1282 
besides this the importance of the observed phenomenon for the 
theory of electrons and for our knowledge of the nature of magne- 
tism sufficiently justified a repetition of the investigation. In the 
Berlin experiments the phenomena were so much complicated, that 
I tried to find a more elegant method, by which part of the sources 
of error could be eliminated, so that it might surpass the former 
method in clearness and convincing power. 
I have succeeded in rigorously separating the effect which I should 
like to call the EiNsrein-effect, from the secondary phenomena. This 
time again I have not had in view an accurate quantitative deter- 
mination; yet it may be mentioned that the quantitative agreement 
between experiment and theory is quite satisfactory. At the same 
e 
time a way is opened for a later accurate determination of —. 
Mt 
§ 1. Principle of the new method. 
In the former method a cylinder that could perform torsional 
vibrations, was brought into an alternating magnetic field of about 
the same frequency as that of the free oscillations of the cylinder. 
From the amplitude of the observed vibrations of the cylinder 
the existing moment of momentum could be caleulated. As we have 
seen the disturbing effects consisted in: 
1. the action of the static terrestrial field on the alternating hori- 
zontal components of the magnetism in the cylinder ; 
2. the action of the alternating field of the coil on the horizontal 
components of the statie magnetism in the cylinder; 
3. the action of the horizontal components of the alternating field 
of the coil on the alternating magnetization of the cylinder. 
As we remarked already, we need not fear any disturbance by 
effect 3, as it has double the frequency of the Ernsrrin-effect. We 
must keep in mind however, that disturbances of this kind will 
appear as soon as the dynamo gives e. g. an asymmetrical sinusoidal 
current. A very small asymmetry already may reach the value of 
the remaining part of the terrestrial tield (which is never completely 
compensated). An asymmetry of the current may give rise to a 
disturbing effect having the period of the free vibrations of the cylinder. 
The characteristics of the new method are the following: 
ist a very slow resonance, which enables us. to investigate the 
difference in phase between the effect and its cause; 
2nd a complete elimination of the sources of error 2 and 3. For 
this purpose the iron cylinder is not placed in the magnetic field 
of a fixed coil, but the wire is wound on the cylinder itself. As 
