191 
Eelafioii of riiD-riif i<> pafenfuil (/rdfUciif for different widths 
of the io)}i><(ifi(ni rhatnhcr, the riirreiits heinrj small: 
potential gradients in volts per centnnetre : currents in 
arbitrary units, rednced to common viarimum. 
A\'ij>rn OK Chami{ki{. 
^"5 inin. () imii. 
roteiitial ,. , Potential r. „„. 
Gradient. Cradieiit. 
1) mm 
1,000 
34 
11 
8 
6 
4 
2 
1 
1 
1 
8 
9 
65 
6 
66 
75 
92 
50 
06 
59 
385 
20 
400 
341 
308 
294 
285 
274 
252 
235 
221 
196 
147 
106 
47 
1,000 
35-7 
12 
9 
7 
5 
2 
1 
•47 
•47 
•50 
•75 
•89 
•98 
•51 
•02 
400 
335 
302 
292 
283 
272 
247 
224 
169 
108 
4 
Potential 
Gradient. 
1,000 
36^0 
16-4 
9^83 
7-81 
5-83 
3-10 
2^23 
1-36 
•84 
•04 
Current. 
400 
333 
300 
290 
283 
271 
244 
220 
174 
134 
68 
These figures are plotted, as far as 36 volts per cm., in 
curves A, B, and C of Fig. 1- An open scale is chosen so as 
to show the separation of the curves at low values of the field, 
when n is not very small. 
These figures and curves shovv- that the ratio of the cur- 
rent at any particular strength of field to the saturation cur- 
rent is almost independent of the shape of the ionisation 
chamber, when the current is small. As this seems an im- 
portant point, we have made many experimental tests of it. 
We give below the details of one such test, in order to illustrate 
the methods employed and their degree of accuracy. 
The arrangements were the same as those just described, 
and the special object of the experiment was tne determina- 
tion of the degree of saturation under a certain moderate 
potential gradient in the case of chambers of two different 
depths. The depths were reckoned in turns of the screw, which 
raised the upper from the lower plate of the chamber : eleven 
turns =1 cm. The currents were allowed to run into the 
electrometer for 10 seconds. The electrometer was not dead 
beat, and therefore the first and second resting-places on the 
scale were observed, and the mean taken- For example, the 
second line, marked i, of the subjoined table, shows that the 
first deflection was to 57^47 cm., and then back to 54^92, zero 
being 4,700. The experiment repeated gave 57*42 to 54^92, 
and asfain 57*47 to 54^93. The leak was also measured witii 
