SPECIFIC AND GENERAL COMBINING ABILITY 353 



the values which would be expected on the basis of the known general com- 

 bining abilities of these two lines or individuals and the maternal ability 

 of the female parent. As Lush (1948) has pointed out, apparent specific 

 effects, or what animal breeders usually call nicking, also can be a conse- 

 quence of Mendelian sampling, of inaccurate estimates of the additive genetic 

 values of the two parents, and of environments affecting the progeny which 

 are different from the average environments in which the general combining 

 abilities and the maternal abilities were estimated. 



Genetically, specific combining ability is a consequence of intra-allelic gene 

 interaction (dominance) and inter-allelic gene interaction (epistasis). We shall 

 assume in this paper that we can estimate only the joint effect of dominance 

 and epistasis. As an illustration of specific combining ability let us suppose 

 that we know that the general combining ability with respect to weight in 

 swine line A is +10 pounds at 154 days, and that the general combining 

 ability plus maternal ability of line B is -|-5 pounds at 154 days. Then if an 

 indefinitely large number of progeny of the cross A X B has a mean of +7 

 pounds, the specific effect for this cross is 7 — 10 — 5 = —8. 



SELECTION FOR GENERAL AND SPECIFIC 

 COMBINING ABILITY 



Under some circumstances selection would be largely for general com- 

 bining ability, and in other circumstances for a combination of general and 

 specific combining ability. For example, those selecting sires for use in a 

 large artificial breeding ring are interested primarily in obtaining sires with 

 the highest general combining ability with respect to the population of cows 

 and environments in which the bulls are to be used. On the other hand, those 

 wishing to employ crosses among inbred lines for commercial use select for a 

 combination of general, maternal, and specific effects. 



Now let us consider some of the problems involved in selecting for general 

 and specific combining ability. There are reasonably good solutions to some 

 of these problems, but almost none for others. Some of the questions which 

 are involved are : 



1. Given a particular set of records how can one best estimate the general 

 combining abilities of individuals, families, or lines, and how can one best 

 estimate the value of the progeny of a specific cross between families or in- 

 bred lines? 



2. What proportion of the breeder's resources should be put into a testing 

 program? For example, if he is dealing with inbred lines, what proportion of 

 his resources should be employed in the making of lines and what proportion 

 in testing them for general and specific combining ability? 



3. Having decided on the size of the testing program, what kind of tests 

 should be made? For example, should lines be tested in topcrosses or in line 

 crosses or in some combination of these two procedures? Also what use should 



