218 ANNUAL REPORT SMITHSONIAN INSTITUTION, 19 3 8 



and theory will undoubtedly be reformed as soon as the oversight is 

 discovered. We may say, therefore, that the practice of physics is 

 to reject the logically unobservable. 



But now, the unobservables whose rejection has caused all the con- 

 troversy belong to neither of these classes. Let us fix our attention 

 on the example of absolute simultaneity, with which the discussion 

 began. This, as we know, is rejected, and that cannot be because 

 it is practically unobservable, for physics reeks of the practically 

 unobservable. Nor is absolute simultaneity logically unobservable. 

 We can conceive that the universe might be such that two events at 

 different places might occur at the same time in an absolute sense, and 

 that this fact might be observable. I should mention, however, that 

 this has been denied, so I will presently show in more detail why abso- 

 lute simultaneity cannot be regarded as logically unobservable, but for 

 the moment I will take it to be established in order not to interrupt 

 the main course of the argument. We cannot, then, escape from the 

 conclusion that absolute simultaneity belongs to a third class of 

 unobservables, which we shall see is what I have called the class of 

 physically unobservables; and the recognition of this class commits 

 physics to the assumption of omniscience. 



To see that absolute simultaneity is physically unobservable, let 

 us look at the obstacle that prevents us from observing it. Why can 

 we not say, in an absolute sense, that two events occurring at dif- 

 ferent places are simultaneous? It comes down to the fact that we 

 can know of the events only through some agency which travels from 

 them to us and takes time to do so. Consider two events — say, the 

 impact of a meteorite on a lunar crater and the outbreak of a new star 

 in the Milky Way. We know of these events when we see them (or 

 their effects) but we do not see them at the moments at which they 

 occur because the light which makes them visible takes time to travel ; 

 and although, perhaps, we observe them on the same evening, one may 

 have occurred 100 years after the other. We can, of course, allow for 

 this by calculating how long the light has taken to travel, but when we 

 do so, according to standard methods and principles, we find that the 

 results depend on the way in which we happen to be moving with 

 respect to the bodies on which the events occurred. Furthermore, 

 we cannot distinguish in an absolute sense between one state of motion 

 and another; so far as we can determine, all states are equally valid 

 or equally invalid. Hence, we do not know what allowance to make 

 for the time of travel of the light, and, therefore, cannot determine 

 absolutely whether the events were simultaneous or not. The word 

 "absolutely" is important. We can determine quite definitely if the 

 events are simultaneous if we assume that we are at rest or that we 

 are moving in any definitely specified way; but what we cannot do is 

 to justify our assumption if it is challenged. 



