K. Pp:arson 
125 
It is very difficult to dr;iw any definite and safe conclusions from the very irregular 
distribution of variability. Taking first the Cambridge graduates, we see that the 
probable error of the standard deviation is in round numbers about "2 for the first 
three classes and '1 for the poll men in head length ; about "15 to "1 for the same 
groups for head breadth and about '1 for the first three classes and '06 for the 
poll men in the case of the cephalic index. It is difficult, on the basis of such 
probable errors, to assert any sensible differences in the class variability. Looking 
at the series as a whole, we might say with hesitation that possibly 2nd and 3rd 
class men are more variable in a very slight degree in their head measurements 
than either brilliant honours men or pass men. 
Turning to the school children we again see differences in variability which arc 
often within the probable error of the differences, but occasionally we note con- 
siderable divergences. They are difficult to account for, and they do not in any 
case run parallel with those of the Cambridge graduates. But one general result 
holds, with two exceptions out of 16 cases, the quick intelligent boys and girls are 
less variable, and very dull boys and girls more variable than the general popula- 
tion. The exceptions are the very dull boys' cephalic index and the quick intelli- 
gent girls' auricular height. Even in the latter case the variability of the very dull 
girls is sensibly greater than that of the quick intelligent girls. We may therefore 
say that with a single exception, and that within probable error limits, the quick 
intelligent are less variable than the very dull. Turning to the Cambridge grad- 
uates, we see that with the same exception — cephalic index — the 1st class men 
are less variable than the poll men. It would accordingly seem probable, that 
intellectual brilliancy is a more closely selected class than special dullness. Or, 
perhaps, it would be safer to say that intellectual power is more closely associates! 
with one physical grade than dullness, which is compatible with a wider range of 
head measurements. 
Generally it will be seen, on looking at Tables IX and X or Figs. 5 and 6, that 
the length of head is more closely associated with intelligence than the breadth, and 
the breadth than the auricular height. Thus the statement of certain anatomists, 
that the auricular height is probably the most important head measurement in 
regard to intelligence is seen to be without statistical basis. The fact that the 
girls differ from both male children and adults in the nature of the relationship 
between intelligence and cephalic index is remarkable. I have tested this result 
in several ways, for example by deducing the intelligence and cephalic index 
correlations from those of breadth and length with intelligence, but I reach the 
same conclusion that there is a real change of sign between this correlation for 
the two sexes, although some methods give the correlation in the girls' case very 
small and positive, while for the boys it remains between — '04 and — '05. 
(5) On the Relationsliip of Ability to other Pliysical and Mental Characters. 
While the relationship of ability to size and shape of the head has been shown 
to be very small, it seems worth while to compare it with the values obtained for 
