In an article on "The Northern Lights," in "Science for All," Vol. II., reference is made to a well-known laboratory experiment in the following words: "We take a glass cylinder, covered at the ends with brass caps, one of which is fitted with a stop-cock, which we can screw to the plate of an air-pump. To the brass caps we now attach the terminals of a powerful induction coil, but as yet we perceive no result. We now begin to exhaust the air from the cylinder, and as the exhaustion goes on we soon see a soft tremulous light beginning to play about the ends of the cylinder; and this, when the air is sufficiently rarefied, gradually extends right through the cylinder. As we continue the exhaustion these phenomena will be reversed, the light gradually dying away as the exhaustion increases. We shall at once perceive how very much this resembles an aurora on a small scale, and so we have electricity suggested to us as the agent which produces the aurora." Farther on in the same article we find that: "Aurora displays usually take place at a great height--sometimes so high as 300 miles--while their average height is over 100 miles. At such heights the air must be extremely rarefied, and we should be disposed to expect that the electric discharge could not take place through it."
Now, at the beginning of this experiment, it must be granted that light was passing freely through the glass cylinder from side to side, and also that, when the electric current was turned on, the electricity was passing freely through the air in the cylinder though it was not visible. It could not pass through the glass on account of its being a non-conductor. Then, when the air had been partially exhausted from the cylinder, and the "soft tremulous light" began to appear about its ends, it is clear that some interference with, or change in, the free passage of light through it must have been produced, both transversely and longitudinally, which occasioned the difference in the appearance of the light and caused its tremulous motion. And as change in the appearance of the light extended through the length of the cylinder as the exhaustion increased, and finally died away--light, change and all--when it approached more nearly that of an absolute vacuum, we cannot help concluding that the light disappeared because there was no medium left in the cylinder, of sufficient density at least, through which it could pass; which, of course, means that light cannot pass through a vacuum any more than electricity can.