C. D. Fawcett 
437 
L or L' knows the other two for his population as far as the errors of random 
sampling allow him to trust his measurements*. Of course the result might be 
different could we deal with 500 to 1000 crania of one sex and race, but there is 
small hope of such numbers in the majority of cases. It is inexpedient, while 
there is so much to be done, to multiply measurements which differ by quantities 
of the order of the probable error of their determination"!". 
(9) On the Variability of the Naqada Race. 
Before we consider the actual frequency distributions and the graphical 
graduations of some of the more important characters, it may be well to examine 
the numerical values of the variations as a whole. These are given with the 
probable errors in Table XL 
We have here for the first time a moderately extensive table of the variability 
of the characters of the lium;in skull, and further the probable errors are given in 
each case so that the significant differences can be discriminated. The misfortune 
is that there is so little material available for comparative racial purposes. Only 
one or two organs have so far had their variability measured for anything like 
a long series of i-aces. The capacity of the skull has been fairly completely dealt 
with by Pearson |. We extract the following coefficients of variation for com- 
parison : 
s 
9 
s 
? 
l7th Century English 
7-68 
8-15 
Etruscans 
9-58 
8-54 
Parisian French 
7-36 
710 
Egyptian Mummies 
8-13 
8-29 
Italians 
8-34 
8-99 
Naqadas 
7-72 
6-92 
Germans 
7-74 
8-19 
Negroes 
7-07 
6-90 
Ainos 
6-89 
6-82 
Modern Egyptians 
8-59 
717 
Low Caste Panjabis 
7-24 
8-99 
Polynesians 
8-20 
5-55 
Kanakas 
7-37 
6-68 
Andamanese 
5-04 
5-59 
It will be seen at once that the Naqadas show no exceptional variability, 
but have their men and women somewhat less equal in variability than is 
the case with the Egyptian mummies or the Negroes. The Naqada values lie 
between those of the Negroes and those of the Egyptians but they are not, 
except in the case of the women to the negresses, really close to either. 
* This point had been already demonstrated for the cephalic index, see Pearson : The Chances of 
Death, Vol. i. p. 270 footnote. 
t Thus although L should generally be greater than L' and 100i?/L, 100 HjL less than 100 DjL', 
and 100 HjL', still the difference of the samples on which L or L' are taken is quite sufficient in itself 
to upset such relations. Of course it is only possible to get L' when both auricular passages and one 
or other eye-socket remain perfect ; thus we can usuallj' find L for more skulls than L'. I should 
therefore be inclined to suggest that F and L' should be dropped in future systems of measurement. 
X The Chances of Death, Vol. i, pp. 328 — 349. The short series of Naqada skulls given on p. 339 
of that work must now be replaced by the complete series given in this paper. 
