406 G. F. SPRAGUE 



ard lines in yield, and in resistance to root lodging and stalk breaking. On the 

 basis of these results it was suggested that early testing might be a valuable 

 tool in a breeding program. However it was pointed out that the method 

 might be of limited value under some conditions. This warning has to some 

 extent been ignored and some have assumed that the early testing procedure 

 is useful at any stage of the breeding program and with any parental ma- 

 terial. 



Additional trials of the early testing procedure have been conducted by 

 Dr. John Lonnquist (1950) of the Nebraska Station. In this experiment a 

 series of selected plants from a strain of Krug were self pollinated and out- 

 crossed to a series of plants of the same variety. When test cross performance 

 data were available two samplings were made. One consisted of the group of 

 lines exhibiting the highest topcross yields and the second group those 

 exhibiting the lowest topcross yields. 



In each group in subsequent generations selection was practiced in both 

 directions. In the high group the phenotypically most desirable and least 

 desirable plants were self pollinated and outcrossed to the tester. In the low 

 group again the most and least desirable plants were selfed and outcrossed. 

 This plan had to be modified somewhat as inbreeding progressed, since seed 

 was not always obtained on the least desirable plants. The group actually 

 used were the least desirable plants which could be propagated. After each 

 test cross generation the selection of lines to be continued was based on 

 combining ability. The single cross WF9XM14 was substituted for Krug 

 as the tester parent after the original series of test crosses. 



The results obtained during the first four selfed generations clearly indi- 

 cate that topcross combining ability can be readily modified by a combination 

 of selection and testing (Fig. 26.2). In the high group selected for high 

 combining ability, the average topcross yields of all lines represented in- 

 creased from 98.6 to 107.5 bushels. In the high group selected for low com- 

 bining ability after the Si yields decreased from 98.6 bushels to 93.3 bushels. 

 In the low group selected for high combining ability after the Si generation 

 yield increased from 85.9 to 94.0. Where selection was practiced for low 

 combining ability in each generation, yields decreased from85.9to77.9bushels. 



Thus selecting for high combining ability for three additional generations 

 when the original lines exhibited poor combining ability produced S4 lines 

 which were not significantly different in combining ability from those of the 

 high group selected for a similar period for poor combining ability. Selection 

 in the low group therefore would be largely wasted effort. Continued selec- 

 tion and testing after the Si would be most profitable for only those lines 

 exhibiting the highest Si topcross combining ability. 



Limitations of Early Testing 

 Three papers have been published which are somewhat critical of the value 

 of early testing. These will be reviewed briefly. Payne and Hayes (1949) 



