The frequency of the vibration, and the quickness of succession of the sparks between the knobs, affect to a marked degree the appearance of the streams.  When the frequency is very low, the air gives way in more or less the same manner, as by a steady difference of potential, and the streams consist of distinct threads, generally mingled with thin sparks, which probably correspond to the successive discharges occurring between the knobs.  But when the frequency is extremely high, and the arc of the discharge produces a very loud but smooth sound—­showing both that oscillation takes place and that the sparks succeed each other with great rapidity—­then the luminous streams formed are perfectly uniform.  To reach this result very small coils and jars of small capacity should be used.  I take two tubes of thick Bohemian glass, about 5 centimetres in diameter and 20 centimetres long.  In each of the tubes I slip a primary of very thick copper wire.  On the top of each tube I wind a secondary of much thinner gutta-percha covered wire.  The two secondaries I connect in series, the primaries preferably in multiple arc.  The tubes are then placed in a large glass vessel, at a distance of 10 to 15 centimetres from each other, on insulating supports, and the vessel is filled with boiled out oil, the oil reaching about an inch above the tubes.  The free ends of the secondary are lifted out of the oil and placed parallel to each other at a distance of about 10 centimetres.  The ends which are scraped should be dipped in the oil.  Two four-pint jars joined in series may be used to discharge through the primary.  When the necessary adjustments in the length and distance of the wires above the oil and in the arc of discharge are made, a luminous sheet is produced between the wires which is perfectly smooth and textureless, like the ordinary discharge through a moderately exhausted tube.

I have purposely dwelt upon this apparently insignificant experiment.  In trials of this kind the experimenter arrives at the startling conclusion that, to pass ordinary luminous discharges through gases, no particular degree of exhaustion is needed, but that the gas may be at ordinary or even greater pressure.  To accomplish this, a very high frequency is essential; a high potential is likewise required, but this is a merely incidental necessity.  These experiments teach us that, in endeavoring to discover novel methods of producing light by the agitation of atoms, or molecules, of a gas, we need not limit our research to the vacuum tube, but may look forward quite seriously to the possibility of obtaining the light effects without the use of any vessel whatever, with air at ordinary pressure.