138 THE ANATOMY OF SCIENCE 



mechanism. We do not demand precision in such 

 a machine. It might, indeed, be so nicely tooled 

 and so uniformly operated that a coin placed 

 head-up in the machine would always land 

 head-up. This is the very opposite of what we 

 desire; we must introduce a concept alien to 

 mechanics and in a sense hostile to mechanical 

 concepts, — the idea of randomness. 



The level chance, the random distribution, 

 suggest in many ways the notion of straightness 

 in a line or evenness in a plane. Indeed, we 

 might emphasize this correlation by defining a 

 straight line in the following peculiar manner. 

 Suppose that a room contains a number of ob- 

 jects moving about in a random way. If now 

 between two fixed points in the room we con- 

 sider an imaginary line drawn, this line tvtlU 

 from time to time pass through one of the mov- 

 ing objects, and we may count the number of 

 these intersections over a long period of time. 

 The line between the two fixed points which cuts 

 the smallest number of objects during this long 

 interval of time is the straight line. However, 

 if the objects were not in absolutely random 

 motion, but were, owing to gravity, for example, 

 more frequently in the lower part of the room 

 than in the upper, then the line cutting the least 



