SUPPLEMENTARY STUDIES ON THE DIFFER- 

 ENTIAL MORTALITY WITH RESPECT TO 

 SEED WEIGHT IN THE GERMINATION 

 OF GARDEN BEANS 

 Dr. J. ARTHUR HARRIS 

 Carnegie Institution of Washington 



I. Introductory Remarks 



In an earlier paper 1 I have shown that in field cultures 

 the mortality of seeds of Phaseolus vulgaris is not 

 random, bnt selective with respect to seed weight. 



From the available data it appeared that both the 

 upper and lower weight classes are more heavily drawn 

 upon in the mortality than is the modal region of the seed 

 weight distribution. So delicately balanced is this mor- 

 tality of the two extremes in the particular series of 

 experiments that the mean weight of the survivors in the 

 long run differs not at all from that of the population 

 from which they are drawn, while their variability, either 

 absolute as measured by the standard deviation or rela- 

 tive as expressed by the coefficient of variation, is dis- 

 tinctly less than that of the original population. 



Now while these results are deduced from such large 

 and representative series of data, that their general 

 validity for the specified conditions seems beyond much 

 question, their substantiation has appeared to me for 

 two reasons most desirable : 



(a) The demonstration of selective elimination was 

 somewhat indirect. Comparisons could not be made 

 between the physical constants of the seeds which devel- 

 oped and those of all exposed to risk, or between the con- 

 stants of those which actually failed and those which 

 actually developed, but were necessarily drawn between 

 the constants of the general population from which the 

 seeds were taken for planting and those of the seeds 



Weight Occurring in FicM ('ultmvs of Ph,». ,,! us ruh/oris." Amek. Nat.. 

 46: 512-525, 1912. 



683 



