226 Variatio)i and Cort'elation of the Human Skull 
Comparing the coefficients of variation with those of the Naqada race, we find : 
Character 
English 
Naqada 
Male and 
Female 
Male 
Female 
Male and 
Female 
5-57 
7-46 
6-77 
7-34 
^: 
8-71 
7-62 
7-19 
8-11 
11-73 
9-93 
8-47 
9-40 
/■ 
6-23 
5-73 
5-31 
5-72 
From tliese figures we may conclude that the English mandibles are more 
variable than the Naqadas in all the characters except W^. 
Summing up generally for the consideration of the variability, it is clear that 
our Whitechapel skulls have very much the same general degree of variation as 
the Bavarian. There is certainly no marked difference which would allow us to 
assert that this group is less homogeneous than the Altbayern of Professor Ranke. 
It would indeed be rash to assert that they are more or less variable than the 
Naqada. Far more data must be reduced for variability before we can see clearly 
where we stand. But when once tables of variation have been made out for 30 or 
40 fairly long series of skulls we shall probably get a fairly good scale of relative 
homogeneity for many characters and so be able more easily to detect marked 
heterogeneity in cranial series. 
(10) On the Nature of the Frequencij Distrihidion. 
I will now discuss the variability of the series graphically and analyticall}', 
using the notation of Pearson's memoir on Skew Variation*. I have selected, 
for both sexes, the 12 characters dealt with by C. D. Fawcettf, and have added 
a thirteenth, viz., B', or least breadth of forehead, but I do not propose to plot 
the 26 skew curves representing the distributions of these characters, because 
in most cases the normal curve represents the distribution with sufficient accuracy 
for all practical purposes, as I shall now proceed to show. 
The following table gives the chief analytical constants of the skew curves for 
the 13 selected characters. The second column gives the number of skulls on 
which the calculation is based ; the third shows the unit in terms of which the 
2nd, 3rd, and 4th moments (/Aj, fis and |Jb^) are calculated ; after the moments follow 
columns giving V/Sj with its probable error, with its probable error, the 
difference between jS^ and the number 3, and the " criterion " 6 + 3/3i — 2/S2 with 
its probable error; the next three columns give the mean, the "mode," or value 
* Phil. Trans. Vol. 186, A, pp. 343—414. 
t Loc. cit., p. 442. 
