244 FRESNEL. 



ble extent, if not the absolute, yet the relative velocity 

 of a ray ; it is to make observations on it when, in the 

 annual course of the earth, its motion is directed towards 

 the star from which the ray proceeds, and again when it 

 is diametrically in the opposite direction. In the former 

 case, it is as if the velocity of the ray was increased by 

 the whole of that of our globe ; in the second, the nu- 

 merical difference has the same amount, but the velocity 

 is diminished. Now no one is ignorant that the velocity 

 of the earth's revolution is quite comparable with that of 

 light ; being in fact about a ten-thousandth part of it. 

 And again ; to observe a star, towards which the earth 



O 



is moving, and then one from which it is receding, is to 

 operate upon rays whose velocities differ from each other 

 by one five-thousandth part. Such rays ought to be 

 unequally refracted ; the theory of emission furnishes 

 the means of expressing in numbers the amount of 

 the inequality ; and we may easily see that it will far 

 exceed the small errors of observation. Now precise 

 measures have completely negatived such calculation ; 

 the rays proceeding from all stars, in whatever region 

 they are situated, undergo precisely the same refraction. 



The disagreement between this theory and experience, 

 could not be more manifest, and from that moment the 

 system of emission seemed to be overturned from its very 

 foundations. Nevertheless, this definitive sentence has 

 been suspended by the aid of a supposition which I can 

 explain in two words ; it consists in admitting that 

 incandescent bodies emit rays with all sorts of velocities, 

 but that a special and determined velocity is necessary 

 to make them rays of light. If a ten-thousandth part of 

 increase or diminution in their velocity takes away from 

 rays their luminous properties, the observed equality of 



