290 JAMES F. CROW 



tween selection and mutation pressures. This assumption probably is not 

 strictly correct for any population. Any equilibrium involving occurrences 

 as rare as mutations must be slow of attainment. Hence many if not most 

 populations must not be at equilibrium. Probably the most common way 

 in which a population gets out of equilibrium is by an alteration of the breed- 

 ing structure or population number so that the effective amount of inbreeding 

 is changed. If the change in population structure is such as to increase the 

 amount of homozygosity, a new equilibrium is reached comparatively rapidly 

 through the elimination by selection of the recessives which have been made 

 homozygous. On the other hand, if the change in population is such as to 

 decrease the amount of homozygosity a new equilibrium is attained only 

 through the accumulation of new mutations. This is an extremely slow 

 process. 



Since the return to equilibrium is much slower when the population 

 changes in the direction of less inbreeding, it follows that most populations 

 which are out of equilibrium will be out in the direction of having too few 

 detrimental recessives. Therefore the effect of fluctuations in population 

 size and breeding structure will be on the average such as to increase the 

 fitness of the population. For this reason, the average loss of fitness per locus 

 is probably less than the mutation rate. Fisher (1949) has pointed out that 

 if the yield of a crop is near a "ceiling," the relative effect of each factor con- 

 ditioning yield becomes less. There will be a similar tendency for the popula- 

 tion to be out of equilibrium because of the slowness of occurrence of the 

 mutations required to bring the population to the new equilibrium level. 



Another factor also pointed out by Fisher is that complete lethals and 

 highly deleterious factors contribute to the mutation rate but, at least in 

 grain crops, have no appreciable effect on yield since they are crowded out 

 by other plants. 



All of these factors make the 5 per cent figure an overestimate, so it should 

 be regarded as a maximum. The true value may be much less. In this con- 

 nection Fisher (1949) said: 



... it would appear that the total elimination of deleterious recessives would make less 

 difference to the yield of cross-bred commercial crops than the total mutation rate would 

 suggest. Perhaps no more than a 1 per cent improvement could be looked for from this 

 cause. Differences of the order of 20 per cent remain to be explained. 



These considerations make it difficult to explain, in terms of the domi- 

 nance hypothesis, cases in which two equilibrium populations produce hy- 

 brids with considerable heterosis, or in which crosses between inbred lines 

 average appreciably more than the randomly mating populations from 

 which they were derived. 



This discussion is relevant only when the character is measurable in terms 

 of selective value. For yield characters subject to any high degree of artificial 

 selection an increase in yield is probably accompanied by a greater proper- 



