238 Mr. G. Owen on the Discharge of 
observations is given in the table. The second column gives 
the resistances inserted in series with the filament in addition 
to the 20-ohm iron wire resistance supplied with the lamp. 
The results obtained for R=30 ohms, and for R=0O on 
March 1st, have been plotted and are shown in diagrams 1 and 
2 respectively. 
Table II. has been drawn up from Table I. and diagrams 
1 and 2 to show how the ratio of the positive and negative 
leaks changes with increasing temperature. 
TaBe II. 
| Potential on Filament= +600 Volts. 
/ | Ratio of Negative to 
| Positive Leak. 
Date. R ohms. 
SI 7s eais chaw ne cc gueaineee 100 | 53 
fon Sais atvice ios wiv pieces ced 60 286 
Nl ne ee hee eS, 30 449 
sod akselkanieemecs Crousieee Filament fully glowing. | 833 
SO) ssdist tech an ceamace do. : 1250 
111) Oc) oil Rea eas ee do. 2000 
The third column shows that the negative leak increases 
with the temperature much more rapidly than the positive 
leak. The last three numbers in the same column point also 
to a rapid decay of the positive leak with continued heating. 
The latter point will be more fully discussed in section 6, 
It will be noticed in Table I. that a negative leak is obtained 
with a positive potential of 40 volts on the filament. With 
the filament fully glowing at atmospheric pressure, it was 
found that +68 volts were just sufficient to stop the negative 
leak. <A potential of +70 volts gave a positive leak. The 
reason for this is that there was an alternating potential- 
difference of about 70 volts between the ends of the filament, 
together with the fact that the negative leak was much greater 
than the positive leak. 
(2) Variation of the Negative Leak with the Potential at 
Different Pressures ; Lonisation by Collision. 
On March 9th a series of observations were made to deter- 
mine the variation of the negative leak with the potential at 
pressures ranging from atmospheric pressure to a high 
vacuum. The temperature was maintained constant through- 
out the series by means of a thermocouple attached to the 
filament. 
The results obtained are given in Table III. The numbers 
within the brackets in the first column refer to the order in 
which the observations were made. 
