1904] SHULL—PLACE-CONSTANTS FOR ASTER. 373 
disk-florets were studied quantitatively, and the results compared 
with those of the earlier study. 
Twelve successive collections were made from the same plot, and 
it was found that the earliest collection had low mean numbers, that 
the mean values then leaped quickly to a maximum, falling gradually 
to near the end of the season, and that the last collection exhibited a 
tise, the rise in mean values at the beginning and at the end of the 
season being in disagreement with the conclusion reached in my 
eatlier study. In general, the first head to bloom on any stem has 
the highest number of parts possessed by any head produced by that 
stem, and the last to bloom has the lowest number. The low mean 
numbers at the beginning of the season are due to the precocious 
flowering of the weakest individuals, and similarly the rise at the end 
of the season is due to the belated flowering of a few very vigorous 
individuals, 
Comparison of the results with those of 1900 show that the mean 
Values in 1903 were 10-12 per cent. lower than in 1900, and that 
*ccompanying these low mean values there are a strong positive 
ewing of the curves, a remarkable rise in the coefficient of varia- 
llity, and a considerable increase in the coefficient of correlation. 
The difference in the mean values for the two years is attributed 
10 less favorable climatic conditions in 1903, chiefly to low tempera- 
ture and low light-intensity in the month of June. 
The skewness is due to the unequal sensitiveness of individuals to 
“ ve of environment. It is positive because the proportion of 
nsetvative Individuals is small. In direct or physiological variation, 
a . skewness indicates great sensitiveness and “prophetic” 
~ “Ness indicates slight sensitiveness to the changes of environment. 
egg great increase in the coefficient of variability is due to an 
in the standard deviation and a decrease of the mean. The 
Tatatig Seeffcient of variability is not satisfactory in cases of skew 
_~» and the value of o alone should be used as the measure of 
Ses of variability in one and the same population. ~ 
icing) 8€S In the coefficient of correlation may be due either to an 
change of correlation or to the introduction of a greater or less 
til heterogeneity, The latter is probably responsible for the 
hoted in this species, 
