248 



THE AMERICAN NATURALIST [Vol. XLV 



All of these propositions have now been tested in a 

 limited way. In 1910 nine different self -fertilized fami- 

 lies were compared with nine crosses between sibs within 

 the same self-fertilized family; ten crosses between sibs 

 in F 1 families were compared with ten self-fertilizations 

 in the same F 1 families ; seven families were raised as 

 first generation hybrids between individuals belonging 

 to different self-fertilized families; and ten families 

 were grown, in which self-fertilization had been entirely 

 precluded during the past five years. The average height 

 of plants in decimeters, the average number of rows per 

 ear, and the average yield in bushels per acre, in these 



An examination of this table indicates to me that on 

 the whole hiy self-fertilized families are not yet quite 

 pure-bred; for the sib crosses give on the average a 

 slightly greater height, number of rows per ear, and 

 yield per acre than the corresponding self-fertilized fami- 

 lies, as shown by a comparison of the first two columns 

 of the table. The same fact is apparent from a com- 

 parison of the "Fj X self" and "F, X Sibs" columns, 

 except that in this case the heights and number of rows 

 per ear are essentially equal while the yield per acre is 

 significantly higher in the sib-crosses than in the self- 

 fertilized families. An alternative explanation of these 

 slight differences between the results of self-fertilization 

 and of sib-crosses may attribute them to an injurious 

 effect of self-fertilization, but in any event such injurious 

 effect must be exceedingly slight as compared with the 

 stimulating effect of heterozygosis. My practise of 

 choosing for seed the best available ears tends to delay 

 the attainment of complete genotypic purity, and this 

 fact favors the view that whatever advantages the sib- 



