5 2 A PURE-LINE METHOD IN CORN BREEDING 



gradual reduction of the strain to a homozygous condition; and (3) 

 that the object of the corn-breeder should not be to find the best pure- 

 line, but to find and maintain the best hybrid combination. 



The continuation of these studies during the past year have given 

 still further proof of the correctness of the first two of these propo- 

 sitions, and besides has given unexpected suggestions for a new method 

 of corn breeding by which the essential feature of the third proposition 

 may be realized. It is my purpose to discuss this new method briefly 

 in the following pages. I will first, however, describe the results of the 

 past year's experiments in so far as they bear upon the points in 

 which we are interested here. For convenience I will refer to the 

 two self-fertilized families contrasted in my paper last year as "Strain 

 A" and "Strain B." It will be remembered that these two families 

 resulted from the self-fertilization of different, apparently equal, indi- 

 viduals ; but that notwithstanding this fact, they differed from each 

 other in height and stockiness of stems, width and greenness of the 

 leaves, length of shank of the ears, appendages of the husks, quality of 

 the grains, and the number of rows of grains on the ears. (See fig, i.) 



In addition to the parallel cultures of self-fertilized and cross- 

 fertilized families which have been continued from the beginning of 

 these experiments in 1904, I had during the past season the F t offspring 

 of a cross between two sibs in Strain A, and two families representing 

 reciprocal crosses between Strain A and Strain B. It was observed 

 that every one of the mentioned characteristics which distinguished 

 Strains A and B, remained constant distinguishing features in the pure- 

 bred families, but in regard to the number of rows on the ears, it is 

 now obvious that Strain A has the normal mean number 8, as compared 

 with 14 in Strain B, for in this year 89 per cent of the ears produced by 

 Strain A had only 8 rows of grains, though the selection of ears for seed 

 in this strain during three years was for 12 rows on the ear, and only 

 in the last year was an 8-rowed ear used because a suitable 12-rowed 

 ear was not available. This result is a striking confirmation of the 

 suggestion made last year that according to the law of regression the 

 occurrence of a mean number of rows less than 12 in Strain A indicated 

 that the normal number of rows for this strain is 10 or possibly only 8. 



The cross between two sibs in Strain A was grown beside the self- 

 fertilized family belonging to the same strain, and these two families 

 were so similar during the entire period of their development that they 

 were considered identical, but at the end of the season it was found 

 that the cross-bred family was a trifle taller and produced over 30 per 



