IX— X] 



Introduction 



xxv 



Values of the "generalised probable error" coefficients are given in Table X 

 for n = l to 11, and by means of a smooth curve the results may probably be 

 extended to ?! = 1 5. The values found for this extension are : 



Illustration (i). Let us consider long bone data for Frenchmen. 1 = F= femur, 

 2 = H = humerus, 3 = T= tibia, 4 = R = radius*, then by formula (xviii) p. xxiv : 



R = 



Further in cms : 



1, -8421, -8058, -7439 



•8421, 1, -8601, -8451 



•8058, -8601, 1, -7804 



■7439, -8451, -7804, 1 



to, = 45-23, o-, = 2-372, 

 to 2 = 33 01, a 2 = 1-538, 

 m, -36-81, 0-3 = 1-799, 

 to 4 = 24-39, o- 4 =l-170. 

 What is the chance that the following individual may be considered French ? 



J" = 36-97, #' = 26-82, T'= 3056, i?' = 20-68. 

 The deviations in terms of their standard deviations are : 



x x = (F' - to,)/o-, = - 3-482, x, = (//' - ?« 2 )/o- 2 = - 4-059, 

 x 3 = (T - mj/tr, = - 3-474, g> 4 = (R' - to 4 )/o- 4 = - 3-171. 



Further : 



R a 



s =3-7810, -g? = 6-5496, 



R 



R. 2l 

 R 



I! 



R 

 ft 



4-3406, 



^ = 3-6508, 



^23 



= 20231, ^- 3 = 11404, ^ = 02130, ^=21946, 



= 2-3175, 



-74 

 R 



= 0-6842. 



Whence 



X 2 = 16741,035 and x = 40916, 



n is even, hence: I JN = ^iM 4 ^! 6 ) = ^2 



■n-.m,, 



* For particulars of these length measurements the reader must consult R. S. Proc. Vol. 61, 

 pp. 343 et seq. and Phil. Trans. Vol. 192, A, p. 180. 



B. d 



