*4 



Records of the Indian Museum. 



[Vol. XXIII, 



From Biometric Table II, Id +a)= 99 32 443 



l(i-a)= -0067557 



Thus only about 7 in 1000 will have greater variability. 



Again, 



Selected Non-Caste S.D. =60*6563+ 5453 mm. 



Selected Caste S.D. =53 8 ±'4429 

 Anglo-Indian Difference = 6*8563 + 7025 



Selected Caste Variability is thus significantly greater. 



We conclude from our comparative study of variabilities that 

 Anglo-Indian Variability though high is not sufficiently so to enable 

 us to assert that the material is heterogeneous. The Anglo-Indian 

 sample is however markedly more variable than the Rislay Samples 

 of Indian Castes and Tribes. 



I shall now consider a series of modern European races for 

 which reliable data is available. 



Modern European Races. 



rum. 



Anglo-Indian S.D. 

 Average European S.D. 



=67*385 mm. 

 =65775 



Anglo-Indian Difference = i'6i 

 S.D. of S D. = 275 



Anglo-Indian excess in terms of S.D. 

 = i-6i 275=0-5855 



S.D. 



Aberdeen students (493; 59'4» 



Cyprus (585) 61 -6 



Cambridge students (1000) 64-6 



U.S.A. recruits (25,898: 65-6 



Albanians (140) 657 



N.S.W. Criminals (2871) 65-8 



Oxford students (959) 66 - i 



Germans (390) 66-8 



Crete (318) 67-5 



I'.ng. Criminals (300 •) 68 -i 



Eng. Fathers (1078) 687 



4. Sons (1078) 69-4 



Thus the Anglo-Indian variability is only 161 mm. greater 

 than average variability of European races. We have however 

 included no less than five different English samples. If we retain 

 tlie largest English sample ( >ooo criminals) we get the Mean varia- 

 bility to be 057,75 mm. with a S.D. of 2*513 mm. The Anglo- 

 Indian excess is 2*1 mm. or in terms of the S.D. is 0*79586. 



We conclude that Anglo-Indian Variability is of the same 

 order as modern European variability. 



Conclusions. 

 I have proposed five distinct tests of " homogeneity 



I 'flu. Frequency distribution should be homotvpic. 



II It should resisl statistical dissection; 



III Subsamplea should not differ significantly; 



IV The general nature of the distribution should be similar 

 to homo iUS distribution. 



V The Variability should not differ significantly from the 

 av< Variability of homogeneous races. 



