CHANCE. 75 



accidental errors arise, and we attribute them to 

 chance, because their causes are too compHcated and 

 too numerous. Here again we have only small causes, 

 but each of them would only produce a small effect ; 

 it is by their union and their number that their effects 



become formidable. 



V. 



There is yet a third point of view, which is less im- 

 portant than the two former, on which I will not lay so 

 much stress. When we are attempting to predict a 

 fact and making an examination of the antecedents, 

 we endeavour to enquire into the anterior situation. 

 But we cannot do this for every part of the universe, 

 and we are content with knowing what is going 

 on in the neighbourhood of the place where the fact 

 will occur, or what appears to have some connexion 

 with the fact. Our enquiry cannot be complete, and 

 we must know how to select. But we may happen 

 to overlook circumstances which, at first sight, seemed 

 completely foreign to the anticipated fact, to which 

 we should never have dreamed of attributing any 

 influence, which nevertheless, contrary to all anticipa- 

 tion, come to play an important part. 



A man passes in the street on the way to his 

 business. Some one familiar with his business could 

 say what reason he had for starting at such an hour 

 and why he went by such a street. On the roof a 

 slater is at work. The contractor who employs him 

 could, to a certain extent, predict what he will do. 

 But the man has no thought for the slater, nor the 

 slater for him ; they seem to belong to two worlds 

 completely foreign to one another. Nevertheless 

 the slater drops a tile v/hich kills the man, and we 



