If the remaining rays are allowed to fall on a glass plate silvered and divided by a diamond fixed on a dividing machine into small rectangles of equal dimensions, there will be observed variations in the reflecting power according to the orientation of the rectangles, under conditions entirely comparable with the experiment of Garbasso.
In order that the phenomenon be produced it is necessary that the remaining waves should be previously polarized.

This is because, in fact, the mechanism employed to produce the electric oscillations evidently gives out vibrations which occur on a single plane and are subsequently polarized.
We cannot therefore entirely assimilate a radiation proceeding from a spark-gap to a ray of natural light.

For the synthesis of light to be realized, still other conditions must be complied with.

During a luminous impression, the direction and the phase change millions of times in the vibration sensible to the retina, yet the damping of this vibration is very slow.

With the Hertzian oscillations all these conditions are changed--the damping is very rapid but the direction remains invariable.
Every time, however, that we deal with general phenomena which are independent of these special conditions, the parallelism is perfect; and with the waves, we have put in evidence the reflexion, refraction, total reflexion, double reflexion, rotatory polarization, dispersion, and the ordinary interferences produced by rays travelling in the same direction and crossing each other at a very acute angle, or the interferences analogous to those which Wiener observed with rays of the contrary direction.
A very important consequence of the electromagnetic theory foreseen by Maxwell is that the luminous waves which fall on a surface must exercise on this surface a pressure equal to the radiant energy which exists in the unit of volume of the surrounding space.