RESEARCHES IN MAGNETISM. 
631 
of these progressive changes, and may then study the differential effects of heating 
and cooling, after the magnetic changes have also become sensibly cyclic. 
I have made a number of experiments of this kind with the view of seeing whether, 
after the magnetic changes become sensibly cyclic, they exhibit static hysteresis with 
respect to the changes of temperature. Since there is clear evidence of such 
hysteresis in changes of magnetism produced by other means, namely by changes of 
magnetic field, and also by changes of stress, I expected somewhat confidently to find 
hysteresis in the relation of magnetism to temperature. But in this expectation I 
have been altogether disappointed. 
§ 115. A long iron wire was fixed inside a vertical glass tube, on which was wound 
a magnetising solenoid. The tube was connected by india-rubber piping at one end 
to one or other of three small boilers, capable of supplying a steady current of steam, 
alcohol vapour, and sulphuric-ether vapour respectively, or to a water-cistern filled 
with cold water. The other end of the tube led to a water vessel which served as a 
condenser. An adjacent mirror magnetometer, level with the top of the iron wire, 
measured the changes of magnetism in the usual way. The method of procedure 
was this:— 
Steam and cold water were alternately passed through the tube many times, until 
the magnetic state of the wire was observed to change from one to the other of two 
nearly steady values. Then readings of the magnetometer were taken during the 
passage through the tube of (l) cold water; (2) ether vapour; (3) alcohol vapour; 
(4) steam; (5) alcohol vapour; (6) ether vapour; (7) cold water. This completed 
a cycle of temperature changes in which two intermediate points were fixed during 
each of the processes of heating and cooling. 
This method was adopted in order that the magnetised iron might be exposed 
sufficiently long to an atmosphere of one definite temperature to give it time to 
assume that temperature throughout, and so avoid any possibility of error due to the 
sluggishness with which changes of temperature took place. The stream of vapour 
was in every case kept up until the magnetometer reading became perfectly steady. 
In some experiments, in place of a magnetised wire, a magnetised iron pipe was 
used, surrounded by a non-conducting jacket, and steam and other vapours were 
passed through the iron tube itself. This had the advantage that the position of 
the magnetised iron, with respect to the magnetometer, could be maintained more 
absolutely constant and unaffected by temperature changes than when the magnet 
consisted of a wire inside a glass tube. The difficulty of fixing the magnet so 
that the process of heating and cooling it might not produce displacements whose 
effects on the magnetometer were comparable with the true change of magnetism, was 
somewhat formidable on account of the smallness of the whole range of magnetic 
change caused by alternations between atmospheric temperature and 100' C. 
The small range of magnetic change also made it needful to raise the magneto¬ 
meter to a condition of great sensibility. For this purpose a directing magnet 
4 m 2 
