32 Dr. E. N. da C. Andrade on the Electrical 



§ '.). Experiments at Reduced I'ressures. 



This resuH renders doubtful the work of Gr. Ebert *, who 

 carried out experiments at low pressures (76 to 26 cm. Hg) 

 and found that in this range the velocity of migration is 

 independent of the pressure. But he records that the 

 potential gradient in the flame rises with decreasing pressure, 

 which is highly improbable, in view of the well-known in- 

 creased conductivity of gases at lower pressures t- It there- 

 fore seemed advisable to repeat the experiments at low 

 pressure. The flame and electrodes were those used in the 

 high-pressure experiment: they stood on a smooth plate 

 provided with the necessary inlets for gas and air, and were 

 covered by a large bell-jar, which could be rapidly exhausted 

 by means of a large oil pump, which was kept running con- 

 tinuously. The platinum wires used to measure the potential 

 difference in the flame were 0*2 mm. in diameter, and experi- 

 ments were done with them at two different distances from 

 one another (6*4 mm. and 5*1 mm.). If a luminous streak 

 of the metallic vapour be brought between them, not 

 touching either, there is no perceptible change in the differ- 

 ence of potential between them, provided the streak i- not 

 too strong. We can therefore measure the difference of 

 potential in the flame itself. 



It was first shown that the potential gradient in the flame 

 was proportional to the potential of the insulated electrode, 

 as was to be expected. This indicates that the wires were 

 giving reliable measurements of the gradient : E berths 

 measurements are very discordant on this point, and seem to 

 indicate experimental errors J. 



The pressure was then varied. With a eon-taut potential 

 of the insulated electrode the gradient in the flame decreased 

 rapidly with decreasing pressure. 



* G. Ebert, Heidelberg Diss., 1911. 



t There are also contradictions in the measurements recorded iu 

 Ebert's paper which point to the measurements of potential gradient 

 being unreliable. 



X It appears thai Elbert made measurements without varying the 

 distance apart of the probe wires, the fixed distant used by him being 

 very -mall (2 linn, or less). With bo small a distance, especially if the 

 wires are not very thin, there may he many disturbing influences: the 

 Same between tin 1 wires is omen cooled, and possibly some kind of a 

 mutual action takes place between the wires, which action would 

 probably extend to greater distances with decreasing pressure. 1 tried 



to measure with a very -mall distance apart (1*6 mm.) and could get DO 



satisfactory results. 



