1291 
positive as negative. Developing in a series we are led to an 
expression of the form 
t 
7 
q sin 2 
with positive or negative amplitude . 
It is seen by this, that the second effect is always a quarter of 
a period in phase behind effect (1, 3). 
If all the three effects exist, we may replace (@) and (2) by a single 
alternating couple, which determines the motion of the cylinder. 
Its amplitude is Wp? + q’, while its phase lies between those of 
(1, 3) and 2, 
We may use this to separate the effects (1, 3) and 2. When, in 
effecting the compensation II, we come nearer and nearer to 
the right position of the coil A, the amplitude q and at the same 
time Vp? ae 7 will decrease. After the coil has passed through the 
position in question, 17 p+ will increase again (the resonance 
in the 2"? effect being then caused by the action of the compen- 
sating coil on the small magnet, which now exceeds that of the 
residual field on the cylinder). If we succeed in reaching a minimum 
of resonance, the effect (1,3) will be free from 2 and this will be 
confirmed by the phase of the effect. Since the beginning when the 
second effect had the greatest influence till the minimum, when it has 
disappeared, the phase must have changed by a quarter of a period. 
It remains to separate the effects 1 and 3 (a, /). For this purpose 
we may change by 180° the azimuth of the suspension arrangement 
of the cylinder. For the same state of magnetization the couple 
with which the residual field acts on it will then change its sign 
and the action of the compensating coil on the magnet fixed to the 
cylinder, if we leave the current in the coil unchanged, will likewise 
be reversed. So the compensation, when it has once been reached, 
_will remain. But it is evident that both the effects 8a and 34 will 
be reversed. Both the remanent magnetism (Ol in fig. 2) and the 
difference J/—J/’ (that is to say their horizontal components) are 
turned through 180°, while the remaining tield has not changed. 
It is clear that the change of azimuth does not influence the first 
effect. Thus we are able to make the effects 1 and 3(a, 6) act in 
the same or in the opposite sense’). 
ain reality ‘the turning by 180°, by which among other things the distance 
between the magnet AB and the coil K was altered more or less, required so 
many changes in the apparatus that after it a new compensation was necessary. 
This however alters nothing in’ the above considerations on the change of sign 
of effect 3(a, b), with which we are here concerned. 
