54 



Life and Letters of Francis Galton 



Macdonell's values, obtained by a far more refined and accurate method, indicate 

 — especially when we remember that they are for a very different popu- 

 lation — how successfully Galton solved his problem. Doubtless he was some- 

 what aided by the fact that anthropometric physical measurements are far 

 more nearly normal than many other variates. Had his distributions been 

 more skew, his median estimates would not have given as accurately the 

 correlation coefficients. We can now compare the mean or median values 

 and the standard deviations as found from the quartiles with later results : 



Considering the difference of social class in the three series, Galton's 

 results can hardly have exception taken to them, except in the case of the 

 variabilities of Head Length and Head Breadth. These are excessive, but 

 as we have not the original tables from which the quartiles were determined, 

 it is not possible to investigate wherein they are anomalous f. 



The degree of accordance reached by Galton's process may be illustrated 

 by his tables for Stature and Knee Height : 



The figures in brackets give the numbers of individuals upon whom the 

 observed medians of the arrays were determined. It will be observed that 

 the accordance between observation and theory is again very good. 



* For a. general hospital population: Stature = 170-59 (Biometrika, Vol. iv, p. 126). 

 f Galton says "The head length is the maximum length measured from the notch between 

 and just below the eyebrows " (p. 137). Is this the glabella? 



