754 



THE AMERICAN NATURALIST [Vol. XLVII 



Comparison of the relative variability in weight of 

 seeds giving abnormal germinations with that of those 

 failing to germinate at all, B-C, gives : 



isons to differences at least 2.5/7: 



M llVr-h.-.'- in -tjl 



All differences i 12 -3.04 



Differences at least IE are : 



The reduction in variability in passing from C to B is 

 clearly significant. 



Consider now the difference between seeds germinating 

 abnormally and those germinating normally, A-B: 



Or taking the usual minimum standard of statistical 

 significance : 



There is no certainty of any reduction in relative vari- 

 ability here. But turning back to the standard deviations 

 we find that there were fair indications of a lowering of 

 variability. 



