74 



Records of the Indian Museum. 



[Vol. XXIII, 



duakd by the Gaussian curve. We can now safely apply the theory 

 oi Errors (which is based on the Gauss-Laplacian Probability Integral) 

 to judge the likelihood of deviations from the Mean. 



Anglo-Indians V= 4*0672 



Average V = 3*5700 



Anglo-Indian Difference = '4972 

 Now the S.D of V= 4 '5450 

 Thus, P.E. of 7= ±3676 

 Anglo-Indian Difference = '497 2 ±'3^>7 b 



D 4972 



X~- = — — - ='QI 

 " '5450 



From Biometric Table II, |(i+a) = -8i 85 88 



J(l— a) = *l8 14 12 



Thus we find that no less than 18*14% of " homogeneous ' 

 races will have larger Coefficients of Variation than Anglo-Indians. 

 The Anglo-Indian Coefficient of Variation is not significantly 

 greater than the average Coefficient of Variation of the whole series. 



(b) Selected Series. 



We obtain the following distribution of the Coefficients of 

 Variation for 55 selected 1 races (unit of grouping =*2). 



Distribution of jj selected Coefficients of Variation. 





27 



2-9 



3'i 



33 



3"5 



37 



3 9 



4'1 





• '.roup 



to 



to 



to 



to 



to 



to 



to 



to 



Total. 





2-9 



3-i 



3'3 



3*5 



37 



3-9 



4'i 



4*3 





Frequency 



3 



5*5 



1 '5 



9-0 



17-5 



9*5 



4-0 



50 



55 



We get, 



Mean Coefficient of Variation 



Standard Deviation of Coefficient of Variation 



P.K. of Mean V 



The other constants are: — 



JU = 3 22 2() 45 



fj ? = 1 59 62 01 



^=29*89 12 I 



0, = -06 61 53 

 ft = 2 '97 93 



= 357' 



- '359° 

 = 2421 



• It will be noticed thai the extreme values nave been automatically excluded 

 li\ our principle ^rejection of unreliable value 



