Stout & Boas: STATISTICAL STUDIES IN CICHORIUM 337 
per head. If there be a noted differentiation among the flowers 
in a head, the number of both ray- or disk-flowers may vary. 
The compact inflorescence terminating a branch is itself considered 
as a unit structure in the study of partial variability upon which 
the value of the individual is to be based. The study of flower 
number in the Compositae, particularly the variation in the num- 
ber of ray-flowers, has received much attention. The biometrical 
treatment inaugurated by Quetelet and Galton has been utilized in 
the descriptions and analyses of the performance of a species as a 
whole and of various populations and lines within a species. These 
studies have not always recognized the extent and kinds of individ- 
ual and partial variability that may exist, or the degree to which 
differentiation may be recognized as operative in what appear as 
chance variations. The reasons for this will be obvious from the 
following considerations of the points of view and aims which 
influenced and guided different investigators in their studies. 
REVIEW OF LITERATURE PERTAINING TO NUMBER 
OF FLOWERS PER HEAD IN THE COMPOSITAE 
I. LUDWIG’S EVIDENCE THAT FLOWER NUMBER IS A SPECIFIC 
CHARACTER 
Ludwig was one of the first to investigate intensively botanical 
subjects by biometric methods and his studies were especially 
directed to problems relating to number of flowers per head in 
various Compositae. He also investigated species of Umbellif- 
erae as to the number of rays per umbel. His interest at first 
centered on the analysis in Galtonian terms of flower number 
for species as such. In the Compositae he studied those species 
which have both ray- and disk-flowers, but confined his obser- 
vations almost exclusively to ray-flowers. 
Ludwig did not realize the importance of the behavior of the 
individual plant, hence his method involved counts of heads col- 
lected indiscriminately. From such data curves were constructed 
to determine the behavior of the particular species, and especially 
to determine the highest mode or the maximum frequency of 
number per head. In the collection of data therefore, for the 
most part, no special recognition was given to individual variability 
and, of course, partial variability was ignored. 
