378 MR. S. M. JACOB ON 









Vioo/ 



(— - 



V too/ 



y - n y { J^ ] _ ¥ 



IOO> 



which gives 





I 

 I 



v(-)-*/ 



Wioo/ ' 



where v' is the measured coefficient of variation 



i Viooy i 



! VlOO/ 



I+£ a )1 



Vtoo/ 



100 



This gives the true coefficient of correlation in terms of the measured values and 

 the known 'standard ' error k. 



In the cases dealt with here, for the Kharif harvest v' has a typical value of 20, 

 and for the Rabi harvest 40. 



Now if we put k = I this would be equivalent to saying that in the estimation of 

 matured areas of crop the probable error was about 8*5 %. 



This is perhaps somewhat larger than might be met with in practice, but I fear, 

 it is a not very unreasonable value to adopt. 



Accordingly, for the Kharif harvest r / = i*35*f approximately. 



And for the Rabi harvest r' = i - i2 r approimately. 



Thus the coefficients of -5 and 7 would be increased to "675 and 784 respec- 

 tively. Of course no stress can be laid on particular figures till we are in a position 

 to say what the percentage error of measurement actually is, but they indicate the 

 type of correction which would be applicable, and point to higher values of the 

 correlation coefficient being obtained with increased accuracy of measurement. 

 Further, the assumptions made here must be borne in mind. At best they can be 

 only partial representations of truth, as it is very probable, for instance, that the 

 actual errors made in measurement have a skew distribution. 



