1292 
As to 1 and 3a it is easy to find out which of these two cases 
occurs. We need only consider the direction of the deflexion that is 
obtained, when the current caused by a positive reversal is sent 
through the windings of the compensating coil only and not through 
those of the evlinder. 
Suppose e.g. the deflexion to be to the positive side. Then before 
the positive reversal, the current flowing through the coil has given 
rise to a couple in the negative direction acting on the- magnet AB 
by the compensating coil. But that couple served to neutralize the 
action of the residual field on the magnetism of the cylinder. This 
action therefore had the positive direction and the same may be 
said of the action of the residual field on the remanent magnetism 
after a positive reversal. Thus effect 3a has a direction opposite to 
that of effect 1. 
§ 7. Description of the experiments. First different alterations of 
the apparatus were tried in preliminary experiments, but though 
these gave clear indications of the effect we were seeking for, we 
did not obtain perfectly satisfactory results. Then however we suc- 
ceeded in making a series of experiments, by which the existence of the 
effect became quite evident. At least I should think it difficult to 
interpret in another way the result I am going to describe. | regret 
that these observations could not be repeated, as they were made 
on the last day that was at my disposal for this work. 
The period of the pendulum was found to be 1,856 sec., that of 
the cylinder 1,912 sec. The damping of the cylinder was so great 
that the amplitude diminished to half its original value in 39 half 
oscillations, while by means of an added known moment of inertia 
| found 0,814 gr. em* for that of the cylinder. 
The equation of motion of the cylinder for free vibrations is 
da , da 
Te ree 
where w is the angle of deflexion, and g the coefficient of resistance 
divided by the moment of inertia, while , denotes the frequency 
of the free vibrations as it would be without damping. A solution 
of this equation is: 
«= C e—*9't cos2x—, 
0 
An 
Vn? ig 
Let us now introduce an Sl force which produces an accelera- 
T= 
0 
