234 



NATURE 



[July 8, 1897 



Bttnsen Flame. 

 Sensitiveness of electrometer = 607 scale divisions per volt. 



§ 3. If the leakage in these experiments were proportional 

 to the difference of potential, then the product of mean differ- 

 ence of potential into time should be constant for the same 

 deflection from metallic zero. Taking the numbers obtained 

 for the 300 scale divisions of deflection in virtue of the Bunsen 

 flame, we have : — 



Positive charge Negative charge 



Thus it is proved that the leakage between two plates, each 

 10 square centimetres in area, i centimetre apart when the 

 fumes from a Bunsen burner pass between them and round 

 them, is approximately proportional to the difference of potential 

 between them, when that difference is above 20 volts and 

 up to 94 volts, the highest we have tried ; but that, below 

 20, it diminishes with diminishing voltages more than according 

 to simple proportion. 



§ 4. To determine the currents which we had in our arrange- 

 ment, we took a movable plate of a small air condenser charged 

 to a known potential, and applied it to the insulated terminal 

 of the quadrant electrometer. In this way we found that a 

 quantity equal to 0*15 electrostatic unit, gave a deflection of 300 

 scale divisions. Hence in the experiments with the Bunsen 

 flame and with a potential of -f 94 volts kept on the uninsulated 

 copper plate, the current to the insulated copper plate opposite 

 to it, when 300 scale divisions was reached in 5 seconds, was — 



-^ X i = io~^^ ampere. 



3 X io9 ° 



= mikro-ampere. 



lOOOOO 



§ 5. One of us about the year 1865, when occupied in ex- 

 periinenting with the latest form of portable electrometer, found 

 that if it was held with the top of its insulated wire (which was 

 about 33 centimetres long) a few inches below a gas-burner, a 

 charge of electricity, whether positive or negative, given to this 

 wire was very rapidly lost. The disinsulating power of flames 

 and of hot fumes from flames was well known at that time, but 

 it was surprising to find that cold air flowing up towards the 

 flame did somehow acquire the property of carrying away 

 electricity from a piece of electrified metal immersed in the cold 

 air.i Circumstances prevented further observations on this very 



1 We have recently (June 1897) found the following statement, in Worth- 

 ington's communication to the British Association (1889 Report, pp. 225, 

 227) '• On the Discharge of Electrification by Flames"; . . . "the observa- 

 tion seems to have been made by Priestley, that the discharge takes place 

 •with apparently equal rapidity, if the rod be held at the side of, or even 

 below, the flame at the distance of, say, five centimetres " The four words 

 which we have italicised are clearly erroneous, as we find enormously greater 

 leakage five centimetres above a flame than five centimetres below it : but it 

 is very interesting to learn that Priestley had found any leakage at all 

 through air five centimetres below a flame. 



NO.. 1445, VOL. 56] 



interesting result at that time, but the experimetjt was repeated 

 with a portable electrometer in December of 1896, and we were 

 made quite sure of the result by searching tests. During April 

 and May of the present year observations were again made by 

 meansof(i) a multicellular electrometer reading up to 240 volts, 

 and (2) a vertical electrostatic voltmeter (Fig. 3, p. 235) reading 

 up to 12,000 volts. A steel wire 43 centimetres long was fixed 

 to the insulated terminal of the multicellular electrometer, with 

 its needle-point vertically below an ordinary gas-burner, as shown 

 in Fig. 2. 



§ 6. By means of a small carrier metal plate (a Coulomb's 

 proof plane) a positive or negative charge was given to this wire 

 and the quadrants of the multicellular till the reading on the scale 

 was 240 volts. The leakage was then observed {a) with gas not 

 lit, {b) with gas lit at different vertical distances above the point 

 of the wire. We found that there was rapid leakage when the 

 flame was one centimetre above the wire ; and the times ot 

 leakage from 240 volts to about 100 volts increased as the flame 

 was raised to greater distances above the point ; or, otherwise, the 



TWl 



r \ 



Fig. 



rate of fall of potential in one minute from 240 volts diminished 

 as the distance of the flame above the point was increased. 

 When the vertical distance of the flame above the point was 15 

 centimetres, or more, the time of leakage from 240 volts was 

 practically the same as if the flame was not lit at all. A 

 plate of metal, glass, parafiin, or mica, put between the point 

 and the flame, diminished the rate of leakage. The leakage 

 from 200 volts during the first minute is given in the following 

 table, for different distances of the flame, with no intervening 

 plate. 



* We sometimes found the multicellular electrometer to insulate so well 

 that in five minutes there was no readable leakage from 240 volts. 



