122 



distributed after the lapse of 10 years, consider fîrst what 

 will be their condition at some earlier epoch, say at the 

 end of the fifth year. 



We may admit that a certain trader A will then only 

 possess a capital of 100 £, while another may possess 

 100.000 £. 



Now if a certain cause of gain or loss comes to operate, 

 what will happen? 



For instance: let the price of an article in which both 

 A and B hâve invested their capital rise or fall. Then it 

 will be évident that, if the gain or loss of A be 10 £, 

 that of B will not be 10 £, but 10.000 £; that is to say 

 the effect of this cause will not be independent of the 

 capital, but proportional to it. 



I think that anybody will admit that it is somewhat 

 the same with the efîect of nearly any other cause. The 

 effect will not be the same for the small and the large 

 capitals. Therefore: the wealth curve will certainly not 

 be a normal curve, but a skew one. 



12. Skew-curvc machine. On the same lines as 

 G a 1 1 o n's apparatus for the normal curve, I hâve 

 devised a machine, which will illustrate the genesis of the 

 frequency-curve for the particular case that the effect of 

 the various causes is strictly proportional to the absolute 

 dimensions. 



If we might admit that ail the causes of gain and loss 

 are proportional to the total capital, then this curve would 

 give an exact représentation of the wealth distribution. 

 Probably it will give some rude approximation. The appa- 

 ratus was constructed under the supervision and according 

 to the directions of Prof. M o 1 1 and is embodied in the 

 collection of instruments belonging to the botanical labora- 

 tory of our university. Those who are familiar with 

 Gai to n's apparatus will readily understand the présent one. 



Like G a 1 1 o n's apparatus it is (see fig. 7) a frame 



