RECURRENT SELECTION AND OVERDOMINANCE 455 



tests if higher levels of specific combinability could be accumulated by recur- 

 rent selection. There is no need to await incontrovertible evidence of over- 

 dominance; indeed even if it were in hand the direct test would still be 

 needed. 



The second proposal was that if recurrent selection for specific combinabil- 

 ity should be important, selection within and among inbred lines had been 

 greatly over-emphasized. The inbreeding interphase could be abandoned. 

 This would provide an enormous saving in time and otherwise, particularly 

 with poultry and other livestock. Curiously, some reviewers have described 

 the proposed breeding plan as a "laborious method." 



Grain yield of corn depends appreciably on resistance to new and sporadic 

 diseases, insects, and adverse environmental complexes. Here it would seem 

 that overdominance is not likely, but that selection within and among inbred 

 lines is yet of real value. Significant resistance where it exists will eventually 

 be identified in continuing a stable line. Selected crosses will be generally 

 superior insofar as the several resistances are dominant and matched com- 

 binations are found. Here again I am not certain that rapidly recurring 

 progeny tests without inbreeding may not be equally or more effective in the 

 main. One resistant line among some hundreds of susceptible ones in an epi- 

 demic provides a striking field illustration — perhaps a deceptive one. 



Breeding plans to accumulate specific combinability may be designed in 

 many ways, the better ones to be determined by actual tests. Testers might 

 best be the male parent of the hybrid in some cases, or the female parent in 

 others. The inbreeding interphase may be omitted or included in any prac- 

 ticable degree. It has been thought that the problem of the preceding para- 

 graph might be met well enough by direct selection in the crossbred lot and 

 selection among So testcrosses. But in some cases there might be an advan- 

 tage with Si or So testcrosses. With So or S2 some of the selection may be for 

 general combinability, for higher frequencies of genes which are favorable in 

 any combination. 



The early view (Hull, 1945a, Proposition 7) was that where aA is generally 

 intermediate to aa and AA, A should be in high frequency, in improved 

 varieties. Not much further opportunity for improving combinability would 

 remain. 



Crow's viewpoint, as he has presented it here, seems to be that without 

 overdominance long continued selection in any form would have carried 

 favorable alleles to high frequency in equilibrium with reverse mutation, 

 where heterozygosity is infrequent and heterosis not large. 



If recent shifts of environment or of emphasis in artificial selection should 

 have provided important loci with intermediate gene frequencies, Crow's 

 argument may not be germane. Here I may venture an opinion (Hull, 

 1945b) that without overdominance rapidly recurring mass or ear-row selec- 

 tion should continue to surpass contemporary selected Fi crosses of homozy- 



