282 
Journal of Agricultural Research 
Vol. XXVII, No. 5 
These represent the fundamental units on which the conclusions are 
based. Since rows i and 5 and 3 and 7 do not differ in variety but 
only in the origin of the seed used, the four rows (1, 3, 5, and 7) have 
been combined for a more extensive comparison of Pima and Meade 
cotton. 
This has given a series of semi-independent subsamples. The 
concordance of the results from these subsamples is one of the strongest 
arguments for the validity of the conclusions based on the series as a 
whole. 
In comparing the differences between the parental types, or between 
the parental types and the hybrids, we must remember that the two or 
three determinations on the same duplet or triplet will be rendered more 
or less alike by (a) the heterogeneity of the plot upon which the plants 
were grown and by (6) the differences in the time at which the samples 
were taken. Thus we should expect a correlation between the constants 
of the individuals of the same duplet or triplet, notwithstanding the 
fact that these individuals belong to different varieties of cotton and 
are merely rendered similar by community of environmental influences. 
The consequence of these correlations for the arguments of the present 
section of the investigation is that the probable error of the differences 
between the two parental types, and between the two individual parental 
types and the hybrid form, will be greatly reduced. We have, therefore, 
always calculated the probable error of the difference between the series 
of constants with due regard to the intensity of correlation between 
them. 
In the analysis of the data we may first present the frequency distri¬ 
butions for the physicochemical measurements made. We shall then 
proceed to a more detailed comparison of the physicochemical constants 
by means of statistical constants. 
We may now consider the frequency distributions of the constants 
for the individual classes of plants. The depression of the freezing 
point may be expressed in units of 5 per cent of the molecular lowering 
taken as A = 1.86. The results are given in Table VI. 
Inspection of these distributions shows clearly that the constants for 
Pima are somewhat higher than those for Meade. The numbers are 
rather too small to indicate clearly the nature of the distribution. In 
some cases they are wholly skew, but in general there is a suggestion of 
a more or less symmetrical distribution about a central mode. The fre¬ 
quencies for the hybrid suggest a lower minimum than that found in 
either of the parent forms. 
The distribution of the measures of specific electrical conductivity is 
shown in Table VII. The distribution of the ratios of specific electrical 
conductivity to the freezing point lowering is represented in Table VIII. 
The results for specific electrical conductivity show that the Egyptian 
cotton has a somewhat higher maximum than that found in the com¬ 
parable Upland series. It is difficult to decide merely from inspection 
of the distributions the relative value of the ratio k/A in the Upland, in 
the Egyptian, and in the forms. 
The frequencies for the hydrogen-ion concentrations are shown in 
Table IX. Apparently the values of P H are somewhat higher in the 
Upland than in the Egyptian form. The relation of the hybrid to the 
two parental forms can not be determined by inspection merely. 
