phenomenon. There is one consequence of the emission theory to which I should like briefly to allude, because I think it is more in accordance with the actual properties of light than the view to which we should be led if we took the electromagnetic theory in the form in which it is usually presented. The active agents on the emission theory are discrete particles, a ray of light consisting of a swarm of such particles, the volume occupied by these particles being only a very small fraction of the volume through which they are distributed. The front of a wave of light would on this view consist of a multitude of small bright specks spread over a dark ground; the wave front in fact is porous, and has a structure. Now on the electric theory of light as usually given, it is tacitly assumed that the electric force is everywhere uniform over the wave front, that there are no vacant spaces, and that the front has no structure. This is no necessary part of the electric theory, and I think there is evidence that the wave front does in reality much more closely resemble a number of bright specks on a dark ground than a uniformly illuminated area. Let me mention one such piece of evidence. If a flash of light, especially ultra-violet light, fall on a metal surface, negatively electrified corpuscles are emitted from the surface; but when we measure, as we can do, the number of these, we find that only a most insignificant fraction of the number of molecules passed over by the wave front have emitted these corpuscles. If the wave front were continuous then all the molecules of the metal exposed to the light would be under the same condition, and although, like the molecules of a gas, the molecules might possess very different amounts of kinetc energy, this difference would be nothing like sufficient to account for the enormous discrepancy between the number of molecules struck by the light and those which emit corpuscles. This discrepancy would, however, easily be understood if we suppose that the wave front is not continuous but full of holes, so that only a small number
