128 ESSAY ON PROBABILITIES. 



1,000,000 of times more in the same way. Hence ex- 

 perience can never, on sound principles, be held as fore- 

 telling all that is to come. The order of things,, the 

 laws of nature, and all those phrases by which we try to 

 make the past command the future, will be understood 

 by a person who admits the principles of which I treat 

 as of limited application, not giving the highest degree of 

 probability to more than a definite and limited continu- 

 ance of those things which appear to us most stable. No 

 finite experience whatsoever can justify us in saying that 

 the future shall coincide with the past in all time to come y 

 or that there is any probability for such a conclusion. 



CHAPTER VII. 



ON ERRORS OF OBSERVATION, AND RISKS OF MISTAKE. 



IN every measurement, as well as in unassisted esti- 

 mation, the observer is liable to error ; the only dif- 

 ference being that the mistakes of careful instrumental 

 measurement are likely to be less than those of esti- 

 mation. That which we call estimation means guess 

 formed by a person whose previous habits and ex- 

 perience are such as to make it very likely that he can 

 tell nearly true that which would require instruments 

 to obtain with great approach to accuracy. To illustrate 

 this distinction, imagine a certain small length, say 

 about twelve inches, to be presented to a large number 

 of persons, who are required to write on separate bits 

 of paper how many inches and tenths of inches it 

 appears to them to contain. If these persons had been 

 used to estimate lengths by the unassisted eye, it would 

 be extremely probable, 1. That the average of their 

 guesses would be very near the truth. 2. That their 

 widest limits of error would be small. If their habits 



