Limits to the Effects of Selection. 119 



this are not supplied by the Hnear variation of characters 

 which are already present. 



2. The Duration of Progress. The view that linear 

 variation is unlimited in the sense that the changes which 

 can be brought about by selection in the course of cen- 

 turies are greater than those which can be attained in the 

 course of a few years, is absolutely without foundation. 

 We are speaking of course of the improvement of a 

 single character considered by itself. As a matter of fact, 

 2 or 3 years under favorable conditions, or 3 to 5 under 

 ordinary ones, are quite sufficient. The further prosecu- 

 tion of selection serves merely to keep the race at the 

 level which it had reached, unless special circumstances 

 arise (see sections 6 and 7). 



3. The Limits of Selection. There is just as definite 

 a limit to selection as there is to linear variation. The 

 limits of the latter can be extended by dealing with a 

 greater number of individuals; but it takes an enormous 

 increase in this number to sensibly extend these limits. 

 It is the same with selection; but this at least has the 

 advantage that the number of individuals dealt with can 

 be diminished by discarding those of no value. 



It is often stated that variation in a given direction 

 can be increased bv selection in that direction. Observa- 

 tions, or exact information in support of this statement 

 are not given. There is of course an appearance of 

 change owing to the elimination of the less valuable 

 individuals. As a matter of fact in accuratelv recorded 

 cases the very reverse is found to be the case ; that is to 

 say, that it becomes gradually more difficult to effect any 

 change until finally it becomes impossible. 



Darwin''s view that plants increase in variability in 

 the first years after they are brought into cultivation 



