140 Controversial Questions. 



the plant is highly correlated with the number of super- 

 numeraiy carpels. The thickness of the stem, the height 

 of the plant, but especially the weight of the fruit afford 

 a good index of strength. If we arrange the individual 

 plants of a bed, in a row according to this latter character, 

 we find that they form an almost regularly ascending 

 series with regard to the monstrosity. 



It follows therefore that a selection with reference 

 to the elaboration of the carpellary crown is by no man- 

 ner of means without relation to the general nutritional 

 condition of the plant. On the contrary such a selection 

 is merely a selection of the most highly nourished. 



It is just the same if we deal with the //n'n //^-varia- 

 tions. It is only the feeblest specimens which are desti- 

 tute of fully developed supernumerary carpels though 

 as we have said they possess 1 or 2 rudimentary ones ; 

 their capsules are often so small that they do not contain 

 good seed. If we look for capsules with better seeds we 

 find more pronounced traces of the monstrosity. Retro- 

 grade selection or selection in a minus direction obviously 

 consists therefore in a choice of the weakest individuals. 



Nevertheless, in both these cases selection has the 

 effect that it usually has. From the seeds of self-fer- 

 tilized plants with a very large number of metamorphosed 

 stamens we get offspring with this character highly de- 

 veloped; from seeds from fruits poor in supernumerary 

 carpels, a race exhibiting this monstrosity only to a very 

 slight extent. Continued selection for a few generations 

 intensifies this effect, provided that the conditions under 

 which the plants were grown are average ones and that 

 the experiments are carried out on a large scale extend- 

 ing over several square meters of ground. 



External conditions, then, exert corresponding in- 



