( 669 ) 
Each series of measurements consists in turn of two branches. 
First by tentative approximation from both sides for all values of 
the field those values of the azimuth of the electromagnet are found 
for which the couple is a maximum. In this way two azimuths are 
found which are symmetrical with respect to the major axis of the 
ellipsoid and which exert couples of opposite sign. This determination 
can be made accurately to within 0,5° to 1°, which is quite sufficient. 
Then follows the true measurement in which the magnet without 
current is placed in one of. these positions, the circuit is closed and 
immediately afterwards the deflection is read. As soon as this is 
done, the circuit is broken, the magnet is placed in the symmetrical 
position; once more the current is allowed to flow and the new 
deflection is obtained. Since these operations occupy only a short 
time, the after effects in the spring are of no account. The difference 
between the scale readings gives twice the value of the couple to 
be measured, independent of the residual magnetism remaining after 
the current was broken, which however occasioned only an extremely 
small couple. The field was given as a function of the current indi¬ 
cated by the ammeter. For these observations the same ammeter 
(Siemens and Halske instrument, no temperature coefficient) was used 
which was employed in the study of the field. This method of 
evaluating the field was quite sufficient for our purpose. The distance 
between the pole-pieces was read off the divided cylinders of the 
magnet and was verified by passing between them callipers which 
had been previously adjusted to the desired distance. The fields given 
above are corrected for the demagnetizing fields of the ellipsoids. 
§ 4. Corrections and controls ; auxiliai'y measurements. The inherent 
magnetism of the holder is not so weak that the corrections neces¬ 
sary for it may be neglected. On that account a series of measurements 
was made with no ellipsoid in the holder at ordinary and liquid 
hydrogen temperatures. With the weaker spring we found: 
TABLE I. 
Correction for the magnetism of the holder. 
ordinary temperature 
t = 20°.2 K. 
4000 gauss 0.18 cm. 
8000 0.29 
0.26 cm. 
0.48 
0.61 
12000 
16000 
20000 
24000 
0.36 
0.43 
0.73 
0.50 
0.86 
0.98 
0.57 
