536 
PROFESSOR J. A. EWING ON EXPERIMENTAL 
been impossible bad the successive tests involved the unknown and not inconsiderable 
differences of magnetic quality which different specimens would have possessed or 
different annealings given. 
In each case the magnetising force due to the solenoid was raised by steps to the 
same value, 34‘2 c.g.s. units, and lowered by steps to zero. The curves of Plate 57, 
fig. 3, show its relation to 33 at the centre of the rod in the various cases. The full 
lines are the “ on ” curves, or curves showing the relation of 33 to the solenoid’s 
magnetising force during its application: the dotted lines are the “off” curves, and 
show the values of 33 during decrease of the solenoid’s magnetising force. The number 
affixed to each curve gives the ratio of length to diameter in the test it belongs to. 
The abscissas give the magnetising force due to the solenoid, which of course diff ers 
from <§ by the amount of the variable field which the rod exerts upon itself. To avoid 
confusion in the figure, a portion only of each pair of curves is drawn, except in the 
case of the longest and shortest rods, for which the curves are drawn in full. The 
greatest magnetisation reached was sensibly the same for the longest rods and for 
those of intermediate lengths. At the beginning of each “ on ” curve a small amount 
of initial magnetism will be noticed, which is the part of the previous residual 
magnetism not shaken out by tapping. 
§ 16. It appears that the rod 300 diameters long differs little in its magnetic 
behaviour from an indefinitely long rod or from a ring. Its residual magnetism is 
85 per cent, of the total, and its greatest value of /x is about 3500. The “on” curve 
of the 200 diameters rod falls not much below that of the 300 diameters rod, but its 
“ off” curve is notably different near the conclusion, where the demagnetising influence 
of the ends becomes sharply apparent. The result is to reduce the residual magnetism 
to 60 per cent, of the total. In the 150 diameters rod the residual magnetism falls 
to 39 per cent., in the 100 diameters rod to 20 per cent., in the 75 diameters rod to 
9 per cent., and in the 50 diameters rod to about 6 per cent. The “off” curves of 
the shorter rods are distinguished by a long straight descent towards the axis of 33, 
showing a sensibly uniform rate of demagnetisation during the later part of the with¬ 
drawal of the externally applied force. 
The comparative difficulty of magnetising a short rod is shown by fig. 3 in a way 
which requires no comment. If we were to infer values of /x and k from the test ot 
the 50 diameters rod, neglecting the fact that the magnetising force due to the 
solenoid is not the whole of 35, we should find for the maximum of /x 656, and for that 
of k 52 ; and these coefficients are much more nearly uniform for various values of the 
magnetising force than when we determine them in a legitimate manner by observ¬ 
ing the magnetisation of very long rods or of rings. 
§ 17. From this and other experiments, I concluded that even when dealing with 
the softest iron, we may take a rod whose length is not less than 300 diameters as 
giving results scarcely different from those given by a ring or a longer rod, and in subse¬ 
quent experiments rods were almost exclusively used whose length was from 300 to 400 
