454 FRED H. HULL 



the plan — by reducing one of the reciprocal groups to the single-cross tester 

 alone. That tester is to be continued indefinitely. Another way to intensify 

 the operation was to increase the frequency of recurrence of selection. This 

 was done by adopting the general principle of early testing, by abandoning 

 the inbreeding interphase of each cycle, by testing So plants rather than S3 

 lines or higher. Inbred lines, including the tester lines, of the second reciprocal 

 group were intercrossed to provide one crossbred group of So plants. Re- 

 peated selection within this crossbred group for combinability with the per- 

 manent unrelated tester is the proposed plan. It is only for practical reasons 

 that one homozygous line is not employed as the tester for field corn. With 

 sweet corn a line tester might well be used. 



The working definition of specific combinahility employed in designing the 

 foregoing breeding plans was about thus: that part of the genetic superiority 

 of specific Fi crosses of homozygous lines which is not transmitted into or 

 through general recombinations. The concurrent definition of general com- 

 binability then is: that part which is transmitted into and through general 

 recombinations. That these definitions are perhaps inadequate for analyses of 

 variance does not necessarily mean that they are not admirable for the other 

 purpose. 



ShuU, East, and others who isolated inbred lines and crossed them discov- 

 ered that inbreeding did little or no irreparable harm to the germ plasm. 

 Gametes of inbred lines hardly differ basically from gametes of crossbred 

 varieties. The inbreeding effect is very nearly or entirely a zygotic phenome- 

 non. Vigor genes in both homozygous and heterozygous associations were 

 obeying Mendel's first law of non-contamination. All of this was an important 

 discovery. 



ShuU in addition invented selection for specific combinability, which was 

 certainly something new under the sun; yet to be generally recognized as one 

 of the great inventions. Shull was led, I suspect, to this invention by the 

 empirical evidence before him, not by considering the more abstract concept 

 of heterozygosis. Shull must have recognized very soon that reconciliation of 

 his invention with his knowledge of genetics required heterozygosis, and per- 

 haps the more inclusive heterosis. 



RECURRENT SELECTION FOR SPECIFIC COMBINABILITY 



A little more than thirty years later the inevitable invention of recurrent 

 selection for specific combinability was made from matter-of-fact empirical 

 considerations as outlined above. Again it seemed necessary soon afterwards 

 to embrace some theory of heterozygosis for reconciliation with genetics. 

 The breeding plan was presented (Hull, 1945a) with confusing emphasis on 

 the abstract concept of overdominance, I fear, and too little emphasis on the 

 actual motivation. 



May it be said now that the first proposal was to determine with direct 



