Stout в Boas: STATISTICAL STUDIES IN CICHORIUM 353 
not considered as correct because homotyposis was involved. 
The true parental correlation, according to Pearson’s conception, 
was higher. By accounting for homotyposis the value was raised 
from the average of 0.20 to a value lying between 0.35 and 0.40. 
It would seem that much of the difficulty here experienced in 
attempts to make exact determination of values, even for popu- 
lations such as Pearson studied, lies in treatment of all the varia- 
tions as ‘‘chance.” Although Pearson definitely recognizes that 
lateral flowers are differentiated from terminals, there is no attempt 
to determine values for such partial variability. 
In further studying heredity of number per capsule in the 
poppy, Pearson ('o6) sought to avoid the difficulties previously 
encountered in estimating the individual when multiple observa- 
tions involving partial variability were made. He attempted to 
do this by “confining the attention to the first or principal flower.” 
In 1903 and 1904, crops were grown from seed of random samples 
in I5 different localities and treated as populations. Differences 
in mean and in variability were found which were attributed to 
effects of environment as affecting individual variability and 
which were so great as to be “not directly comparable." It was 
possible to determine parental correlations for these results in 
only one population; a crop grown in 1904 from parents of a 1903 
crop, the two crops, however, were grown at different localities. 
The raw correlation was only 0.1717. 
Pearson therefore concludes that the determination of heredity 
even for such an easily measured quality as the number of stig- 
matic bands in pods of the poppy is exceedingly complex and 
difficult, and he now questions "whether the apical flower is as 
true a measure of individuality as the totality of flowers on the 
plant” (706, р. 400). 
Pearson is here concerned with population studies and in 
intensity of parental correlation for rather mixed populations. 
His treatment and results suggest and in fact reveal many sources 
of variability. His rather uncertain results raise very definitely 
the question of how to value adequately a numerical character 
which exhibits elements of both chance and differential variability 
for both partial and individual variations. 
Bateson ('or, 'o3) questions the аши of Pearson's distinc- 
