408 



G. F. SPRAGUE 



Payne and Hayes stated that: 



The extent of relationship between the performance of F2 test crosses and of the per- 

 formance of their F3 progenies in test crosses leads the writers to conclude that in these 

 studies there was some doubt of the practical value of early testing for combining ability 

 as a means of selecting desirable sources of F3 lines. By a test however of relatively few 

 F3 lines it was possible to select F3 lines that seemed to be a desirable source for improving, 

 or substitution for certain inbred lines in Minhybrid 608. 



It may be well to emphasize again that the only claim made for early test- 

 ing was that it enables the separation of a population into two groups on the 

 basis of combining ability. Also, continued selection in the more desirable 

 group will yield a larger number of high combining lines than will the less 

 desirable group or a random sample of lines selected solely on the basis of 



TABLE 26.3 



FREQUENCY DISTRIBUTIONS OF YIELD IN BUSHELS PER ACRE FOR 



1 TO 3 F3 PROGENIES DERIVED FROM F. LINES OF A116 X L317 CROSSED 



WITH 4 DIFFERENT TESTERS (AFTER PAYNE AND HAYES, 1949) 



phenotype. The frequency distributions of test cross combining ability for F2 

 and Fg progenies seem to fulfill this claim very nicely. In the table that fol- 

 lows, each F2 distribution has been divided into the higher yielding 50 per 

 cent and the lower yielding 50 per cent. The distribution of F3 test crosses for 

 each of these subgroups was taken from their paper. The results are pre- 

 sented in Table 26.3. 



The writer would conclude from these distributions that the testing of F2 

 would have been a desirable practice. Within each test cross series it would 

 have permitted of the discarding of a considerable number of lines. If the 

 number of F.s's to be tested had been held constant and all of the lines to be 

 tested derived from the higher yielding F2 subgroup, even greater progress 

 might well have been expected. 



The results obtained in this study are exactly those to be expected under 

 the postulates of early testing. Early testing obviously cannot be used as a 



