Gr. W. Robinson and W. E. Lloyd 153 



Similarly in the case of a chemical analysis, we have for six borings: 



Percentage deviation Chances of result being ivitliin 



from true value limits of such deviation 



4-4 per cent. Evens 



5-5 ,, 3 to 2 



6-3 ,, 2tol 



7-5 „ 3 to 1 



8-3 „ 4 to J 



8-8 ,, U to 2 



From these results it is seen that an accuracy of 5 per cent, in 

 mechanical analysis is ensured a probability of 4 to 1 by doing a duplicate 

 analysis on six borings. For survey purposes this is probably sufficient, 

 since it is not conceivable that variation in the amount of any fraction 

 corresponding to 5 per cent, (relative to the amount of the fraction) 

 could have any effect on the properties of a soil. 



In the case of chemical analysis it does not seem that the same 

 accuracy can be expected. For survey purposes however the accuracy 

 is probably sufficient. In the case of a critical study of one soil however 

 it would be necessary to reduce the errors much more by repeating 

 analyses and increasing the number of borings. 



(Received March 'ird, 1915. 



11—2 



