it is advisable to use fairly large cylinders. The cylinders used in the following experiments were 27.4 cm. in diameter, and 10 cm. or more in height. For liquids, the cylindrical glass trough used has a diameter of 25 cm.
The tube of the radiator is fixed and points to the centre of the graduated circle. The vertical central line of the cylinder passes through the centre of the circular platform.
The Receiver.—The receiver is a modified form of the coherer. In a rectangular piece of ebonite a narrow groove is cut out. In this groove bits of coiled steel springs are arranged side by side, only one layer deep. In this way a linear receiver is constructed with a sensitive surface 2 cm. in length and 4 mm. in breadth. By means of a screw, the springs may be gradually compressed, reducing the resistance. The coherer is in a circuit with an aperiodic D'Arsonval galvanometer and a copper-iron cell. The galvanometer has a resistance of 300 ohms, and the voltaic cell has an E. M. F. of about 0.45 volt. A Daniell cell is sometimes used, with a resistance box as a shunt; the E. M. F. can thus be adjusted to suit the sensitiveness of the receiver. When the spiral spring coherer is freshly made, it is over-sensitive. On the second day it settles down to a fair condition, though for about half an hour its action is rather unsteady; but afterwards the sensitiveness becomes fairly uniform. It will maintain this state under favourable conditions for nearly an hour, after which it begins to lose its sensitiveness. It must also be borne in mind that the sparking balls also are undergoing deterioration. The sensitiveness of the coherer may be partially restored by slightly raising the E. M. F. of the circuit. In this way it is
