SELECTION 327 



in F, hybrids in terms of dominance and linkage (p. 231, 2). The fact 

 that different genotypes give diverse results when crossed is of immense 

 practical significance. 



The Ear-to-row Method. This has been the method of commercial 

 corn improvement for many years and it is well illustrated by the Illinois 

 corn breeding experiments, which have been going on continuously for 

 over 20 years. The original purpose of the experiments was to produce 

 new strains which would be more valuable as a source of feed for livestock. 

 It was found that there was considerable variation in the relative amounts 

 of protein and carbohydrates in the grains of different ears. Accordingly 

 selection was begun with the object of increasing the protein and reducing 

 the starch content of the grains; also of decreasing protein and increasing 

 starch. As oil was worth three times as much as starch per unit of weight, 

 selection for higher oil content was also begun. A low oil strain was 

 started for comparison and such corn was soon found to be desirable for 

 the production of pork and beef of high quality. 



The work was begun by Hopkins who picked out 163 ears of a local 

 strain known as Burr's White, made a chemical analysis of a few grains 

 from each ear, and on that basis sorted them into four classes, viz., high 

 and low protein and high and low oil. The strains were grown in isolated 

 plots from the beginning. After 9 years of selection it was found to be 

 necessary to prevent inbreeding. Accordingly in the tenth and succeed- 

 ing years about 24 ears were selected for each plot and one row was 

 planted from each ear, then the even numbered rows were detasseled. 

 Subsequent selections were .made from the detasseled rows, the first 

 consideration always being high yield. Usually 20 ears were taken from 

 each of the six higher yielding rows, or 120 ears for each plot. These 

 were tested by chemical analyses and the most extreme variants in the 

 desired directions were selected for the next planting. 



The results in general have been more regular in the high and low oil 

 series than in the high and low protein strains. In the latter there seems 

 to have been no very decided effect of selection after the first 10 years. 

 Similarly there has been no continuous advance in the low oil strain since 

 the seventeenth year of selection, but in the high strain the per cent, of oil 

 has continued to increase slightly. The progressive effects of selection in 

 the four series are graphically illustrated in Figs. 133 and 134. That the 

 striking results depicted in these graphs were not caused by environ- 

 mental conditions was proved by planting mixed plots with two grains of 

 "high" and two of "low" corn in each hill so arranged that the resulting 

 plants could be identified. This test, according to L. H. Smith, was made 

 for three successive years, and subsequent analyses showed that under 

 these conditions the different strains maintained their distinguishing 

 chemical characters. 



