398 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1964 



e' e 



v=o 



TO OBSERVER 

 E B C 



AD = L 



FA D 



NON-RELATIYISTIC 



E B 



Sin-e-= 



(b) 



FA D 



RELATIYISTIC 



(a) 



Figure 8. — (a) Appearance of moving cube, (b) Appearance of rotated cube. 



It now appears that we have all been wrong. 



Fortunately, the basic ideas have not been wrong — only the inter- 

 pretation of what we would actually observe have been mistaken. The 

 error was recently pointed out by J. Terrell (1959), and expanded on 

 by V. F. Weisskopf (1960). It is one of those embarrassing tilings 

 which appears obvious after it is pointed out to you. 



Suppose we consider a cube which is moving at right angles to the 

 observer's line of sight (fig. 8). The observer takes an instantaneous 

 photograph of the cube, so what he records is the position of those light 

 rays which reach the photographic plate at one instant of time. 



Our first impulse is to say that we would simply see one side of the 

 cube — the square ABCD. However, we must keep in mind the fact 

 that the points E and F are farther away from the photographic plate 

 than points A and B. Therefore light from B which leaves the cube at 

 a certain instant will reach the plate at the same time as light from E 

 which left the cube L/C seconds earlier. But during that time the 



