INBRED LINES FOR HETEROSIS TESTS? 331 



By definition, epistasis is universal in the sense that expression of every 

 gene is to some degree dependent on and modified by the effects of genes in 

 other sets of alleles. Epistasis would include fixed multij)licative or propor- 

 tional effects of each gene on the expression of non-allelic genes. Such 

 epistasis, although unlikely to be important, would be of a highly predictable 

 sort and would disappear if phenotypes were measured in log scale units. 

 A potentially much more important sort of epistasis would be that involved 

 whenever a phenotypic maximum is associated with an optimum genetic 

 intermediate (Wright, 1935). Here a given gene may have either a positive 

 or a negative selective value, depending on whether the individual's average 

 genotype is above or below the optimum genetic intermediate. 



Some of the evidence concerning the kind of genetic variability with which 

 we must deal in seeking to improve economic characters of swine has been 

 considered earlier (Dickerson, 1949, 1951) and may be summarized here as 

 follows: 



Inbreeding and Crossbreeding Effects 



Proportion of heterozygous loci has a major influence on total perform- 

 ance, affecting most the highly important but lowly heritable characters for 

 which selection has been consistently in one direction. Take for example, an 

 intra-season comparison of 538 inbred and 325 linecross litters from the same 

 lines in four projects of the Regional Swine Breeding Laboratory (Dickerson 

 et al., 1947). This showed a decline in performance per 10 per cent increased 

 inbreeding of litter amounting to 2.6 and 7.8 per cent, for litter size at birth 

 and weaning, respectively; 2.6 per cent for pig weight at 154 days of age; and 

 11.4 per cent for total weight of litters at 154 days. Similar estimates per 10 

 per cent increased inbreeding of dam, based on sixty-three inbred and fifty 

 linecross dams at the Iowa Station, were 2.1 and 5.0 per cent for litter size 

 at birth and weaning; 1.6 per cent for pig weights at 21 days; and 5 per cent 

 for total weight of litters at 154 days. 



Results from studies of regression of j)erformance on inbreeding of dam 

 and litter within line and season (Blunn and Baker, 1949; Stewart, 1945; 

 Comstock and Winters, 1944; and Hetzer et al., 1940) agree quite well with 

 the figures given. Inbreeding of dam and litter together greatly depresses 

 prolificacy, suckling ability, pre- and post-natal viability and growth rate, 

 and particularly their product — total litter weight. Inbreeding effects on 

 carcass composition, body conformation, and efficiency of food utilization 

 were relatively minor (Dickerson et al., 1946). 



The results of the earlier crossbreeding exi)eriments have been summarized 

 by Lush (1939) and Winters (1936). When the mean of the two purebred 

 stocks crossed is compared with the crossbred litters, the results of many ex- 

 periments summarized by Carroll and Roberts (1942) indicate that the 

 average performance of crossbred individuals is increased about as much as 

 it would be by a 10 per cent reduction of inbreeding (see Table 21.1). More 



