12 
Influence of Weight of Seed on Plant 
" light," or at most into " heavy," " medium " and " light." The meaning of such 
crudely denned terms of course differs from experiment to experiment. They 
may, with certain limitations, enable one to say which class of seeds gives the 
best results. They do not permit comparisons of the advantages to be gained by 
seed selection in different varieties. They do not allow of the writing of a general 
formula enabling one to predict the yield from seed of given weight. Yet such a 
formula is precisely what is needed. The advantage of sorting seed — if there be 
an advantage— depends not only upon the increased yield (or increased uniformity 
of the crop in certain cases) but also upon the cost of carrying out such a selection. 
In deciding upon the stringency which is profitable in the seed selection, the 
practical breeder should know the exact weight to be attached to each factor. 
(3) The experiments have not been carried out in a way to make possible 
the calculation of the probable errors of the results. In many cases the experi- 
ments have been small, and the conclusions are open to serious question, both on 
the ground of possible experimental errors and on the ground of the probable 
errors of random sampling. 
All of these difficulties can be overcome by the application of the modern 
statistical methods to the problems. A wide series of such biometric constants 
deduced from carefully conducted physiological investigations ought to be of great 
service to the man dealing with the practical problems of agriculture. 
II. Material and Methods. 
The materials for this study are drawn from three varieties of garden beans — 
the White Flageolet, White Navy and Ne Plus Ultra. Altogether there are 
twenty series grown under most diverse environmental conditions. Their history 
has already been given* for another purpose and need not be repeated here. The 
symbols used to designate the different lots are the same in both papers ; hence 
the reader may make any comparisons which he sees fit. 
All of the seeds planted were, as far as could be determined by inspection, 
perfect in form and development. Each was individually weighed and classified 
in a uniformly graduated scale. These individually weighed and individually 
labelled seeds were then mixed and planted at random in rows by varieties, the 
rows in their turn being scattered over the field, more or less at random, in order 
to counteract by chance distribution any influence of the possible heterogeneity of 
the substratum upon the characters of the plants. All the determinations of 
number of pods produced were made on individual plants. 
Thus the probable errors for the statistical constants for weight of seed 
planted, number of pods per plant, and for the correlation between the weight of 
* Harris, J. Arthur: "A First Study of the Influence of the Starvation of the Ascendants upon the 
Characteristics of the Descendants." Airier. Nat., 1912. In press. 
