473 Messrs. Haseltoot and Kirkby: Electrical fects 
the same conditions of pressure. This hope was not realized, 
as will be seen from the following table of results :— 
Taeve LV’. 
! ; f Heddleities eA 
p | p" T We Q q 
80 60 20) 3 83.) 2oic th Saas 
84 63 21 88 | 55 | 0087 | 
88 68 20 83 "805 ‘0045 | 
The last observation is the mean of four experiments in 
each of which the field of force was identical and the amount 
of the mixed gases varied by only 15 mm. The deflexions 
obtained were all small, giving values of g ranging from 
(0034 to (0064. Except in the first experiment of the Table, 
the amount of air present was less than 1 mm. It appeared, 
too, that a smaller percentage of the gas was exploded than 
with the previous apparatus. But some uncertainty arises 
from the difficulty of estimating the amount of water-vapour 
formed, since it diffused slowly through the pressure-tubing 
which led to the pentoxide of phosphorus. 
With regard to the experiments generally, there are two 
quantities which it is sought to measure, the quantity of 
gas exploded and the amount of electricity formed. We con- 
sider that the experimental errors may be estimated at 5 per 
cent. in each case. The chief difficulty in measuring the 
former quantity was to make the proper allowance for the 
water-vapour formed by the explosion which is not absorbed 
immediately by the phosphorus pentoxide. It was observed 
that the pressure p diminished by as much as 3 mm. when 
the gases were lett undisturbed for a considerable time. A 
neglect of this would involve an error of about 8 per cent. 
in the fall of pressure as calculated above. Accordingly, 
suitable corrections were applied to the pressures observed 
immediately after opening the pinchcock EH, the effect of 
which would be that errors exceeding 4 per cent. could 
scarcely occur. This estimate, however, does not apply to 
Table 1V. 
In measuring the charge of electricity there was the possi- 
bility of a leak “of electricity across the ebonite plug through 
which the wire passed, and of a misreading of the instrument. 
The former was obviated in the case of the electrometer- 
observations in the manner already described—in the case of 
