Wave-trains through Layers of Electrolyte. 535 



The first trials of the apparatus were made with dilute 

 solutions of copper sulphate, some five or six centimetres of 

 solution being the greatest depth used, as the dish was only a 

 shallow one. One of the first curves obtained is given in 

 fig. 2, and may serve as a type of two or three similar ones. 

 The curve is by no means logarithmic, and if the conductivity 

 of the solution be calculated from a pair of points on it arbi- 

 trarily chosen near the beginning and end it comes out very 



It appeared possible that this irregularity might be due to 

 the existence of interference-effects analogous to Newton's 

 rings or the phenomena of thin plates by transmitted light, 

 especially in view of a somewhat similar effect observed just 

 previously by Mr. Barton in the same laboratory*; but I 

 felt doubtful at first whether it might not be ascribed to some 

 other disturbance — to the fact that a part of the radiation 

 passes round outside the jar instead of through the electrolyte, 

 or to mutual induction between the two halves of the long 

 wire circuit in the garden. The second disturbance proved, 

 however, to be nonexistent, and the first quite inadequate to 

 explain such a distortion of the curve. 



To see if the effect were really due to interference phe- 

 nomena, it was necessary to determine a transmission-curve 

 for a great length of liquid, and consequently to avoid ab- 

 sorption as far as possible. Distilled water presented itself 

 naturally as the best liquid for the purpose. 



A tall glass cylinder, about 114 centim. high and 12 centim. 

 internal diameter, was used as the containing vessel, the wires 

 being strained straight down the middle of it. and passing 

 through holes drilled in the bottom. 



In order to avoid having to remove all the water between 

 every pair of readings, the method of observation was slightly 

 altered : each point was compared with one previously deter- 

 mined, instead of directly with the zero. As before, ten or 

 twelve readings were taken at every point to cover variations 

 in the effectiveness of the primary sparks. As a specimen of 

 the usual variations, I give two series of observations in 

 Table I. The first was for the determination of the ratio 

 between the intensities transmitted at 40 and at 55 centim., 

 the second for the determination of the same ratio at 60 and 

 80 centim. 



* Proc. "Roy. Soc. liv. p. 85 (1893). 



