04: 
NATURE 
[APRIL 22, 1897 
charcoal was kept away from the funnel till all flame had dis- 
appeared. Then the vessel was put in position, and the 
deflections obtained in two experiments were— 
51 scale divisions positive in 10 minutes. 
100 
” ” ” 2»? ” 
§7. Next an experiment was made with the burning charcoal 
put in position while a flame was visible. The flame remained 
visible for 7 minutes, and in that time a negative electrification 
of 34 divisions was obtained. Then the deflection came back 
to the metallic zero in one minute, and in 10 minutes more a 
positive electrification of 87 divisions (0°78 
volts) was obtained. 
§8. Glowing coals taken from the 
fire and put at once in the vessel in 
position, repeatedly gave negative elec- 
trification ; but when they were kept 
away from the funnel till all flame 
had disappeared, the electrification ob- 
tained was slightly positive. Glowing 
coals remained glowing a very short 
time after all flame ceased, and the 
smallness of the observed effect is prob- 
ably due to this cause. 
§9. A few experiments have also been 
tried to find to what positive potential 
the flame must be raised so as to over- 
come the negative electrification it gives 
to the air. Hitherto the only flame 
tried was a spirit flame. The positive 
electrode of a secondary cell was put 
into the flame of the lamp, and the 
negative electrode was joined to the iron 
funnel and to the case of the electrometer. 
The results obtained are not very regular, 
but we found that one storage cell was not sufficient to over- 
power the electrifying effects of the spirit fame. With one cell 
we got 45 divisions negative in 10 minutes, instead of 109 
divisions with metallic connection; with two cells we got 10 
divisions positive in 10 minutes ; and with six cells we got 83 
divisions positive in 4 minutes. 
_S 10. The filter, pump, and worm were now removed, and 
two plates—one of polished copper, and the other of polished 
zine—were fixed 0’9 centimetre apart in a block of paraffin, as 
represented in Fig. 2. The arrangement was such that either 
plate could be insulated, while the other was kept in metallic 
connection with the case of the electrometer. Observations 
were made to find the deflection from metallic zero with one 
plate insulated, and fumes from different flames and burnings 
at different distances from the plates passing up between them. 
This may be called the fumes-zero. When the top of the flame 
was within 5 or 6 centimetres from the plates, the results 
were very irregular. The results in the following table for 
spirit flame are in accordance with what Maclean and Goto 
obtained from unguarded fumes from a spirit lamp 30 centi- 
metres below the plates, as stated in their paper published in 
the Philosophical Magazine for August 1890. The effect is of 
the same kind as if the plates were connected by a drop of 
water. 
1 Kelvin, ‘ Electrostatics and Magnetism,” $$ 413 414, PP. 332) 333. 
NO. 1434, VOL. 55] 
Sensitiveness of the electrometer 136 scale divisions per volt. 
| | | He 4 
Distance of} Metal con- | DCE 
top of flame/nected to in-| fi 3 ra Drealan 
Flame. below the |sulated termi-| *U™&S2Er° oa 
; and metallic volts. 
plates in cen-|nal of electro-| * coal 
timetres. meter. eh AEE = 
ivisions. 
Spirit lamp 23 Copper S81 pos. 0'60 
» ¥ Zinc | Ior neg. 074 
AD II Copper 53 pos. 0°39 
» = = Zinc 76 neg. 0°56 
Paraffin lamp | | 
without glass | 
AUNME! fe e.2 | ses 7 Copper | 141 pos. 104 
» ” Zinc 135 neg. IOI 
3 15 Copper 90 pos. 0°66 
5 oa 35 Zinc 103 neg. 0°76 
, 3 neg 
a 23 Copper | 108 pos. 0-79 
+ ae On Zinc 112 neg. o0'$2 
” 30 Copper 83 pos. | 0°61 
=i) ; =) Zinc 83 neg. | 0°61 
§ 11. To observe the leakage between two parallel metal 
plates, the zinc plate was removed, and a polished copper plate, ° 
equal and similar to the other copper plate, was substituted for 
it. The distance between their parallel planes was 0°9 cm. 
The experiments were conducted as 
follows: One pair of quadrants of 
the electrometer, with one of the 
copper plates in metallic connection 
with it, was insulated. There was 
now no deviation from metallic zero. 
A small charge, positive or negative, 
was given to it, producing a deflec- 
tion of about 450 scale divisions. 
This corresponds to over 9 volts, as 
the sensitiveness of the electrometer 
now used was 48'2 scale divisions 
per volt. In two or three minutes 
the ordinary leakage of the arrange- 
Fie 3 
ment was observed. This did not amount 
to more than one division, or at most two 
divisions, per minute. Then the flame was 
lit, and readings were taken every half- 
minute. This was done with the variations 
in the funnel described in the last column 
of the following table, and illustrated by 
Fig. For comparison, the numbers in 
the following table show the leakage for two 
minutes after the reading was 300 scale 
divisions (6°2 volts) from metallic zero. This 
gives us the leakage at diminishing electric 
potentials during the time of observation. 
We intend to continue these experiments, 
and to arrange to find the leakage at different constant 
electric pressures. ‘ 
§ 12. The marked difference in the leakage obtained when the 
horizontal tube was of small bore (3°8 ems.) and when it was of 
larger bore (15°3 cms.), may be contrasted as indicated in the 
last four results given for spirit flame. We also tried how long 
the fumes retained this conductive quality, but in every case we 
found that the leakage stopped in less than a quarter of a minute 
after the flame was extinguished, or removed from the bottom 
of the funnel. Closing the top and bottom of the funnel imme- 
( diately after the flame was removed, we still found that the 
