220 ANNUAL REPORT SMITHSONIAN INSTITUTION, 19 3 8 



learning of an event at the moment of its occurrence. If, then, two 

 separated events occur at the same instant, it is conceivable that we 

 should observe them at the same instant. Now no one has denied 

 that simultaneity of observation is significant; it is only simultaneity 

 of occurrence of the events that is in question. Hence, since simul- 

 taneity of occurrence is conceivably deducible from simultaneity of 

 observation, it also would be significant if the logically unobservable 

 alone were excluded. 



We can now, at last, give a rigorous form to the principle which is 

 the subject of our inquiry. It is this: that only that which would be 

 observable if we were able to use known means of observation to the 

 known limits of their possibilities, is significant. Our description of 

 the universe must describe nothing else, must imply the existence of 

 nothing else, must imply the possible existence of nothing else. If we 

 do not accept this principle, we must reject relativity and a consider- 

 able part of the quantum theory as worthless illusion. 



The next step, clearly, is to examine the credentials of this principle 

 on general rational grounds, but before doing so I want to give an- 

 other example of its application, in order to emphasize the fact that 

 it is not an unimportant appendage of physical theory, but the very 

 mainspring of the most prominent modern developments. Heisen- 

 berg's uncertainty principle is perhaps the best-known example, but 

 I will not deal with that because it is too closely bound up with other 

 factors which there is no time to consider. I choose instead an idea 

 which stresses the point still more forcibly because it is not generally 

 regarded as exemplifying the principle in question, but is attacked or 

 defended on quite other grounds. The principle has taken root so 

 deep in the minds of physicists that they employ it unconsciously, 

 and justify their action by arguments which appear to others either 

 incomprehensible or absurd. I am speaking of the idea that the 

 physical universe is finite but boundless. This idea can be made 

 intelligible in 5 minutes when presented as an example of the prin- 

 ciple of rejection of the physically unobservable, and I believe that 

 those who accept it are convinced of its rationality because they have 

 already accepted that principle. They are not aware of this source 

 of their conviction, however, and therefore have to justify their belief 

 by saying that space is "curved," that it "bends back on itself" — an 

 "idea" which I do not think it is humanly possible to grasp except as 

 a metaphor of the kind one meets with in the "metaphysical" poetry 

 of the seventeenth century. Whether or not that psychological diag- 

 nosis is accurate, however, is unimportant; the main point is to see 

 that, in terms of our principle, the idea that space is finite and bound- 

 less is intelligible without calling on such unimaginable notions as 

 curvature. 



