No. 510] HEREDITY AND VARIATION 



331 



selection. But finally we reach a stage in which all but 

 the largest race have been excluded. Thereafter we can 

 make no farther progress. In vain we choose for breed- 

 ing the largest specimens of the lot ; all belong to the same 

 race, so that all produce the same progeny. Selection 

 has come to the end of its action. 



It is well known that this is a course of events com- 

 monly observed in selective breeding. Improvement oc- 

 curs for a time, then stops. We might have produced 

 the final result at once, in our infusoria, by merely isolat- 

 ing at the first selection the largest individual of the 

 entire lot; its progeny would have produced at once a 

 pure race of the largest size attainable. Selection here 

 consists simply in isolating already existing races; it 

 produces nothing new. 



Thus the facts in Paramecium furnish an excellent 

 illustration, in the simplest possible form, of the prin- 

 ciples of breeding for improvement so convincingly set 

 forth in de Vries 's recent work on plant breeding, and in 

 his other writings. 



It is well known that in inheritance extremely marked 

 parental characters appear less marked in the progeny; 

 the progeny of extreme parents are on the whole nearer 

 the mean than were the parents. This fact is spoken of 

 as regression. The reason for this is clear in such a 

 collection as Fig. 5. Suppose we breed from the very 

 largest specimens of the entire collection. These are the 

 largest individuals of the largest race. They produce, 

 as we have seen, like any other specimens of that race, 

 progeny of merely the mean size for that race. The 

 progeny will then of course be on the average smaller 

 than their parents, but they will be above the mean size 

 of the species as a whole, since they belong to the largest 

 race. Thus 4 'regression is not complete"; the progeny 

 diverge from the parents toward the general mean of the 

 species, but do not reach it. If however we consider a 

 single pure race alone (Fig. 4), and breed from the ex- 



