174 HAROLD H. SMITH 



homozygotes. The composition of parents, Fi, and selected inbred is shown 

 below with arbitrary "size" values assigned to each. 



Pi = XKX' (4) + y V (4) +Z'Z' (2) =10 



Po = X'X^ (2) + Y^V (2) -\-Z'-Z-' (6) =10 



Fi = X'X' (4) + I'l ¥' (2) ^Z'Z- (5) =11 



sel. = X'X' (4) + Y'V (4) -^Z-Z'- (6) =14 



Although the difficulty in selecting superior inbreds would become 

 greater with increasing numbers of effective segregating units, the following 

 advantages of selfed over crossbred systems would enhance the opportunity 

 for success: (1) lack of deleterious recessives, (2) less preponderance of 

 dominant favorable alleles, (3) homozygous pairs of alleles are superior, as a 

 result of an adaptive evolutionary process, to heterozygous combinations. 

 Naturally inbred organisms are products of historical evolutionary processes 

 in which harmonious systems of homozygous loci have been selected to 

 attain optimum adaptation. These considerations favor the expectancy 

 and practicability of obtaining maximum advance through selection and 

 inbreeding with self-fertilized organisms. 



SUMMARY 



There were two general purposes in conducting these experiments: First, 

 to demonstrate that by selection following intervarietal hybridization in a 

 self-fertilized organism, inbreds could be produced which transgressed the 

 character expression in parents and Fi; secondly, to investigate the relation 

 between estimated heritability and the actual results of selection. 



An inbred selection of Nicotiana rustica which transgressed the Pi and Fi 

 characteristics in plant height, node number, and leaf length was obtained. 

 The heritabilities for these three characters were calculated to be 54.9 per 

 cent, 12.4 per cent, and 11.2 per cent, respectively. The gains (in terms of 

 standard deviations) due to selection were 1.74, 2.42, and 0.91, respectively. 

 Some possible explanations for the lack of direct proportionality between 

 heritability and gain were discussed. 



The number of effective segregating factors for each of the three characters 

 studied was estimated to be of the same order of magnitude and relatively 

 few. Non-isodirectional distribution of -|- and — genes in the parent varieties 

 contributed to an underestimation of this number. 



Non-allelic interactions were apparently not an important source of 

 variation, as indicated by scaling tests and evidence from double cross means. 



Reasons for expecting greater advances by selection and inbreeding, as 

 contrasted to the use of first generation hybrids, in naturally self-fertilizing 

 genetic systems were reviewed. 



