982 
soon as it appears for the first time at A and C' (fig. 1) or D (fig. 
2) is still very weak at these points. Thereupon its intensity increases 
and at B and D it gradually disappears again. 
The electrolytic current control mentioned above, is not sufficiently 
continuous for galvanometric observations. By using the following 
method, however, we could obtain a very continuous current variation, 
though the total magnitude of it was limited. 
The magnetising coil M (fig. 3] is put in series with a precision 
VODVOVO | 
ammeter A, rheostat /?,, key S, commutator G, and battery K. 
Moreover M is shunted by the variable resistance ft, while S is 
put in parallel to a platinum wire DH, the latter being placed ina 
glass tube. This tube has two connections. The one at D goes through 
the tap # to a Gaedepump the other at H to a long capillary glass 
tube. After S is closed, we can, by variation of A, and R,, and 
with the aid of the commutator G submit the ferromagnetic wire 
C a few times to a complete magnetic cycle. For the sake of 
clearness the wire C’ has been drawn in fig. 3 outside the coil M, 
though actually of course its place is inside. Thereupon, after the 
pump has been working for some time, the current through M is 
made zero and the key S is opened. With the aid of the resistances 
the platinum wire DH is brought to a dull red heat and the magne- 
tising current is brought to a value which brings the ferro magnetic 
substance on the steep part of the hysteresis curve. After the tap 
has been closed, the pressure of the air in DH slowly increases; 
due to the leakage through the capillary tube, the platinum wire 
gets colder, and the magnetising current through J/ increases very 
Fig. 3. 
