EARLY TESTING AND RECURRENT SELECTION 403 



(1935) which seemed to lend considerable support to the general ideas men- 

 tioned above. Remnant seed of 14 lines from the variety Lancaster and 14 

 lines from the variety lodent representing eight generations of selfing were 

 chosen in Jenkins' study. These 28 lines represented a random sam[)le of the 

 lines from these two varieties which had survived the eight generations of 

 inbreeding. Two sibs were chosen to represent each generation, one repre- 

 senting a selected ear in the direct line of descent and the second a discarded 

 sib. These 56 ears were grown ear-to-row, and pollen from 10-12 plants of 

 each line were mixed and applied to ten ear shoots of the tester variety Krug. 



Due to variation in stands and the unfavorable season neither the sam- 

 pling of plants within a strain nor the topcross parent was as adequate as 

 planned. Only 12 of the lines originally chosen were represented in each of the 

 eight generations of selfing. The yield trials of the topcrossed progeny were 

 grown in 1932. Information on several important problems is presented in 

 this paper, but the items of most importance in the present discussion deal 

 with the performance of the lines after successive generations of selfing. In 

 the lodent series, represented by seven lines, the mean square associated 

 with generations was not significant. In the Lancaster series, represented 

 by five lines, the variation associated with generations was significant but 

 there was no indication of a consistent trend. 



On the basis of these results Jenkins concluded that, "The inbred lines 

 acquired their individuality as parents of top crosses very early in the 

 inbreeding process and remained relatively stable thereafter." Since this 

 paper was published, several people have assumed that the stability men- 

 tioned by Jenkins was synonymous with homozygosity, and therefore experi- 

 ments demonstrating segregation in F2 or F3 were disproof of this stability. 

 However Jenkins took particular pains to point out that the stability he was 

 assuming did not arise from homozygosity, but was a sampling phenomenon. 

 This sampling stability, if confirmed, makes the early testing procedure 

 even more attractive, but stability is neither assumed nor required as a pre- 

 requisite for early testing. 



Results from Early Testing 

 Experiments on early testing were started in Missouri in 1935. However 

 due to unfavorable seasons, no critical data were obtained until 1938. The 

 experiments were continued in Iowa in 1939 and subsequent years. The re- 

 sults of these studies were summarized in 1946 (Sprague, 1946). Some 167 

 selected So plants from a strain known as Stiff Stalk Synthetic were self pol- 

 linated and outcrossed to the double cross tester parent Iowa 13. The yields 

 of the test crosses ranged from 61.8 to 100.8 bushels per acre. Four of the 

 test crosses were significantly lower yielding than the synthetic parent, and 

 two were significantly higher yielding than the tester parent. The plants 

 chosen for selfing represented a carefully selected group on the basis of 



