or thirty flashes, however, the receiver lost its power of automatic recovery, and the sensitiveness had then to be restored by tapping. An interesting observation was made to the effect that on the last occasion the receiver responded without previous tapping, a rumbling noise was heard in the telephone which lasted for a short time, evidently due to the re-arrangement of the surface molecules to a more stable condition, after which the power of self-recovery was lost.
The state of sensibility described above is more or less transitory, and is induced, generally speaking, by a somewhat unstable contact and low E.M.F. acting in the circuit. In the majority of metals, the normal tendency is towards a diminution of contact resistance by the action of electric waves. The occasional increase of resistance, in general, disappears when the pressure and E.M.F. are increased. But in the case to be presently described we have an interesting exception, where the normal state of things is just the reverse of what prevails in the majority of metals.
Alkali Metals.
In the following investigations the radiator is a platinum sphere 9·7 mm. in diameter. The coherer was placed at a short distance, so that the intensity of incident radiation was fairly strong.
Potassium.—In working with this metal, the exceptional nature of the reaction became at once evident. The effect of radiation was to produce an increase of resistance. The pressure of contact was adjusted till a current flowed through the galvanometer, the galvanometer spot of light being at one end of the scale. On subjecting the receiver to radiation the spot of light was deflected to the opposite end, exhibiting a great increase of resistance. When the pressure and E.M.F. were suitably adjusted a condition was soon attained, when a flash of radiation made the spot of light swing energetically in one direction, indicating an increase of resistance: the receiver, however, recovered instantaneously with the cessation of radiation, and the spot violently swung back to the opposite end, indicating the normal current that flows in the circuit. This condition was found to persist, the receiver uniformly responding with an increase of resistance followed by automatic and instantaneous recovery. To prevent oxidation, the receiver was kept immersed in kerosene. When the receiver was lifted from the protecting bath, it still continued to respond with an increase of resistance, but with a gradual loss of power of automatic recovery. This power was again restored on again immersing the coherer in kerosene. The receiver in vacuo, or under reduced hydrogen pressure, would have been preferred, had the necessary appliances been available.
Sodium—As we pass from potassium to the neighbouring metals, there is a gradual transition of property as regards the nature of response to electric waves. With sodium the adjustment is a little
