316 



THE AMERICAN NATURALIST [Vol. XL VI 



But concerning the influence of the feeding or starving of 

 the parent upon the characteristics of the offspring, we 

 have little direct experimental knowledge. 



It seemed expedient therefore to neglect for the mo- 

 ment the problem of the various edaphic and metereolog- 

 ical factors which determine the characteristics of the 

 individual and to ascertain whether the subjecting of 

 parent plants (or parents and earlier ascendants) to dif- 

 fering environmental conditions has any influence upon 

 the characteristics of the offspring. It was therefore 

 only necessary to find fields in which the soil barely sus- 

 tained a given variety and others which produced a luxu- 

 riant growth. The first would represent for the species 

 in question starvation fields. 



The judgment of the relative richness of the plots by 

 their actual productiveness is justified by our ignorance 

 of the nature of soil fertility. 



The reader who is inclined to criticize this method of 

 approaching the problem as very coarse may be reminded 

 of the following points : 



(a) The complexity of the problem of soil fertility is 

 such as to preclude a trustworthy evaluation of the par- 

 ticular factors determining the productiveness of any 

 parcel of ground. 1 For this reason I have purposely 

 omitted all but the barest descriptions concerning the ex- 

 perimental plots employed. 



(b) Artificial soils or water culture media of known 

 chemical composition were carefully considered and ruled 

 out. In the first place, the technical difficulties seemed 

 almost unsurmountable. Again, it seemed desirable to 

 carry on the experiments under conditions as nearly as 

 possible identical with those to be met with in practical 

 agriculture. Chemically prepared nutrient solutions are 

 useful in the physiological laboratory, but they do not 

 occur in practical farming, while soils which are "sterile" 



