DECLINE OF DETERMINISM EDDINGTON 147 



lunacy commissioners) has seen to it that the correspondence is 

 sufficient for practical needs. But we cannot rely on the corre- 

 spondence, and in physics we do not accept any detail of the picture 

 unless it is confirmed by more exact methods of inference. 



The external world of physics is thus a universe populated with 

 inferences. The inferences differ in degree and not in kind. Famil- 

 iar objects which I handle are just as much inferential as a remote 

 star which I infer from a faint image on a photographic plate or an 

 " undiscovered " planet inferred from irregularities in the motion 

 of Uranus. It is sometimes asserted that electrons are essentially 

 more hypothetical than stars. There is no ground for such a distinc- 

 tion. By an instrument called a Geiger counter electrons may be 

 counted one by one as an observer counts one by one the stars in 

 the sky. In each case the actual counting depends on a remote 

 indication of the physical object. Erroneous properties may be 

 attributed to the electron by fallacious or insufficiently grounded 

 inference, so that we may have a totally wrong impression of what 

 it is we are counting ; but the same is equally true of the stars. The 

 rules of inference are the laws of physics; thus the law that light 

 travels in straight lines enables us to infer the location of distant 

 objects ; and so on. In fact a law of physics can be used either way — 

 to predict an effect from a cause or to infer a cause (i. e., a physical 

 object embodying certain properties) from an observed effect. 



In the universe of inferences, past, present, and future appear 

 simultaneously and it requires scientific analysis to sort them out. 

 By a certain rule of inference, viz, the law of gravitation, we infer 

 the present or past existence of a dark companion to a star; by an 

 application of the same rule of inference we infer the existence on 

 August 11, 1999, of a configuration of the sun, earth, and moon, 

 which corresponds to a total eclipse of the sun. The shadow of the 

 moon on Cornwall in 1999 is already in the universe of inference. 

 It will not change its status when the year 1999 arrives and the 

 eclipse is observed; we shall merely substitute one method of infer- 

 ring the shadow for another. The shadow will always be an infer- 

 ence. I am speaking of the object or condition in the external 

 world which is called a shadow; our perception of darkness is not 

 the physical shadow, but is one of the possible clues from which its 

 existence can be inferred. 



Of particular importance to the problem of determinism are our 

 inferences about the past. Strictly speaking our direct inferences 

 from sight, sound, touch, all relate to a time slightly antecedent; 

 but often the lag is more considerable. Suppose that we wish to 

 discover the constitution of a certain salt. We put it in a test tube, 

 apply certain reagents, and ultimately reach the conclusion that it 



149571—33 11 



