No. 546] INFLUENCE OF STARVATION 



325 



fourth variety, BG, which must be reserved for a later 

 paper. . 



The following method was adopted for counteracting 

 the possible hetereogeneity 10 of the fields upon which 

 the plants were grown. 



The different strains must be subjected to as nearly 

 as possible a random sample of the conditions afforded 

 by any plot. This end is secured by labeling each seed 

 individually and then scattering those of a particular 

 series quite at random over the field. If, then, certain 

 spots are somewhat more fertile or slightly moister 

 than others, all lines will have equal chances of being 

 represented there. If this were not done an undetected 

 differentiation in the substratum might induce quite de- 

 ceptive differences in the crops. 



In these experiments I did not, unfortunately, work to 

 quite this degree of refinement. For technical reasons, 

 it was desirable to have each of the varieties planted in 

 separate rows. Each seed was placed in an individually 

 labeled envelope and the envelopes of a series thor- 

 oughly shufiled. The series were then planted in rows, 

 which were scattered as nearly as possible at random 

 across the field. By this means an almost but probably 

 not quite random distribution was secured. 



3. Collection of Data 



The recording of the data from the mature plants was 

 an onerous but relatively simple process. 



As noted above, the plants were wrapped individually 

 at harvest time when as nearly dry as they could be 



