332 . THE AMERICAN NATURALIST [Vol. L 



reliance can be placed on the categories adopted. But 

 the figures given indicate that Hoshino has estimated too 

 high the coupling strength, and that more probably it 

 does net exceed 5 : 1. This is not due to any inaccuracy in 

 Hoshino 's calculations, but would follow only if the 

 hypothesis suggested in this paper is substituted for 

 Hoshino *s hypothesis. 



If I have correctly interpreted Hoshino 's observations, 

 flowering time in peas is clearly a Mendelian unit-char- 

 acter, entirely devoid of dominance, so that a strictly 

 intermediate hybrid form is the commonest end-product 

 of a single cross between early and late varieties. Fur- 

 ther, segregation is imperfect so that blending results, 

 which becomes more and more complete with each genera- 

 tion of inbreeding. From the incompleteness of the blend- 

 ing in the F x zygote and so the imperfection of the segre- 

 gation in the Fj gametes, it follows that many different 

 types of F 2 zygotes are produced, some of which are 

 practically constant (homozygous) particularly those at 

 either extreme of the series (the "early constant," 

 "pseudo-early constant," "late constant" and "pseudo- 

 late constant") and also at two intermediate points ("in- 

 termediate early" and "intermediate late"). 



Other F 2 zygotes, resulting from the union of gametes 

 quite dissimilar, produce a highly variable F 3 progeny, 

 but one which will give rise to F 4 families individually 

 less variable for two reasons: (1) because the process 

 of blending continues and so gametes produced by the 

 same zygote become more like each other than were the 

 parent gametes of that zygote, and (2) because hetero- 

 zygotes under self-fertilization tend to produce about 

 50 per cent, of homozygous offspring, while homozygotes 

 produce only homozygous offspring. 



The entire population therefore will in accordance with 

 recognized Mendelian principles gradually resolve itself 

 into relatively constant self -fertilizing lines. But because 

 of the slow but continuous blending which occurs, these 

 pure lines will in a very few generations form a complete 



