220 Professor A. S. Eddington [Feb. 1, 



In so far as it is a philosophical theory, it is no more than a legitimate 

 and useful point of view. I now pass on to a Generalized Principle 

 of Relativity, in which we must be content at first: to be guided by a 

 natural generalization of these results, hoping later to be able to check 

 our tentative conclusions by experiment. 



If we analyze any scientific observation, distinguishing between 

 what we perceive and what we merely infer, it always resolves itself 

 into a coincidence in space and time. A physicist states that he has 

 observed that the current through his coil is 5 milliamperes ; but 

 what he actually saw was that the image of a wire thrown by his galva- 

 nometer coincided with a certain division on a scale. He measures the 

 temperature of a liquid, but the observation is the coincidence of the 

 top of the mercury with a division on the thermometer. If then we 

 had to sum up the whole of our experimental knowledge, we should 

 have to describe it as consisting of a large number of coincidences. 



A complete history of the progress of a particle consists of a 

 knowledge of its path and the time at which it occupied each point 

 of the path. The time may be regarded as an extra co-ordinate 

 corresponding to a fourth dimension, and so the whole history may 

 be summed up by a line in four dimensions representing the particle's 

 progress through space and time. We call this four-dimensional Hue 

 the world-line of the particle. Imagine that we have drawn the 

 world-lines of all the particles, light-waves, etc., in the universe ; 

 we shall then have a complete history of the universe. It will be 

 rather a dull history-book ; the Venus of Milo will be represented by 

 an elaborate schedule of measurements, and Monna Lisa by a mathe- 

 matical specification of the distribution of paint ; still they are there, 

 if only we can recognize them. I have here a history of the universe 

 — or part of it. Unfortunately I was not able to draw it in four 

 dimensions, and even three dimensions presented difficulties, so I 

 have drawn the world-lines in two dimensions on the surface of a 

 football bladder. 



A great deal is shown here which properly speaking is not history 

 at all, because it is necessarily outside experience. As we have seen, 

 it is only coincidences — the intersections of the world-lines — that 

 constitute observational knowledge ; and, moreover, it is not the 

 place of intersection but the fact of intersection that we observe. I 

 am afraid the tAvo-dimensional model does not give a proper idea of 

 this, because in two dimensions any two lines are almost bound to 

 meet sooner or later ; but in three dimensions, and still more in four 

 dimensions, two lines can and usually do miss one another altogether, 

 and the observation that they do meet is a genuine addition to 

 knowledge. 



When I squeeze the bladder the world-lines are bent about in 

 different ways. But I have not altered the history of the universe, 

 because no intersection is created or destroyed, and so no observable 

 event is altered. The deformed bladder is just as true a history of 



