difficulty is further accentuated by the fact that the intensity of the electric radiation cannot be maintained absolutely constant. Attempts have been made in the following experiments to remove, to a certain extent, some of these difficulties.
Two semi-cylinders of glass, with a radius of 12·5 cm., were placed on the spectrometer circle. The plane faces were separated by a parallel air-space. The radiator was placed at the principal focus of one of the semi-cylinders; the rays emerged into the air-space as a parallel beam, and were focussed by the second semi-cylinder on the receiver placed opposite the radiator. Electric radiation was produced by oscillatory discharge between two small circular plates having a diameter of 1·2 cm. and an interposed platinum ball 0·97 cm. in diameter.
The two semi-cylinders were separated by an air-space 2 cm. in thickness; this thickness was found to be more than sufficient for total reflection. The critical angle for glass I found to be 29°. The experiments were commenced with an angle of incidence of 30° (slightly greater than the critical angle). The receiver, which was placed opposite the radiator, remained unaffected as long as the rays were totally reflected. But on gradually diminishing the thickness of air-space by bringing the second semi-cylinder nearer the first (always maintaining the plane surfaces of the semi-cylinders parallel), a critical thickness was reached when a small portion of the radiation began to be transmitted, the air-space just failing to produce total reflection. The beginning of transmission could easily be detected and the critical thickness of air determined with fair accuracy. When the thickness of air was reduced to 14 mm., the receiver
