xxii Tables for Statisticians and Biometricians [V — VI 



Hence using formula (ii) p. xiii : 



(- log 1")= 252-9o315 + -31 [14-99573] 



_ -31 X -9039 ^ .^3333 _ 69x^5239 ^ .^33^^ 

 6 6 



= 252-95315\ 



+ 4-64868- =257-55542. 

 - -04G4ll 

 Hence \QgF=- 257-55542 = 258-44458, 



i?'= 2-7834/10"-^, 

 which measures the improbability required. 



Table V (pp. 12—18) and Table VI (p. 18) 



Probable Errors of Means, Standard Deviations and Coefficients of Vai-iation. 

 (Table V calculated by Winifred Gibson, B.Sc; Table VI by Dr Raymond Pearl 

 and T. Blakeman, M.A. Biometrika, Vol. iv. pp. 385 — 393.) 



If m be a mean, o- a standard deviation and T'^=100o-/»i a coefficient of 

 variation, for a population of n, we have 



Probable Error of Mean 



= -6744898a-/Vn = Xicr (xi), 



Probable Error of Standard Deviation 



= -6744898.o-/V2^ = x-,°" (^^i)> 



Probable Error of the Coefficient of Variation 



= -6744898 Fx|l + 2 (^^yiyv^ (xiii), 



= -6744898/V2n x i/r 



= X-^^ (^iv)- 



Table V gives Xi ^d*^ X"- f°i' ^^^^ value of n up to 1000, Table VI gives i^ for 

 each value of V proceeding by units from to 50. 



When the frequency n is greater than 1000, the tables may still be used by 

 taking out a square factor, which can be divided out at sight. 



Illustration (i). n = 2834 = 4 x 708-5. 



n = 708, xi = 02535 ; n = 709, x. = '02533. 



.-. « = 708-5, xi= 02534, and .-. for » = 2834, 



we have x^ ~ 01207. 



Illustration (ii). In the case of the 900 Bavarian crania of the Illustration (iii) 



to Table II the values 



w = 83-069, o- = 3-432, 



