No. 564] STUDIES ON DIFFERENTIAL MORTALITY 759 



duction is not so evident in the coefficients of variation, 

 probably because of changes occurring in mean weight. 



There is clearly a lowering of both absolute and rela- 

 tive variabilities between seeds which fail to germinate 

 and those which germinate normally, or those which 

 germinate at all, either abnormally or normally. The dis- 

 position which is made of the seeds which give rise to 

 abnormal seedlings does not affect the conclusion concern- 

 ing a reduction in variability due to a differential death 

 rate. 



To what extent this reduction is incidental to a change 

 in mean through elimination preponderantly from one 

 end of the range, as compared with elimination from both 

 the extremes without change of type, must be determined 

 on wider series of data, and probably by the use of 

 statistical methods not yet applied to the problem. 



5. The constants of this paper, taken in connection with 

 data made directly available from other published studies 

 by the key letters, can be used towards the solution of a 

 number of problems not touched upon here. These have 

 been purposely left out of account because they were 

 aside from the present main purpose and because I hope 

 to have much more extensive materials for their solution 

 later. 



6. Concerning the causes of the differences in viability 

 no conclusions can be drawn. I have shown 18 that in 

 general the larger seeds require longer for germination, 

 but the precise relation, if any, of this phenomenon to 

 selective mortality, as well as its explanation in more 

 general physical and chemical terms, are still to be 

 worked out. 



Tumamoc Hill, Tucson", Ariz., 

 April 3, 1913 



is Harris, J. Arthur, "A First Study of the Relationship between the 

 Weight of the Bean Seed, Phaseolus vulgaris, and the Time Required for its 

 Germination." In press. 



