480 Mr. A. B. Meservey on the Potentials required 



and 4 cm. in diameter, was closed at the ends with ebonite 

 disks, through which a concentric wire or tube o£ a given 

 material and a given diameter was passed. The apparatus 

 could be easily taken apart, and another cylinder substituted. 

 As a precaution against leakage and irregular discharge, 

 the inner cylinder was surrounded at the ends by short 

 pieces of glass tubing, leaving about 16 cm. clear for 

 discharge at the middle o£ the apparatus. The passage o£ a 

 current was detected by a P'Arsonval galvanometer of 

 92 ohms resistance, giving a deflexion of 1 mm. on a scale 

 1 m. distant for a current o£ 4*2 x 10~ 8 ampere. The 

 difference of potential between the cylinders was established 

 by a battery o£ small storage-cells, and was measured by a 

 Kelvin electrostatic voltmeter reading to 600 volts, con- 

 necting directly to the electrodes. The voltmeter had been 

 accurately standardized, and during the course of the ex- 

 periments was compared with a standard electromagnetic 

 instrument. The gas was permitted to stand for a clay or 

 more in a drying-tube before being admitted to the apparatus, 

 and there was another drying-tube in close proximity to the 

 spark chamber. 



When it was desired to take a set of readings, the con- 

 nexions Were made so that the difference of potential would 

 be less than the sparking potential ; then the high potential 

 wire was moved up, a cell at a time, till a galvanometer 

 deflexion indicated a discharge between the cylinders. The 

 pressure was determined by the McLeod gauge, the con- 

 nexions reversed, and another reading taken after a short 

 interval. A small quantity of gas was then pumped out or 

 admitted as the case might be, and, after time enough had 

 elapsed to permit the pressure to become steady, the whole 

 process was repeated at the new pressure. Readings were 

 usually started at comparatively high pressure, and the latter 

 was then decreased, a pair of readings being taken after 

 each stroke of the pump, till the lowest desired pressure was 

 reached. Small quantities of gas were then admitted, and 

 the sparking potentials found in each case. The steps were 

 more numerous during this process than during the other, 

 as smaller changes of pressure w T ere possible. 



The readings were often somewhat less regular when the 

 pressure was being changed by admitting fresh supplies of 

 gas than when it was changed by the use of the pump, but 

 the mean position of the curve through the points was not 

 materially altered. It was found that when the apparatus 

 had been unused for several hours, the potential required 

 for the first spark was distinctly higher than that required 



