the fathers. The male used in the Series II matings 

 differed genetically as well as somatically from his father, 

 who sired the Series I young. Not only was he darker, 

 but he also produced darker English young. Yet the 

 father contained only a single dose (one gamete) of Eng- 

 lish pattern and the son derived his English pattern ex- 

 clusively from this same source. Hence the English 

 unit-character had changed quantitatively in transmission 

 from father to son. This seems to us conclusive evidence 

 against the idea of unit-character constancy, or "gametic 

 purity." If unit-characters are not constant, selection 



