220 
Proceedings of the Royal Society of Edinburgh. [Sess. 
known that the application of stress often changes the magnetic condition 
of a metal. The method of demagnetising was a novel one, and consisted 
in connecting the primary coil to an alternator running at 500 revolutions 
per minute. The current was adjusted till it was slightly above the current 
to be employed in the test. The alternator was a very free-running 
machine. On breaking the field switch, its speed and current decreased 
together slowly. The eftectiveness of this method of demagnetisation was 
proved by means of a very delicate galvanometer, hardly a trace of deflec- 
tion beino; obtainable. 
Specimens Tested. 
Three different specimens were tested : one each of cast iron, soft 
wrought iron, and mild steel. 
Experimental Data and Results. 
The cast-iron piece was tested first. At 5 tons per square inch its 
induction curve was practically^ the same as at no compression. At 10 tons 
per square inch, however, a distinct lowering of the curve at all points 
occurred. At 15 tons per square inch the specimen broke. This was seen 
to be due to the shearing action of the two compression blocks AA (fig. 2). 
In the next tests the corners of these blocks were rounded over in order to 
minimise this shearing action. The curves for this test are shown in fig. 3, 
and the figures are given in Table I. No reliability can be placed upon 
this test, since in all likelihood the sudden chano;e in the induction was 
due to a split developing in the piece. This is indicated by its giving way 
at such a low pressure as 15 tons per square inch. 
The piece of mild steel was next tested, and then the wrought-iron 
piece. Both gave good readings up to 50 tons per square inch in each case. 
The figures of these tests are given in Tables II and III, and the induction 
curves in figs. 4 and 5. 
From the nature of the curves in these two cases it was seen that 
the greatest change in induction with compression took place first at 
the low currents and then later at the high currents. That is, in pass- 
ing from curves I to lY and I to VI in figs. 4 and 5 respectively, the 
induction changes are not regular. Had they been regular the various 
curves would have been similar. Owing to this the curves showing the 
relation between compression and induction are drawn and shown in 
figs. 6 and 7. 
Taking fig. 6 first, which gives the relation for the wrought-iron piece, 
