It would even be possible, if we wished, to suggest images capable of representing these contrary appearances.  Various authors have done so.  Thus, M. Boussinesq assumes that the ether behaves like a very rarefied gas in respect of the celestial bodies, because these last move, while bathed in it, in all directions and relatively slowly, while they permit it to retain, so to speak, its perfect homogeneity.  On the other hand, its own undulations are so rapid that so far as they are concerned the conditions become very different, and its fluidity has, one might say, no longer the time to come in.  Hence its rigidity alone appears.

Another consequence, very important in principle, of the fact that vibrations of light are transverse, has been well put in evidence by Fresnel.  He showed how we have, in order to understand the action which excites without condensation the sliding of successive layers of the ether during the propagation of a vibration, to consider the vibrating medium as being composed of molecules separated by finite distances.  Certain authors, it is true, have proposed theories in which the action at a distance of these molecules are replaced by actions of contact between parallelepipeds sliding over one another; but, at bottom, these two points of view both lead us to conceive the ether as a discontinuous medium, like matter itself.  The ideas gathered from the most recent experiments also bring us to the same conclusion.

Sec. 2.  RADIATIONS

In the ether thus constituted there are therefore propagated transverse vibrations, regarding which all experiments in optics furnish very precise information.  The amplitude of these vibrations is exceedingly small, even in relation to the wave-length, small as these last are.  If, in fact, the amplitude of the vibrations acquired a noticeable value in comparison with the wave-length, the speed of propagation should increase with the amplitude.  Yet, in spite of some curious experiments which seem to establish that the speed of light does alter a little with its intensity, we have reason to believe that, as regards light, the amplitude of the oscillations in relation to the wave-length is incomparably less than in the case of sound.

It has become the custom to characterise each vibration by the path which the vibratory movement traverses during the space of a vibration—­by the length of wave, in a word—­rather than by the duration of the vibration itself.  To measure wave-lengths, the methods must be employed to which I have already alluded on the subject of measurements of length.  Professor Michelson, on the one hand, and MM.  Perot and Fabry, on the other, have devised exceedingly ingenious processes, which have led to results of really unhoped-for precision.  The very exact knowledge also of the speed of the propagation of light allows the duration of a vibration to be calculated when once the wave-length is known.