368 BOTANICAL GAZETTE [NOVEMBER 
character under consideration before assuming that either natural 
selection or mutation is involved in any given case of skewness. And 
although this is most strikingly true of plants, it must likewise be 
true of animals, especially of animals having a short life-cycle, so 
that no investigation can be considered as giving satisfactory support 
to any hypothesis of evolution until the sensitiveness of the character 
under consideration to secular changes shall have been determined. 
Perhaps even more remarkable than the skewness and the changes 
in mean value, which have resulted from the less favorable conditions 
in 1903, is the great increase in value of the coefficient of variability. 
Reference to Tables E, F, and G will show that the variability in 
the bracts in 1900 was 12.979 +.241, as compared with 19.928= .345 
in 1903. Corresponding changes are shown in rays and disk-florets, 
from 14.516+.270 to 19.766+.343, and from 12.546+.233 to 21.595 
+.374, respectively. As it has been assumed that the low mean 
values indicate that conditions were less favorable in 1903 than os 
1900, we may accept these changes in the coefficients of variability 
as proof of the hypothesis that when organisms are introduced into 
unusual surroundings or subjectcd to unusual conditions they become 
more variable, and that this would be favorable to any selective 
process which might set in as a result of the change. Before too 
great stress is laid on this conclusion, however, we need to consider 
the nature of the coefficient of variability. The importance of this 
constant lies in the fact that it is an abstract number and therefore 
allows us to compare the variability in characters of different magn! 
tude or even of different quality, as color, form, size, weight, pene 
etc. It consists of two factors, the standard deviation (7) and the 
i. 
mean (M), and is expressed by the formula C. V.= “yy ° 
with changes 
value of the coefficient of variability will change directly h 
g now to the 
of « and inversely with changes of the mean. Turnin 4 upon 
cause of the greatly increased coefficient of variability, so fin pet 
inspecting Tables E, F, and G that the value of ¢ was every rer 
considerably higher in 1903 than in 1900, and at the me rin 
the mean was much lower, so that both factors acted togetne 
producing the high values of the coefficient of variability 
: . measure 
To show that this coefficient is not always a satisfactory 
