In order to solve our problem, therefore, we have to observe in complete darkness stars which are nevertheless near the edge of the sun’s disk. Is that impossible? No. Nature has met our need by providing total eclipses of the sun which may at times be seen from various stations on the earth. At those times the bright disk is hidden for a few minutes behind the disk of the moon. Midday is turned into midnight. We see stars shine out close to the masked face of the sun.
Fortunately, a total eclipse, visible in Africa and South America, was due on May 29, 1919, shortly after Einstein had, on the strength of an argument like that we have just expounded, announced the deviation of the light of the stars when it passed the sun.
Two expeditions were organised by the astronomers of Greenwich and Oxford. One proceeded to Sobral, in Brazil, the other to the small Portuguese island Principe, in the Gulf of Guinea. Some of the English astronomers were rather sceptical about the issue. How could we, until it was proved, admit that Newton was wrong, or had at least failed to formulate a perfect law? But this was proved, and very decisively, by the observations.
These observations consisted in taking a certain number of photographs during the few minutes of total eclipse of the stars near the sun. They had been photographed with the same instruments some weeks before, at a time when the region of the sky in which they shine was visible at night and far from the sun. As everybody knows, the sun passes successively, in its annual course, through the different constellations of the zodiac.
If the light of the stars which were photographed were not bent out of its path in passing the sun, it is clear that their distances ought to be the same on the plates exposed during the eclipse as on the negatives taken during the night some time previously. But if the light from them were bent out of its course during the eclipse by the gravitational influence of the sun, it would be quite otherwise. The reason is as follows. When the moon rises on one of our plains, it is not round, as everybody will have noticed, but flattened at top and bottom, somewhat like a giant tangerine lifted above the horizon for some magic supper. The moon has, of course, not ceased to be round. It merely seems to be flattened because the rays which come from its lower edge, and have to pass through a thick stratum of the atmosphere before they reach us, are bent toward the ground by the refraction of the denser atmosphere much more than are the rays coming from the moon’s upper edge, which pass through a less dense mass of air. Our eyes see the edge of the moon in the direction from which its rays come to us, not in the direction from which they started. That is why the lower edge of the moon seems to us to be raised higher above the horizon than it really is. This deviation is due to refraction.
In the same way a star situated a little to the east of the sun (the rays in this case being curved by weight, not by refraction) will seem to us further away from it. It will look as if it were further east than it really is. Similarly, a star to the west of the sun will seem to us still further from the sun’s western edge.
Hence the stars on either side of the sun will, if Einstein is right, be more widely separated from each other in the negatives taken during the eclipse. In their normal position, on the photographs taken during the night, they will seem nearer to each other.
This is precisely what was found when the photographs taken at Sobral and Principe were studied with the aid of the micrometer. Not only was it thus proved that the light of the stars is bent out of its path by the sun, but it was found that the deviation had exactly the extent which had been predicted by Einstein. It amounts to an angle of one second and three-quarters (1″·75) in the case of a star that is quite close to the sun’s disk, and the angle decreases rapidly in proportion to the distance of stars from the sun. It was a great triumph for the theory of Einstein, and for the first time it gave us some connecting link between light and gravitation.
On the preceding page I compared the curvature of light owing to its weight with the deviation that is caused by atmospheric refraction. As a matter of fact, there were astronomers who wondered whether the agreement between Einstein’s theory and the results obtained during the eclipse was not merely a coincidence: whether the deviation that was recorded was not due to refractive action by the sun’s atmosphere.