mined, and the constancy of the potential of the supporting rod at the same time tested.
The number of scale divisions traversed during the course of a determination of the leak did not in any case exceed ten ; in the case of the smallest rates of leak measured only one scale division was traversed. As one scale division corresponded to 0'31 volt, the poten- tial of the leaking system never fell more than about 3 volts below that of the supporting rod.
The potential of the supporting rod did not itself remain absolutely constant, but fell generally from 1 to 2 volts in the hour, unless a small additional charge was given to the condenser during the course of the experiment ; this was done in the case of the smaller leaks. The correction to be applied for a fall of potential of the support corre- sponding to a given number of scale divisions was determined by first making contact by means of the magnet, then reducing the potential of the supporting rod sufficiently to move the gold leaf across a whole number of scale divisions, and again making contact and observing the new potential of the supporting rod. It was found that to make a change of potential in the leaking system corresponding to one scale division (the charge remaining constant) a change of potential of the supporting rod corresponding to 5 '4 divisions was required ; or the cor- rection to be applied for a fall of potential of the supporting rod corre- sponding to one scale division amounted to - 18 of a division.
The error due to want of perfect insulating power in the sulphur bead was certainly very small. Even when the supporting rod was earthed, so that the difference of potential between the two conducting systems separated by the sulphur was about 90 volts, the movement of the gold leaf, when the apparatus was exhausted, did not exceed one division per hour. Under the conditions of the actual experiment the difference of potential between the two systems never exceeded 3 volts, and with the apparatus exhausted no leak was observed.
The apparatus for the preparation and drying of the gases was sealed on to the bulb in which the leakage was measured, all joints being made with the blow-pipe. A mercury pump served to exhaust the apparatus before the introduction of the gases. The gases were dried by passing them slowly over phosphoric anhydride.
The air was admitted through a cotton -wool filter. The carbonic acid was prepared by heating sodium bicarbonate. The hydrogen used in the first experiments was prepared from zinc and sulphuric acid ; in the later experiments the gas thus obtained was absorbed by palla- dium, from which it was set free by heating. The sulphur dioxide was obtained from a " siphon " of the liquid substance. The chloro- form vapour was obtained from the commercial substance. The temperature ranged from 20 to 24 C. The results are given in the tables which follow.
