IQ22,] P. C. M ahala nobis : Analysis of Stature. 63 



Risley l published the crude measurements of 87 Indian castes 

 and tribes, but he did not calculate a single frequency constant 

 or a single probable error. The size of sample varies from 185 to 

 2, yet eve^ average has been given equal weight on the strength 

 of his authority. The averages published in his book were in many 

 cases hopelessly wrong, in one instance the difference amounted to 

 no less than 60 mm. 



I have just finished calculating the irequency constants for the 

 whole of Risley data for Stature. I hope to publish my results 

 at an early date. Meanwhile I shall use my summary table for 

 purposes of comparison in this paper. 



It should be noted that the present section was already sub- 

 mitted to the press when the Mediterranean data reached me. 

 The Risley data also had not then been reduced. Thus the 

 earlier part of the present section does not include the above two 

 series of data. I have retained a portion of the older work, but 

 have gone over the whole ground again with the inclusion of the 

 new data. 



The Caste data of Risley is substantially differentiated from 

 other samples in showing a significant lower Variability, hence 

 the Anglo-Indian sample is found to be significantly more 

 variable than the Indian Castes and Tribes. Otherwise the 

 inclusion of the new data does not upset the earlier conclusion 

 that the Anglo-Indian Variability, though higher than the general 

 Variability of f ' homogeneous " races, is not significantly different. 

 As a matter of fact Anglo-Indian Variability is just about the 

 same as the Variability of European (in a geographical sense only) 

 races. 



Note on the Retention of Criminal Data. 



It may be objected that a criminal population being substan- 

 tially differentiated from the general population, it is not legitimate 

 to use criminal data for comparative purposes. We can only reply 

 that if there is any fundamental anthropological differentiation 

 this has not }^et been proved to be the case. On the other hand 

 the bulk of available statistical evidence goes to show that there is 

 no such thing as a different criminal type. J. J Craig 2 says of his 

 Egyptan data, ff it may be objected that criminality in itself is a 

 determining factor of selection, but the objection does not hold in 

 Egypt " and he proceeds to explain why. In the case of New 

 South Wales also the same is true. There is no significant differen- 

 tiation of criminals from the general population. 8 



As regards the English convict, we need only refer to the 

 great work on the subject by Dr. Charles Goring (already cited 

 several times in this paper). Goring comes to the conclusion that 

 the Iyombrosian doctrine of criminal types is false. " Criminals as 



''Indian Castes and Tribes" 2 Vols. (1904?) (Superintendent of Govern- 

 ment Printing, Calcutta). 



°- J. I. Craig: loc cit.Biom. Vol. 8 (191 1). 



5 Goring: "The English Convict," (1913), p. 198. 



