I have here a short and wide tube which is exhausted to a high degree and covered with a substantial coating of bronze, the coating allowing barely the light to shine through.  A metallic clasp, with a hook for suspending the tube, is fastened around the middle portion of the latter, the clasp being in contact with the bronze coating.  I now want to light the gas inside by suspending the tube on a wire connected to the coil.  Any one who would try the experiment for the first time, not having any previous experience, would probably take care to be quite alone when making the trial, for fear that he might become the joke of his assistants.  Still, the bulb lights in spite of the metal coating, and the light can be distinctly perceived through the latter.  A long tube covered with aluminium bronze lights when held in one hand—­the other touching the terminal of the coil—­quite powerfully.  It might be objected that the coatings are not sufficiently conducting; still, even if they were highly resistant, they ought to screen the gas.  They certainly screen it perfectly in a condition of rest, but not by far perfectly when the charge is surging in the coating.  But the loss of energy which occurs within the tube, notwithstanding the screen, is occasioned principally by the presence of the gas.  Were we to take a large hollow metallic sphere and fill it with a perfect incompressible fluid dielectric, there would be no loss inside of the sphere, and consequently the inside might be considered as perfectly screened, though the potential be very rapidly alternating.  Even were the sphere filled with oil, the loss would be incomparably smaller than when the fluid is replaced by a gas, for in the latter case the force produces displacements; that means impact and collisions in the inside.

No matter what the pressure of the gas may be, it becomes an important factor in the heating of a conductor when the electric density is great and the frequency very high.  That in the heating of conductors by lightning discharges air is an element of great importance, is almost as certain as an experimental fact.  I may illustrate the action of the air by the following experiment:  I take a short tube which is exhausted to a moderate degree and has a platinum wire running through the middle from one end to the other.  I pass a steady or low frequency current through the wire, and it is heated uniformly in all parts.  The heating here is due to conduction, or frictional losses, and the gas around the wire has—­as far as we can see—­no function to perform.  But now let me pass sudden discharges, or a high frequency current, through the wire.  Again the wire is heated, this time principally on the ends and least in the middle portion; and if the frequency of the impulses, or the rate of change, is high enough, the wire might as well be cut in the middle as not, for practically all the heating is due to the rarefied gas.  Here the gas might only act as a conductor of no impedance diverting the current