ON THE NATURE OF INSURANCE. 251 



and which has hitherto been carefully guarded in the 

 deduction of office results. But there is a much more 

 important question behind. Suppose the calculator had 

 undoubtedly succeeded in exhibiting the real law of 

 mortality, and that it were quite certain the next 

 hundred million inhabitants of Great Britain would die 

 in the manner pointed out in a table. In such a 

 case, many will say, the office may charge the real 

 premiums deduced from the table, with a very slight 

 addition for expenses of management. They may leave 

 the fluctuations to take care of themselves, and trust in 

 the long run. This assertion I now proceed to discuss. 



If the banker of a gaming-table were to follow the 

 same plan, that is, if he were to stake against ah 1 comers 

 with only just enough of advantage to cover expenses, 

 he would infallibly be ruined at last. It might not be 

 in this year nor the next, nor in this century nor the 

 next; but ruined at some time or other he must be 

 (see page 110, and also Appendix I.). If the case 

 of the office managers were precisely analogous to that 

 of the bankers of the gaming-table, I would repeat 

 with as much confidence of the former what I have said 

 of the latter. But, in the first place, the fluctuations of 

 mortality are not, by very much, so great as those which 

 take place in the assortment of cards, nor even so great 

 as those which take place in harvests, in the price of 

 provisions, &c. This is much in favour of the insurance 

 office; but who can say how much? 



In the second place, the fluctuations of mortality 

 have of themselves a tendency to create opposite fluc- 

 tuations. Thus, a very sickly season carries off the weak, 

 and deprives the succeeding years of those who were 

 most likely to have died; causing, therefore, a season of 

 remarkable health. This is a very important item in 

 the theory of the fluctuations of mortality, and there is 

 nothing similar to it in the case of the gaming house. 

 It reduces annual fluctuation itself to a species of regu- 

 larity, and is, perhaps, the sufficient reason for the slight- 

 ness of the total fluctuations. 



