228 



EEPOET — 1888. 



line represent the numbers of coins of each date existing in the circulation at the- 

 end of the most recent period designated 1. At that time let exportation act for a. 

 short period. Its ravages falling unequally on columns 1, 2, 3, &c., the elevation of" 



the hroken lines will represent the new distribution of proportions. Call the new 

 heights h\, h\, &c., the old heights being Aj, h„, &c. At the beginning of the short 



period of exportation at least -— of the initial exportation falls upon column 1. At 



the end of the period at least ^^ of the later drains are taken from that column. 



Sh 



Now — is greater than — Hence for the ivhole period at least —- , of the total 



all 



Sh 



Sh' 



h' 



exportation falls xipon column 1 ; -—, being the proportion of coins dated 1 exist- 

 ing in the circulation at the end of the period of exportation- 

 It is thus that Jevons argues, or must be supposed to argue : that, as the couiage 

 of 1863-64 constituted in 1868 about one-fifth of the then existing circulation, 

 therefore at least a fifth of the exportation during the period 1865-67 must have 

 fallen upon the coinage of 1863-64. 



Part II. — Oeiticism and Correction. 



In criticising this method we may adopt the same order as in expound- 

 ing it, and separately consider the reasoning which constitutes the first 

 and the second approximation. (1) The fundamental position of the 

 argument is undoubtedly sound, provided that the proportions derived 

 from the inspection of sample collections are sufiBciently accurate. For 

 the validity of these data (in the case of one of them at least) we have 

 the high authority of Jevons. ' I feel certain,' he says, ' after drawing up 



