UPON ELECTPJCAL DISCHAEGE PHENOMENA IN RAREFIED GASES. 
143 
field, it is reasonable to suppose that the incoming radial streams of positive i(jns 
would have their equivalent in outgoing (that is, from the centre of the hullj towards 
the equator) streams of negative ions. But such was not found to be the case, as this 
last experiment showed. Acting upon a suggestion kindly made by Lord Kelvin, 
this point was still further investigated by means of the following experiment:—^Two 
plane pieces of metal were fixed into the bulb as symmetrically as possible, one on 
each side of the 2-mm. air gap between the pointed ends of the magnetic electrodes. 
There was no glass covering over these electrodes, the bulb used being similar to that 
shown in fig. I. These metal planes were 1'5 centims. long, '5 centim. wide, and 
1 millim. thick, and were placed parallel to one another with a space of 2 millims. 
between them. The question to be tested was whether the previous diselectrihcatioii 
effects were in any wa,y due to want of symmetry. But with this apparatus all tlie 
results previously obtained were easily repeated. The two metallic planes were then 
replaced by a finely pointed wire, which projected into the bulb, and this when 
charged positively became diselectrified when the magnet was excited. 
On returning to the arrangement shown in fig. 6, it was found, as a general rule, 
that the higher the potential to which the body A was charged, the greater for a 
constant pressure became the strength of the magnetic field necessary to produce 
diselectrihcatioii—more especially was this noticeable at very low pressures. The 
relationship is shown by the continuous curve in hg. 7, where the abscissas represent 
the degree to which A was electrihed, and the ordinates are proportional to the strength 
of the magnetic held sufhcient to produce diselectrihcatioii. On the other hand, 
when A in each exjieriment was charged to the same potential the magnetic strength 
siihicient to produce diselectrihcatioii varied with the pressure of the residual gas 
in the manner indicated by the dotted line in the same hgure. 
The results graphically shown by these curves were obtained in tlie following 
manner:—A was connected with the quadrants of a Kelvin electrostatic voltmeter, 
while a hne wire from T^ was securely screwed to the case of the instrument. The 
current supplied to the electro-magnets passed through a variable resistance, and also 
an ammeter which could be read to two places of decimals. It was generally found 
to be an advantage to introduce a small condenser between T\ and Tb in order to 
steady the needle of the voltmeter. The current through the electro-magnets was 
always increased very slowly, and the exact moment at which diselectrihcatioii 
occurred could easily be determined by the sound of the sharp snap which was heard 
when A suddenly lost its charge. 
The regularity of the readings of the ammeter for a constant gas pressure within 
the bulb may be seen from the following examples, where P represents the potential 
in volts to which A was charged, and C the strength of the current in amperes 
through the electro-magnets, just sufilcient to produce diselectrihcatioii (pressure was 
‘006 millim.). 
Each reading was repeated four times. 
