cxciv Tables for Statisticians and Biometricians [XXXVIII XL I 



Clearly these equations will be simplified by taking X Q = 0, or moments about 

 the start of the frequency distribution; in this case x p =ph. Finally, if there be 

 high contact at the 6-terminal, all the // g) & are zero and we have the simple forms: 



/ii' = Vi - A/7i, a , 1*2 - Vz - A 2 ( T V + H z , a), 



t* = v,' - i*W + h * (#3, - #s,), 



Hi = vl - PV + A 4 (slo + H *,a + #4.6). 



Illustration (i). The following data are based on returns for Prussia of infants 

 dying in the first twelve months of life. They cover the years 1877 to 1881. 



Deaths per 1000 infants born. 



Here we have 



p-12, 



>?/ = -3242,9556, w' p _4 = '0471,3156, 

 n' p _ 3 = '0442,9353, 

 w'p_i = -04-20,1297, 

 M'P_I = -0380,6000, 

 ' 



=-0351,7130, 



n 2 ' = -11 44,8409, 

 n,'- -0941,6177, 



w 4 ' - -0808,8384, 

 %' = -0674,0320, 

 |> 6 ' = -0583,3164], ja = 12. 



Is this distribution asymptotic at # = 0? It is clearly abrupt, but not asymptotic 

 at x= 12. We require the mean age of infant deaths and the variability, i.e. the 

 standard deviation of infantile ages at death. 



We' as found from Table XXXVIII runs as follows: 



5 = 0-0, w,' = -0464,7275, 

 5 = 0-1, w e ' = -0465,2959, 

 5 = 0-4, w 6 ' = -055 1,9489, 

 5 = 0-5, w e ' = -0635,6643, 

 5=1-0, w e ' = -221 6,7041. 



