11 Kill FREQUENCY AND HIGH POTENTIAL CURRENTS. 139

occurs. As regards tlie size of the bull), it is ascertained that atordinary or only slightly differing atmospheric pressures, whenair is a good insulator, the filament is heated more in a smallbulb, because of the better confinement of heat in this case. Atlower pressures, when air becomes conducting, the heating ef-fect is greater in a large bulb, but at excessively high degrees ofexhaustion there seems to be, beyond a certain and rather smallsize of the vessel, no perceptible difference in the heating.

The shape of the vessel is also of some importance, and it liasbeen found of advantage for reasons of economy to employ aspherical bull) with the electrode mounted in its centre, wherethe rebounding molecules collide.

It is desirable on account of economy that all the energy sup-plied to the brdb from the source should reach without loss thebody to be heated. The loss in conveying the energy from thesource to the body may be reduced by employing thin wiresheavily coated with insulation, and by the use of electrostaticscreens. It is to be remarked, that the screen cannot be con-nected to the ground as under ordinary conditions.

In the bulb itself a large portion of the energy supplied maybe lost by molecular bombardment against the wire connectingthe body to be heated with the source. Considerable improve-ment was effected by covering the glass stem containing the wirewith a closely fitting conducting tube. This tube is made toproject a little above the glass, and prevents the cracking of thelatter near the heated body. The effectiveness of the conductingtube is limited to verj- high degrees of exhaustion. It diminishesthe energy lost in bombardment for two reasons; first, thecharge given up by the atoms spreads over a greater area, andhence the electric density at any point is small, and the atomsare repelled with less energy than if they would strike against agood insulator; secondly, as the tube is electrified by the atomswhich first come in contact with it, the progress of the followingatoms against the tube is more or less checked by the repulsionwhich the electrified tube must exert upon the similarly electrifiedatoms. This, it is thought, explains why the discharge througha bulb is established with much greater facility when an insulator,than when a conductor, is present.

During the investigations a great many bulbs of different con-struction, with electrodes of different material, were experimentedupon, and a number of observations of interest were made. Mr.