CHANCE. 8i 



cases. We only need to know one thing — that the 

 errors are very numerous, that they are very small, 

 and that each of them can be equally well negative 

 or positive. What is the curve of probability of each 

 of them ? We do not know, but only assume that it 

 is symmetrical. We can then show that the resultant 

 error will follow Gauss's law, and this resultant law is 

 independent of the particular laws which we do not 

 know. Here again the simplicity of the result actually 

 owes its existence to the complication of the data. 



vn. 



But we have not come to the end of paradoxes. I 

 recalled just above Flammarion's fiction of the man 

 who travels faster than light, for whom time has its 

 sign changed. I said that for him all phenomena 

 would seem to be due to chance. This is true from 

 a certain point of view, and yet, at any given moment, 

 all these phenomena would not be distributed in con- 

 formity with the laws of chance, since they would be 

 just as they are for us, who, seeing them unfolded 

 harmoniously and not emerging from a primitive 

 chaos, do not look upon them as governed by chance. 



What does this mean ? For Flammarion's imasfi- 

 nary Lumen, small causes seem to produce great 

 effects ; why, then, do things not happen as they do 

 for us when we think we see great effects due to small 

 causes? Is not the same reasoning applicable to 

 his case? 



Let us return to this reasoning. When small dif- 

 ferences in the causes produce great differences in 

 the effects, why are the effects distributed according 

 to the laws of chance? Suppose a difference of an 



a.777) 6 



