1922.] P. C. Mahalanobis : Analysis of Stature. 85 



(1) I have shown that graduation by the Gaussian (possibly 

 still better by a Type IV curve) is adequate. Anglo-Indian fre- 

 quency distribution is certainly homotypic. Our first test thus fails 

 to show an} 7 sign of heterogeneity in the material. 



(2) Excepting for a very special type of dissection (which is 

 probably a peculiar feature of the particular sample considered) 

 statistical analysis into component groups is not possible. Our 

 second test too fails to detect heterogeneity. 



(3) We have seen that the difference between subsamples is 

 statistically insignificant. Subsamples seem to agree quite well, 

 thus confirming statistical homogeneity of the material. 



(4) The general nature of Anglo-Indian frequency distribution 

 is also similar to other homogeneous distribution. Anglo-Indian 

 distribution is approximately Gaussian with some tendency towards 

 type IV, lepto-kurtosis and small asymmetry. Other known cases 

 of stature distribution show the same characteristics. The fourth 

 test thus supports the view that the present material is homo- 

 geneous. 



(5) I have compared the Variability of the Anglo-Indian with 

 Variabilities of other races in many different ways. 



Anglo-Indians are more variable than the Indian Castes and 

 Tribes but the Variability of the Angio-Indian sample is not signi- 

 ficantly greater than the average Variability of homogeneous 

 samples in general. 



