72 Records of the Indian Museum. [Vol. XXIII, 



It will be noticed that the list contains man}' small samples. 

 It will be better to omit all samples of less than 25. Doing this 

 we find that extreme values have been mostly eliminated by this 

 process of selection, showing that such extreme values were 

 probably in most cases due to uncertainty of sampling rather than 

 to any peculiarity of the population. 



I have also thought it best to exclude Scottish Insane as well 

 as the Dinka group. We have already seen that Anglo-Indian 

 variability is not significantly greater than the average variability 

 of the whole series. The inclusion of any variability greater than 

 Anglo-Indian variability will strengthen this conclusion, rejection 

 of greater variabilities will go against our conclusion. The 

 Insane is manifestly abnormal and may be neglected for the 

 present. Variability of the Dinka people is greater than that of 

 Anglo-Indians, its rejection will thus make the test more rigid. 

 Separate figures for Aswan is also omitted for similar reasons ' 



For the selected series of Standard Deviations 



Selected Mean Stand. Dev. - 59*8929 mm. 

 S.D. of Standard Dev. = £-3504 mm. 



We notice that the selected Mean is almost exactly the same 

 as the Mean for the whole series. We conclude that 60 mm. is 

 about the true average absolute V ariabiliiy of stature for human races. 

 Due to selection the S.D. of Variability is considerably re- 

 duced because the extreme values of Variability have in most cases 

 been eliminated. 



Anglo-Indian S.D. =67*385 mm. 



Selected Mean S.D. =59*893 mm. 



Anglo Indian Difference = 7 49 2 mm. 

 We find the probability : — 



x =»j„=rw 2 = rx8 

 6350 



From Biometric Tables II, i( I+ ")= "88 09 99 9 



j(i — u)= -ii 90 00 1 



Thus ri'9%of all races will have greater variabilities than 

 Anglo-Indians while 23% will differ more from the Selected Mean. 



As judged by a reliable series of standard deviations, the Abso- 

 lute Variability of Anglo-Indians is not significantly greater than 

 the Average Variability of different ,c homogeneous" samples. 



Relative Variability op Stature. 

 We shall now compare the Relative Variability (as measured 

 i>\ the Coefficient of Variation) of our Anglo Indian data with the 

 variability of samples recognised to be homogeneous, 



' J. I. ( 1. in mstrika Vol. 8 (1911), p, 70. 



