193 
a metal j^late over the radium, and tiie difference taken as 
the proper value of the leak for that experiment, a small pro- 
portion only being due to fi rays. The first measurements 
relate to a chamber of depth 6 turns — (i.) under a potential 
gradient of 600 volts for the 6 turns : (ii.) a gradient of 20 
volts for the same distance : (iii.) under 600 volts again. The 
difference between (i-) and (iii.), as shown in the table, was 
due to the variation in sensitiveness of the electrometer. In, 
almost all our experiments this variation has been negligible : 
in this sj^ecial case it was not so, because so large an amount 
of radium was used, viz., 5 mmg. The y rays penetrated all 
the metal casings, and caused a leak in the charge of the 
needle. The leak had an exaggerated influence on the read- 
ings because the capacity of the electrometer was increased by 
the addition, in parallel, of a plate condenser of 1,000 cm. 
capacity. This disturbed the usual balance of the electro- 
meter, in which leakage of the needle's charge had little effect 
on the deflection for a given quantity of electricity. To 
obviate any error from variation of sensibility the results of 
(i.) and (iii.) were averaged, and compared with the result 
of (ii.). It will be observed that successive determinations of 
the same leak were very consistent with each other. This 
implies that almost all the observed effect was due to the 
radium : extraneous influences were very small. 
6 Turns. Zero = -1^700. 
600 volts 
(Metal over Ra) 
t 600 volts 
47 
47 
'i7 
54 
7o 69 U. 11 
no -n t -Mean leak 
63 o9 j 
67 
47 42 47 ( 
92 92 98 I 
-Nett leak = S."v2 = 1,, (say) 
Mean = 91 9 
20 volts 47 
(Metal over Ra) 47 
20 volts 
.13 
50 54 
^^4^ [Mean = 48 ^ 
49 39 49 I ,1 -,.3 I 
43 36 42^^^^«"='-*'^ J 
-Xett leak =695 = I' 
600 
(Meta 
volt.s 
overRa) 
47 
47 
70 
61 
69 
.'i5 
[ Mean = 64 
-Nett 
leak = 
= .S29 = 
= I„ 
600 
volts 
57 
54 
19 16 13 ( ^r 
73 70 66 ( ' ^^^" ~ 
S93 
.Mean value of I,. =840. 
Valne ot I',. =695. 
