S68 H. H. NEWMAN AND J. THOMAS PATTERSON 
treme right of the polygon will be noted several extreme variates, ~ 
so far separated from the others as to be examples of discontinuous 
variation. The rejection of these extreme variates would render 
the three constants, mean, modeand median, practically identical, 
but it would seem inadvisable to take any liberties with the data. 
Rather we would prefer to accept a close approximation to the 
normal curve of frequency as an indication that we are dealing 
with a case of chance variation, little if at all disturbed by com- 
plicating factors. 
From the above array have been calculated the standard devia- 
tion, 14.89+0.31 scutes; and the coefficient of vanes 2.685 
+ ().32 per cent. 
In addition to these facts we are also led by analogy to infer that 
we have in the scutes of the nine bands of armor a variant which 
is inherited in the blended fashion. This inference is borne out 
by the evidence derived from an examination of the mothers 
of the various sets of quadruplets (table 6). Of the fathers we 
unfortunately know nothing. 
B. Comparative variability of males and females 
While the main mass of our statistical data came from an exam- 
ination of baskets made from the dried and shaped shells, a con- 
siderable number of individuals were identified as to sex. The 
arrays dealt with in this section consist of the scute counts of 
animals shipped to us alive, and of the advanced foetuses in our 
collection. The larger number of females is explained by the 
fact that our first interest was centered on the facts of develop- 
ment and hence we ordered only pregnant females. The number 
of individuals is, however, sufficiently large to furnish a basis 
of comparison between the sexes. Table 3 indicates the frequency 
distribution of the variates of the two sexes. It will be seen at a 
glance that the array of males here presented is decidedly more 
variable than that of the females. Part of the disparity may be 
due to the occurrence of one set of foetuses, all of which have a 
scute count of over 600; but even if we arbitrarily exclude this 
aberrant set we do not materially reduce the variability of the 
