78 



Records of the Indian Museum, 



[Vol XXIII, 



Thus as belore the Anglo-Indian sample does not seem to be 

 significantly more variable than homogeneous samples. About 

 16% of homogeneous samples will have a greater Variability. 



(b) Selected Series. 



Let us now select samples greater than 25. We get a total 

 (omitting different age-groups) of 67 samples distributed as fol- 

 lows 



Distributions of y Selected Coefficients 0/ Variations. 



• iroup 



Be- 

 yond 



270 



-2-80 



-2-90 



-300 -3-10 



-3-20 



-3-30 



-3-40 



-3-50 



-3-60 



Frequency . . 



2 



1 



1 



5 1 







2 



5 



15 



7 



Oroup 



-370 



-3-80 



-3 90 



-4*00 j -4' 10 



-4- 20 



-4'3o 



Total. 







Frequency . . 



6 



8 



4 



1 4 



1 



1 



67 





We get, /', = 12-97 78 82 



/'s = x 7'77 4069 



^4=4777° 38 55 

 giving tf,= -14 43 94 + -I2 47 



0a= 2-83 53 67 ±'33 26 



Graduation by the "normal" curve is thus possible and \\v 

 ire justified in using the ''normal" Probability Integral 

 Mean Value of Coefficient of Variation =3'5 8 43±'°297 



Standard I deviation of Co-efficient of Variation = -3602 + -0210 

 It will be noticed that the Mean Value 3*584 is sensibly the 

 Line as we had obtained without including this Mediterranean 

 -lata e.e. 3571. The difference is only '013 while the probable 

 error is certainly greater than '03. Thus 3-58 may be safeK 

 taken as a standard value for the Coefficient of Variation for 

 lure- of homogeneous non-Caste samples. 

 The mean value lor the whole series 3W>5.> is smaller than 

 the mean value tor selected Samples, 3*5843, because in small 



samples the dispersion i^ more likely to be smaller. 1 



I,i.t ns now compare the Anglo-Indian Variability with the 

 above Mean Variability , 



Anglo-Indian Coeff. <■! Variation 4-01. 72 



Mean Selected Coeff. of Variation 358 4.; 



Anglo Indian 1 >iff< ivne- 



= 04«S 20 



1 >n i'i the dependence ■>! Standard Deviation on the w» 



; 10(1915) p. 5?2 and Vol. 11 (1916) p. 277. 



