| 
1904] SHULL—PLACE-CONSTANTS FOR ASTER 369 
of variability, let us assume that conditions had been unusually 
favorable to such a degree as to give curves with the same values of 
¢, but negatively skew. The variability would then be approximately 
the same, but, instead of the coefficient being the same or even 
nearly the same, it would be very much less, owing to the greatly 
1000 
M 
value of the coefficient of variability in cases of skew variation, since 
its values in positively skew curves are not comparable with those 
in curves of the same species or even of the same population, which 
are negatively skew. If the “normal-mean” could be derived from 
skew curves, that might be used instead of the mean in the formula 
for the coefficient of variability, thus making the value of o alone 
indicate the changes of variability from time to time within one and 
© same population. This would be theoretically correct, but it 
must be evident that. the experimental determination of the normal 
mean, except through a long series of investigations upon any popula- 
tion under Consideration, is impossible, even though, as pointed out 
above, the degree and direction of skewness may in some cases give 
4 Tough approximation to it when the sensitiveness of the species 
M question is known. 
Returning now to the question as to the increased variability due 
© changed environmental conditions, we find that the present imper- 
Be eett of varicbility, which would tend to minimize the 
comty When conditions are unusually favorable, would still be 
erably increased by such unusually favorable conditions as 
Tesult in a negative skewness equal in magnitude to the positive 
he of the 1903 curves. We may confidently accept the results 
Study as Proof, therefore, that changes of environment do 
Tesult in increased variability. 
head noted in T1900 that the correlations between the parts in = 
in 1903 : very high, and by reference to Table H it will be a t . 
kin ey Were very considerably higher still, the highest coefficie 
he being that between bracts and rays. The exact meaning 
. SiN the degree of organic correlation is proving a somewhat 
. © Problem at the present time. Lupwic (1901) presents a 
8 “Se of this kind as evidence of racial distinctness between 
increased value of the mean. I do not think that gives a proper 
Would 
