Stout & Boas: STATISTICAL STUDIES IN CICHORIUM 449 
The collection of data from a number of individuals to be 
compared in any sort of statistical study may therefore involve 
several decided elements of error. The collection of the same 
amount of data is not alone adequate, for the two sets may repre- 
sent different relative periods of bloom. Even if data were taken 
for a period of bloom, as for the first ten days, they would most 
often represent different proportions of the entire period of bloom. 
Data from the last period of bloom only would give an unreal and 
apparent similarity between plants, as all plants are more nearly 
alike in the lower numbers per head produced during the late 
season. Data for the first period of bloom would emphasize the 
differences in the higher numbers per head. Indiscriminate col- 
lection of data from individual plants therefore involves so many 
sources.of misrepresentation as to be of little use in formulating 
any conclusions. It is quite possible that much the same condition 
exists in respect to the hereditary studies of such characters as the 
size of flowers (as Goodspeed and Clausen, 715, have suggested) 
or as the size and weight of fruit produced or of any other character 
which is subject to such degrees of partial variability. 
The data for chicory indicate that there is much individual 
variability in such a character as number of flowers per head. Such 
differences as total number of flower heads and length of blooming 
period are quite closely correlated with vegetative vigor and 
variations of this sort are seen even in closely inbred and very 
uniform races. Much more fundamental differences exist espe- 
cially between plants of different races or between races as such. 
The modal number and the computed value of the first day of 
bloom may be quite different. The rate of change as determined 
by values of 6 may be decidedly different, even to the point of 
exhibiting no seasonal decrease or even of revealing an actual 
increase in number as the season advances. While such cases are 
not numerous they are sufficiently frequent (for this the data are 
fully convincing) to indicate that the usual processes of noticeable 
differentiation according to position which result in seasonal de- 
crease may not be in evidence or may even be reversed in their 
operation so that the differences between terminals and laterals 
as to number per head do not appear and hence both terminals 
and laterals show much the same range of variation. These 
