i 823 ) 
measurements whose results had to be used (axes and volume of the 
ellipsoid, constants of the springs, magnetic field, density of the liquid 
oxygen) and also that the shape of the ellipsoid was not perfect. On 
the other hand we must remark that the application of the correction 
for the non-urfiformity of the field might conceivably have diminished 
the correspondence between the results obtained by the twoimethods. 
We have, however, both theoretical and experimental grounds for 
the assumption that this correction remains within the limits of 
accuracy of not more than 0.5% in the case of cylindrical pole-pieces 
with flat end surfaces 90 mm. in diameter and at a distance of 
21 mm. apart. 
TABLE Va. 
Conical pole-pieces 20 mm. apart 
/=- 183Q.0 C. , 
(To determine /? we assumed XgQ 0 ^ = 240.6.10 ). 
H 
gauss. 
Double de¬ 
flection 2d' 
cm. of the 
scale. 
- 
3685 
4.27 
0.935 
58.5 
46i5 
4.96 
0.920 
54.7 
6944 
4.55 
0.9437 
60.8 
9205 
_ 7.96 
0.9400 
59.7 
41280 
11.90 
0.9348 
58.5 
12835 
45.44 
0.9374 
59.1 
14015 
18.26 . 
0.9295 
57.0 
44900 
20.19 
0.9098 
51.9 
| 15585 
24.73 
0.8945 
48.1 
| 16120 
22.87 
0.8802 
44.5 
A graph of £ as function of E was made, which was used for 
the following table. 
Proceedings Royal Acad. Amsterdam. VoL XII. 
55 
