Feb., 1922] 
LA RUE AND BARTLETT — ^ PESTALOZZIA GUEPINI 
83 
Spores produced, but the differences were not correlated with morphological 
characters. At the outset of the work it appeared that the ratio of length 
to breadth might be a more useful character in the differentiation of the 
strains than mere length, but the breadth measurements could not be made 
with sufficient accuracy by an ordinary ocular micrometer to make them of 
value. In the experiments we encountered a very small number of aberrant 
cultures, characterized by slow growth, numerous deformed spores, and 
greatly shortened appendages, which we attributed at the time to careless 
use of mercuric chloride on the part of a native helper. Our supposition 
was confirmed by the immediate return to normality in the following genera- 
tion of the strains affected. There was no basis for throwing these cultures 
out of consideration, since they could not be distinguished from extremes 
due to other factors. Consequently they increase the range of variation in 
generation means of a number of strains, and afford one reason for our state- 
ment, above, that more careful work would merely result in decreasing the 
difference which would be significant in proving the distinctness of strains. 
In all, the comparisons involved thirty-five strains — twenty-two from 
rubber, seven from cocoanut, three from tea, two from oil palm, and one 
from betel-nut palm. Two strains were measured through only four succes- 
sive generations, eight through five generations, sixteen through six, seven 
through seven, and two through eight generations. Since the time in- 
volved was the same for all strains, it is evident that some of them required 
twice as long for spore production as others. The numerical data for all 
cultures are summarized in tables i and 2. 
In deciding how many demonstrably different lines were represented 
in our set of thirty-five isolations, we were of course concerned with making 
decisions as to how significant were the differences of the final means for 
the several strains. It is obvious from simple inspection of tables i and 2 
that more than one strain was involved, for the generation means of each 
single strain fluctuated within a range of variation (on the average) of 
somewhat less than 3 m in spore length, whereas the total range of variation 
of the generation means of the whole set of thirty-five cultures was 13.8 ^t. 
We therefore felt justified in determining the standard deviation of genera- 
tion means by throwing all the deviations together into one series, calculat- 
ing the plus and minus deviations of the generation means each from its 
own strain mean. In this manner we determined an approximate value for 
the standard deviation of the generation means of an indefinitely large 
number of generations of any given strain as 1.23 yi for spore length and 
2.13 )U for appendage length. All the strains were similar enough so that 
these standard deviations might be applied to any strain. In order to 
determine the significance of the differences between strain means, the usual 
formula for the standard error of the difference of means was applied : 
