328 A. J. MANGELSDORF 



type, it is hardly to be expected that the phenotype of an inbred line will 

 afford a wholly reliable indication of its breeding potentialities in hybrid 

 combinations. 



4. We need to keep in mind the limitations that pertain to a rating for gen- 

 eral combining ability. The best "general" combiner thus far discovered in 

 corn is not so general in its combining ability as to be able to combine to ad- 

 vantage with itself or with any other genotype that happens to be afflicted 

 with the same bottleneck genes. At best, a rating for general combining abil- 

 ity can represent nothing more than an average arrived at by lumping a given 

 population of specific combinations. An average derived from a different pop- 

 ulation of specific combinations could result in quite another rating. 



5. If aseriesof inbreds^,5, etc., be crossed with a tester inbred T, we ob- 

 tain the hybrids AT, BT, etc. The yield of AT will be determined by the 

 bottleneck genes in the A T genotype. The yield of BT will be determined by 

 the bottleneck genes in the 5 T genotype. The test cross can tell us which lines 

 combine to best advantage with the tester line, but it cannot reasonably be 

 expected to tell us more than that. It cannot, for example, tell us with cer- 

 tainty what we may expect from A X B. Both A and B may combine to 

 advantage with T, but if A and B each happen to be afflicted with one or 

 more of the same bottleneck genes (not present in T) the yield of the cross 

 AB will suffer. 



6. The failure of a cross between two convergently improved lines to equal 

 the cross between the two original lines from which they were derived cannot 

 be taken as critical evidence for the existence of an overdominance mecha- 

 nism. The benefits which convergent improvement seeks to achieve can be 

 vitiated if a recessive bottleneck gene b, present in only one of the original 

 parent lines, should become homozygous in both convergently "improved" 

 lines. Selection exercised with the object of preventing such an occurrence 

 may be ineffective if b becomes a bottleneck only under the enhanced rate of 

 physiological activity of the A{B) X B{A) hybrid. 



7. During recent years several examples of heterosis reported in the litera- 

 ture have been attributed to the effect of heterozygosity at a single locus. 

 When the amount of heterosis is substantial, it should be possible to verify 

 the validity of the hypothesis by breeding tests. If the two parents are really 

 isogenic, except for the heterosis locus H, and if HiH2 individuals are more 

 vigorous than either homozygote, then by selfing only the most vigorous in- 

 dividuals in each generation it should be possible to retain in one-half of 

 the population the original vigor of Fi even after many generations of selfing, 

 and such a line should continue indefinitely to segregate HiHi, HiH-i, and 

 H<iH2 individuals in a 1:2:1 ratio. 



8. East describes the effect of heterosis as "comparable to the effect on a 

 plant of the addition of a balanced fertilizer to the soil or to the feeding of a 

 more adequate and more chemically complete diet to the animal." The simi- 



