( 817 ) 
The mounting of the apparatus took place with the same pre¬ 
cautions regarding the centring of the whole, the tension of the 
springs, etc. and by a method similar to that which has been described 
in the research upon the ferro-magnetie metals. 
The course of the observations is very simple once everything has 
been set up in position. First, those azimuths of the electromagnet 
are tentatively determined for which the couple in both directions is 
a maximum. It was sufficient to do these experiments two or three 
times with suitably chosen fields, since the azimuth changes but very 
little with the field, and for other values of the field one can without 
danger have recourse to interpolation. After that the series of obser¬ 
vations took place in the following manner: Before making a measure¬ 
ment with any particular current this was reversed a certain number 
of times so as to obtain a well-defined field; we had not here to 
deal with a value of the saturation-magnetization, which changes but 
slowly with the field, but in our case the couple was proportional 
to the square of the field, so that inaccurate values of the field that 
might be obtained notwithstanding the fact that the iron of the 
electromagnet was extremely soft would make their influence very 
strongly felt in our results. Then the electromagnet was adjusted to 
one of the determined azimuths, the torsion angle was read off for 
the two directions of the field, the current broken, the electromagnet 
turned to the opposite azimuth, and so on several times. At the end 
of a series a measurement with one of the first fields was repeated 
as a control. . 7 
Sources of error , difficulties, collections, and controls. 
1. Inhomogeneity of the magnetic field. As will be seen from the 
following discussion this source of error is by far tile most important 
in our case and is indeed the only one that need be taken into 
account. If we assume i 
may be represented by 
the field near the centre of the pole-gap 
expression 
of the form 
(3) 
where H, is the field in the centre, and r and 6 polar coordinates 
of a point in the pole-gap with respect to the centre as origin 
Let us now replace the ellipsoid by a vertical disc whose diameter 
is equal to the major axis of the ellipsoid; by the expression 
for the energy of the magnetized disc in the field and differencing 
it with respect to the angle between the disc and the lines 
we obtain tor the couple caused by the mhomogeneity of the field. 
Cf. P. Weiss and H. Kamerungh 
