236 



MUTATION AND PLANT BREEDING 



60 4 



E 



i 



IS 

 UJ 



5 



UJ 



2 



a. 



UJ 

 XL 



o 



o 

 o 



44 45 



1000-KERNEL 



48 

 1959 



46 47 



WEIGHT gm 



Figure 5. — Correlation of 1,000-kernel xveiglit between 1959 and I960 

 in 31 selected lines pom X-rayed winter barley with regression lines. 

 r = 0.797. 



screened. Again, differences between the lines are highly significant 

 (simple lattice, F value lines/intrablock error: 16.90; F value of the 

 table at 0.1 per cent level: 2.79). Also, the correlation between 1959 

 and 1960 is highly significant (r = 0.652, P > 0.1 per cent). The breed- 

 ers of Haisa II consider this variety to be a pure line within reason- 

 able limits (Vettel and Lein, personal communication). 



The selection experiments described are of a preliminary char- 

 acter. They were conducted to determine the feasibility and value of 

 initiating more exact and intensive experiments, and to gather tech- 

 nical experience. The results have been broached here because they 

 indicate at least that screening for increased kernel size is very simple, 

 if there is a corresponding genetic variability in the starting popula- 

 tion. Experiments are underway in our laboratory to obtain conclu- 

 sive information on the nature of these variants, as well as on their 

 general significance in breeding. 



Besides the experiments described, we started, several years ago, 

 a program of recurrent selection for kernel yield in irradiated proge- 



