Gregory: efficacy of mutation breeding 479 



made in crosses A and D. However, these tests were weak statis- 

 tically. It is interesting to observe that although tested in different 

 years and fields, the low selections tended to remain low and the 

 high selections to remain high. There were only 6 reversals of high 

 and low out of a total of the 23 pairs available from the four crosses. 



g Cross A Cross D 



% 



30 J 



20 J 

 10 



-io_ 



-20. 

 -30. 



Cross mean 





PI P2 PIX P2X F FX PI P2* PIX P2X F FX 



Legend 



*MISSING I \ ^4 AMONG F, PLANT PROG. 



F 2 FAM X (F 6 generation) 



Figure 4.— F 3 variation in yield of P, PX, F, and FX among F 5 plant 

 progenies (in F 6 generation) of high and low F 2 families (selected in 

 F^) from two crosses (A and D). The variation is shown as in Figure 1. 



Of the 17 pairs which held their high-low positions, 8 were PI, P2, 

 or F 2 and 9 were PIX, P2X, or F 2 X. 



Table 8 shows the extent to which the negative regression of 

 s G for F 2 generation progenies on Fi generation family mean has 

 disappeared in later generations. In two of the crosses, A and B, s G 

 was larger in the low selections, while in the other two s G was larg- 

 er in the high selections. This difference in the two pairs of 

 crosses appeared to be associated with the crosses irrespective of 

 the treatment. The s G was larger in the low selections both in PI, 

 P2, and F 2 and in PIX, P2X, and F 2 X in crosses A and B; while 



