of Electricity through Gases. 323 



were placed in this bulb in order that the hydrogen in it 

 might remain damp. After the bulbs had been filled the 

 communication between them was shut off. and the system 

 left for some time, so that the hydrogen in A might get 

 thoroughly dry. The time required to dry the gas, in order 

 to produce the effects described below, varied from two days 

 to a week ; and I am disposed to think that the drying effect 

 of the pentoxide was not exhausted, even after the gas had 

 been in contact with it for these periods. 



The following is the arrangement I adopted for measuring 

 the potential-difference required to produce a spark. The 

 difference of potential was produced by a battery of 600 small 

 storage-cells giving a potential-difference of about 1200 volts; 

 the ends of this battery are connected with C and D (fig. 2), 

 which are two terminals sunk into a slate bed; these terminals 

 are connected by a very large resistance formed by drawing on 

 the slate with graphite a semicircle passing through CD 

 with its centre at E; a metal arm turning about E makes 

 movable contact with the graphite semicircle, the contact 

 with the graphite being made by a metal roller. E and D 

 are connected respectively with the terminals of one of Lord 

 Kelvin's vertical electrostatic voltmeters ; they are also con- 

 nected with the spark-gap in one of the hydrogen bulbs, the 

 terminals of the spark-gap are portions of large brass spheres 

 carefully cleaned. By moving the sliding arm EF from D 

 towards C, the difference of potential at the spark-gap was 

 gradually increased until a spark passed, the electrometer 

 indicating the corresponding difference of potential. 



Observations were made both on the dry and the damp 

 hydrogen. As the result of these observations it appears 

 that there are well-marked differences in the appearance and 

 behaviour of the wet and the dry hydrogen. In the first 

 place the colour of the discharge is different, that of the 

 damp hydrogen being much more purplish than that of the 

 dry. The most striking difference however, which I observed 

 was the remarkable disproportion between the potential- 

 difference required to produce the first spark through the 

 dried hydrogen and that which is sufficient to cause one to 

 follow it immediately afterwards. I found that whether the 

 gas was damp or dry the potential-difference required to 

 produce the first spark was always greater than that re- 

 quired to make another follow it after a small interval. In 

 the damp gas, however, the difference was comparatively small, 

 averaging about ten per cent.; i. e. after the arm EF had been 

 moved from D until a spark passed through the damp hydro- 

 lien, it could not be moved back by more than about j ] of 



