XXXVIII XLI] 



Introduction 



CXCMI 



Thus vi - 3759,224, i/,' = 25-809,801. 



Then from formula (xiii) by continuous process on the machine*: 

 ,' - -5*730,1507, a z ' = -7125,3970, 



,, 3 ' = _ -761)7,52 1 !>:>, <> 4 ' = -5328,4008, 



5 ' = -'1751,1)766. 

 In the same way from formula (xiv) we have: 



V = -0357,51)1X1), 

 V = '0138,6077, 

 6 6 ' = -0049,6656. 

 With these values we find: 



- 3 V 03' + 

 ifo V + 



6 2 ' - - 0048,22!t!i, 

 /V = - -0126,6980, 



= - -0466,8795- 

 = '0029,6085- 

 = '0057,6163, 

 Tk(^'-T5iiV) = - '0000,3601. 

 Hence by (vii) and (viii) we deduce: 

 in' = 3-759,224 - -046,688 + -002,9608 



= 3-715,497, 

 fjiz' = 25-809,801 - -0833,3333- -0057,6163 - -0000,3601 + "0710,6040 



= 25-791,730, 

 o- 2 = fi 2 ' -/V 2 = H'986,812. 



Thus Mean = 3-715,497 months = 1 13'09 days. 



Standard Deviation = 3-462,198 months = 105-38 days. 



We may now put our results together for the guidance of the reader in dealing 

 with similar problems. 



Mean and Standard Deviation of Infantile Deaths in Days. 



It will be clear from these results that the non-asymptotic-correction is better 

 than the asymptotic, but both are better than that obtained by using Sheppard's 

 correction alone, which in this case changes the variability in the wrong direction. 



* It is well to add all the positive terms before subtracting the negative ones in machining. 



