PROFESSOR R, THRELFALL AND MR. J. H. D. BREARLET 
116 
due to residual surface action, though later experiments lead us to think that this 
is not the case. The capacity test was again applied, giving the same result as 
before, and showing that the plates are still in position. It was decided to heat the 
sulphur condenser, and observe change of resistance and capacity as the temperature 
rose. The galvanometer was now more sensitive than before, giving one micrometer 
division of double elongation for T31 X 10~ 12 ampere. With 291 volts no effect 
could be detected. The result is to place the specific resistance above 1*4 X 10 27 
C.G.S. The galvanometer now approaches the electrometer in its power of discrimi¬ 
nating small conductivities. The temperature was 17°'2 C. 
The temperature of the bath was then raised to 50° C., at a rate of about *2° per 
minute, and was kept within a degree on either side for about an hour while the 
resistance and capacity tests were made. The capacity appeared to be exactly the 
same as before— i.e., it did not change by as much as ‘3 per cent., or we should have 
detected its variation. It will be shown later that this probably indicates an appre¬ 
ciable, but small, positive temperature coefficient of the specific inductive capacity. 
The resistance test also failed to show any signs of conduction, and the result may be 
considered to be the same as at 17°*2 C. 
The temperature was then raised to 75° C., and kept there for several hours. A 
slight conductivity showed itself, corresponding to a deflexion of 21'3 divisions, with 
a voltage of about 285 volts. The specific resistance is now 6'8 X 10 25 C.G.S. After 
the battery had been charging the sulphur for about an hour and a half at the 
temperature of 75° the deflexion fell from 21 '3 to about 5 divisions, showing that 
the resistance had increased, say, four times by the continued application of electric 
stress. Tills being the first clear case of conductivity we had encountered, we were 
anxious to note any phenomenon which might present itself. The most striking 
effect was the discontinuity of the conduction. At first we thought that the sudden 
sharp kicks given every now and then by the galvanometer were due to accidental 
causes of some kind, but it soon became clear that the phenomenon was a real one. 
When the voltage is first applied the discontinuity is more marked than when it has 
been on for some time. In the case we have before us the galvanometer would take 
up a definite position of deflexion, and every two or three minutes this lvould suddenly 
increase or diminish by a large amount. The only possible way of estimating the 
conductivity is, therefore, to take a long series of observations and obtain a mean 
value, and that is what was done above. The impression forced on one while 
observing at the galvanometer is that a succession of irregular changes are taking 
place; for instance, that groups of molecules, one after the other, succumb to the 
electric stress, and signalise their destruction by allowing a current to pass. The 
conductivity in these cases, however, never falls to zero. It is, perhaps, not too 
much to say that all the observations are in accordance with a theory that there is 
a small regular and continuous conduction, superimposed on which there is something- 
very like a disruptive conduction, We have since discovered that the discontinuous 
