The Shifting of the Curves of Variability. 551 



creases, and the amplitude of variation increases too 

 (short-fruited race 1891-1894). 



4. The variability-curves remain ahncjst symmetrical 

 (0i=Q2) although their apices are considerably shifted 

 to one side. The deviations from this symmetrical form 

 almost all lie within the limit of ordinary errors of obser- 

 vation. 



§ 5. THE SHIFTING OF THE CURVES OF VARIABILITY 



BY NUTRITION. 



Oenothera Lamarckiana, and other plants as well, 

 may be stimulated to a much swifter and more vigorous 

 growth by planting the seeds in pans and picking out 

 the young seedlings soon after the unfolding of the 

 cotyledons into fairly large pots filled with heavily ma- 

 nured garden soil. One would expect that, by continuing 

 this process for a few generations, it would be possible 

 to increase the mean fruit-length very considerably, in 

 accordance with the principle of nutrition of the parent- 

 plant. 



The experiment to be described fulfils this expecta- 

 tion; the increase made in three years (1892-1894) far 

 exceeds that made in the selection experiments already 

 described, in which the seed was sown in the garden 

 (Fig. 116). 



Let us give a description of this experiment year by 

 year. It began in the spring of 1892 with the seeds (if 

 the species Oenothera ruhrinervis (p. 273) which arose 

 in my experimental garden in 1889: the length nf the 

 fruit in this species is the same as in O. Lauiarckiaua 

 (Fig. 99, p. 446). 



In 1890 the seeds were harvested from a number of 



