provided with the aluminium screen is rendered highly incandescent, while the filament in the other bulb may, with the same potential, not even come to redness, although in reality the latter bulb takes generally more energy than the former.  When they are both connected together to the terminal, the difference is even more apparent, showing the importance of the screening.  The metal tube placed on the stem containing the leading-in wire performs really two distinct functions:  First:  it acts more or less as an electrostatic screen, thus economizing the energy supplied to the bulb; and, second, to whatever extent it may fail to act electrostatically, it acts mechanically, preventing the bombardment, and consequently intense heating and possible deterioration of the slender support of the refractory incandescent body, or of the glass stem containing the leading-in wire.  I say slender support, for it is evident that in order to confine the heat more completely to the incandescing body its support should be very thin, so as to carry away the smallest possible amount of heat by conduction.  Of all the supports used I have found an ordinary incandescent lamp filament to be the best, principally because among conductors it can withstand the highest degrees of heat.

The effectiveness of the metal tube as an electrostatic screen depends largely on the degree of exhaustion.

At excessively high degrees of exhaustion—­which are reached by using great care and special means in connection with the Sprengel pump—­when the matter in the globe is in the ultra-radiant state, it acts most perfectly.  The shadow of the upper edge of the tube is then sharply defined upon the bulb.

At a somewhat lower degree of exhaustion, which is about the ordinary “non-striking” vacuum, and generally as long as the matter moves predominantly in straight lines, the screen still does well.  In elucidation of the preceding remark it is necessary to state that what is a “non-striking” vacuum for a coil operated, as ordinarily, by impulses, or currents, of low-frequency, is not, by far, so when the coil is operated by currents of very high frequency.  In such case the discharge may pass with great freedom through the rarefied gas through which a low-frequency discharge may not pass, even though the potential be much higher.  At ordinary atmospheric pressures just the reverse rule holds good:  the higher the frequency, the less the spark discharge is able to jump between the terminals, especially if they are knobs or spheres of some size.

Finally, at very low degrees of exhaustion, when the gas is well conducting, the metal tube not only does not act as an electrostatic screen, but even is a drawback, aiding to a considerable extent the dissipation of the energy laterally from the leading-in wire.  This, of course, is to be expected.  In this case, namely, the metal tube is in good electrical connection with the leading-in wire, and most of the bombardment is directed upon the tube.  As long as the electrical connection is not good, the conducting tube is always of some advantage, for although it may not greatly economize energy, still it protects the support of the refractory button, and is a means for concentrating more energy upon the same.