72 
Proceedings of the Royal Society of Edinburgh. [Sess. 
consequent variations in the members and the range of the different series 
in Table II. 
The comparison is then made by calculating the coefficient of correlation 
between the positions as given by each measurement or index in turn, and 
the positions as deduced from the composite series. 
The maximum value of unity for any particular coefficient would indi- 
cate that the corresponding measurement or index would give an order pre- 
cisely the same as the composite order deduced from the consideration of all 
these measurements and indices. A zero value would show an absence of 
relation between variation in the measurement and variation in degree of 
development. Thus, the nearer the value of the coefficient of correlation 
for any measurement or index approaches unity, the greater the value of 
that measurement or index as a criterion in determining the relative evolu- 
tionary positions of the various crania. 
In the column headed r in Table IV. are given the values of the various 
coefficients of correlation thus determined. 
A second approximation to the evolutionary order is then arrived at by 
weighting each measurement or index proportionately to the most probable 
value of the corresponding coefficient of correlation. Thus, the series in 
Table IV. are obtained by multiplying the various series in Table II. by the 
corresponding coefficients of correlation. The subsequent procedure is then 
the same as with the latter series. 
A comparison of this new composite order in Table IY. with the first 
approximation in Table II. will show distinct modifications in the relative 
intervals, though the order is unaltered, except that Yeddahs move up to 
equality with Europeans, or to a position slightly in advance. 
These new values for the composite relative positions necessitate, how- 
ever, a recalculation of the coefficients of correlation, so as to arrive at a 
second approximation to the values of the various measurements and 
indices as criteria. 
Table V. gives the necessary adaptations of the new order for this re- 
calculation, these various series being compared with the original series of 
Table II. and the coefficients of correlation calculated as before. 
These new values are given under the heading r in Table VI. 
With these new values a further step has been taken in determining a 
third approximation to the order of the various types. A comparison of 
this approximation with the second shows only very small variations. 
Here, then, the process ends. 
The high position of the Cannstatt skull is discussed by Berry and 
