NEW ESSAYS 365 



men become more skilful by finding countless new con- 

 trivances, while on the other hand the stags or the hares 

 of our time do not become 36 more full of shifts than 

 those of former times. The concatenations [of ideas] in 

 the lower animals are only a shadow of reasoning, that is 

 to say they are only connexions 37 of imagination and 

 passings 7 from one image to another, because in new 

 circumstances which seem to resemble others which have 

 occurred before we 38 expect anew what we 38 at other 

 times found along with them, as if things were actually 

 connected together because their images are connected in 

 memory. It is true that reason also leads us to expect, 

 as a rule, that there will occur in the future what is in 

 harmony with a long experience of the past, but this is, 

 nevertheless, not a necessary and infallible truth ; and 

 our forecast may fail when we least expect it, because 

 the reasons which have hitherto justified it no longer 

 operate 89 . And on this account the wisest people do not 

 trust altogether to experience, but try, so far as possible, 

 to get some hold of the reason of what happens, in order 

 to decide when exceptions must be made. For reason 

 is alone capable of laying down trustworthy rules and 



36 E. reads ' are not.' 



37 E. reads ' a connexion' and l a passing.' 

 88 E. reads < they.' 



39 Cf. Lettre a laEeine Sophie Charlotte (1702) : ' For instance, though 

 we may have observed a thousand times that iron, when it is put 

 by itself on the surface of water, goes to the bottom, we have no 

 assurance that it must always be so. And without referring to the 

 miracle of the prophet Elisha who made iron to swim, we know 

 that we can make an iron pot so hollow that it floats, and that it 

 can even carry a considerable load, as do boats made of copper and 

 tin. And even abstract sciences, like geometry, afford instances in 

 which that which usually happens no longer happens. For 

 instance, we usually find that two lines which continually approach 

 one another ultimately meet, and many people will be ready to 

 take oath that this can never fail to happen. Nevertheless geo- 

 metry makes known to us unusual lines (called, on that account, 

 Asymptotes) which, if prolonged to infinity, continually approach one 

 another and yet never meet.' (G. vi. 505.) 



