570 
Proceedings of the Royal Society 
Having decided upon our variables, we drew out the logical plan 
exhibited in the Tables, and proceeded to fill in the observations 
systematically. We first observed the behaviour of the flame 
under the influence of the charge, and also the mode of passing of 
the spark, and then proceeded to note the readings of the electrometer 
for three separate discharges. It will be observed, that where the 
flame was in a steady state the three readings differ only by small 
quantities, but that where the flame was in what may be called a 
critical condition, the differences are relatively larger. The sign of 
the electricity was changed by interchanging the connections at 
A and B, and that was always accompanied by a simultaneous 
interchange of the connections at H and K, in order that any bias 
due to the electrometer might be eliminated. The mean differences 
of potential obtained were reduced to absolute measure by taking 
the mean of the differences of potential required to produce several 
sparks between two parallel plates at a distance of 6 mm. apart. 
For such a spark the absolute value of the difference of potential 
is 46’52 C.G.S. units.* 
The figures sketched (Plate IV. a) indicate the behaviour of the 
flame, and the path of the spark in 24 representative cases. They 
are drawn in proportion to the exact dimensions. 
The results for a flame of constant height (Table I.) are repre- 
sented on diagram 1. The first conclusion that may be drawn is, 
that the difference of potential is greater when the disc is nega- 
tively electrified, excepting that for small distances there may be 
an equality or even a small difference the other way. The next 
conclusion is, that it is greater for the non-luminous flame at the 
smaller distances, but very much the opposite at the greater dis- 
tances, particularly in the case of the negative luminous. If we look 
at the notes or at the sketches, we shall find decided differences of 
behaviour coexisting with these differences in electromotive force. 
In the case of the negative electrification, the flame in general 
assumes a pointed form, becomes diminished, and loses its luminosity 
partially or wholly, or if already clear, becomes more bluish. But 
at the smaller distances, the flame being able to form a bridge over 
to the disc, remains with its tip in contact, and there are none of 
the above changes. On the other hand, in the case of the positive 
* Trans. Iioy. Soc. Edin., vol. xxviii. p. 652. 
