1905 - 6 .] Electric Oscillations and Magnetic Properties of Iron. 35 
These conditions of field superposition are entirely dissimilar, 
and the recognition of this fact is of primary importance. It 
may be observed that if in the former case (A) the field is 
varied after the superposition of oscillations, the order of super- 
position passes into those under the B conditions, a change of 
field being then superposed upon permanently acting oscillations. 
It is evident that the converse does not hold. 
Experimental Methods under A conditions. 
First, after demagnetisation of the iron, a fixed maximum field 
is put on (by increasing reversals, to secure as far as possible 
symmetry about the zero of induction), and reversed twenty times. 
The plus change of inductions due to the twenty-first reversal is- 
measured. The co-directional oscillations are now superposed, and 
the plus induction change measured. Oscillations and field are 
now put off. Second, the iron is again demagnetised, and the 
same fixed maximum field put on in the same way as before and 
reversed twenty times. The plus induction change due to the 
twenty-first reversal is measured. A single step is now taken to 
any given point on the hysteresis loop, and the minus change of 
induction measured. The co-directional oscillations are now r super- 
posed and the plus or minus reading taken. Oscillations and 
field are again put off. This second process is repeated for a 
sufficient number of points all round the loop. The whole process 
is repeated for transverse oscillations. 
Three curves result from the reduced galvanometer readings 
taken as above described. These are plotted in figs. 1 and 2, 
when the maximum cyclic induction values are (without oscilla- 
tions) B = 780 and B = 5620 respectively. The scale of fig. 1 is 
for both ordinates double that of fig. 2. The ordinates measuring 
induction are in C.G.S. units; the abscissae measuring field, in 
arbitrary units. 
Summary of Results under A conditions. 
At and near extreme cyclic values the superposition of oscillations 
produces for low values of field (see fig. 1) a relatively large 
increase of induction ; for higher values of field (see fig. 2), a 
