hu Hot riathium in Air at Low Pressures. 81 



])ol:ential, whilst the surroiindino- cyhndrical ek'ctrode was 

 connected to one of the quadrants of the electrometer, the 

 other being earthed. In general the experiments were made 

 at such a temperature that the currents could conveniently 

 be measured without iixlding any external capacity to that of 

 the electrometer. The temperature was always below that at 

 which the negative leak became detectable. 



The first experiments were made to determine the w^ay in 

 which the current from the wire varied with the applied 

 potential, the other conditions being maintained constant. 

 The temperature was 442° C. The pressure remained con- 

 stant, and equal to '00625 mm. of mercury. Observations 

 were taken first with no potential on the wire, and then for 

 every additional 40 A-olts up to 400 ; the values of the 

 current"^, in scale-divisions per minute, obtained in this 

 manner are marked thus, x, on the accompanying diagram 

 (PI. V. fig. 1). The observations were then repeated, the 

 potential being continuously decreased by 40 volts down to 

 zero; the values obtained in this w^ay with decreasing potential- 

 differences are marked thus, o. 



It will be seen that the ascending observations give rise to 

 a curve which is slightly convex to the axis of current; on the 

 other hand, the curve which represents the descending observa- 

 tions possesses about an equal curvature, but is concave to the 

 current-axis. The disj^arity of the tw^o series of observations 

 is due to the falling-off of the current from the w^ire with 

 time, a point which is examined in detail later. It wdll be 

 seen that the time-effect can be allow^ed for by taking the 

 mean of the tw^o observations belonging to the same potential^ 

 since this w^ould be — approximately at any rate — the current 

 for that potential at a time half-way between the two experi- 

 ments. This time is the same for every two observations, 

 and is identical with that at w^hich the reading for 400 volts 

 was taken. This reasoning assumes that every observation 

 takes the same length of time, a condition which was 

 approximately fulfilled. 



It will be seen that the current-E.M.F. curve is almost a 

 straight line passing through the origin, so that the current 

 is nearly proportional to the electromotive force. A repetition 

 of the observations gave the same approximately linear 

 relation, and showed that the fact that the mean point for 

 40 volts fell off the curve was due to accident, for it did not 

 occur again. 



Some experiments on the relation betw^een the current and 



* Throughout this paper 1 scale-divisiou is equal to '02 electrostatic 

 units of quantity. 



Fhil. Mag. S. 6. Vol. 6. No. lU. Jal2/ 1903. G 



