RESEARCHES IN MAGNETISM. 
587 
§ 75. Initial form of Curves of Magnetism and Load .—It has been mentioned 
above that when a change from loading to unloading, or from unloading to loading, 
occurs, the new curve starts off tangent, or nearly tangent, to the direction in 
which loads are measured. In other words, the initial change of magnetism appears 
to be indefinitely small relatively to the change of stress whenever the operation 
is reversed from loading to unloading, or vice versa. In the experiments which have 
been already cited the steps by which the load was increased or diminished were 
too large to allow the initial form of the curve to be very well defined, and as 
the matter seemed important as throwing light on the character of hysteresis, a 
set of observations were made in which the process of loading or unloading 
was conducted by very gradual stages at the places where the precise form of the 
curve was to be examined. A number of small weights were prepared, which could 
be added to or removed from the other weights hanging from the wire, wherever it 
was desired to have a number of closely consecutive points in the curve. Great 
care was taken in this experiment to avoid the slightest disturbance of any kind, 
since it was evident that the least trace of vibration would vitiate the results. 
The wire, which was the same as before, and still hanging vertically in the earth’s 
field, was loaded in the usual manner up to nearly 5 kilos.—a place where the on 
curve is very steep. The load was then removed, at first by very small parts, in order 
to determine as precisely as possible the initial form of the off curve. The following 
are the observations. The magnetometer scale was altered before this experiment, 
but the readings, although different from those taken in former experiments with the 
same wire, are still proportional to the total magnetisation. The wire had been 
slightly shaken before the loading began, and consequently the operation was not 
cyclic. 
Here, then, it is clear that the off curve, beginning from a steep part of the on curve, 
starts off as nearly as can be observed tangent to the line of loads. 
Similar observations were made at two other points, namely, at the upper end of 
