1890-91.] Prof. Knott on Interaction of Magnetisations. 125 
slipped through a glass tube a little longer than the internal metal 
wall of the tube on which the well-insulated magnetising coil was 
coiled. This metal wall was used as the return channel for the cur- 
rent, so that all possibility of an appreciable direct electro-magnetic 
effect of the circuit upon the magnetometer was quite excluded. 
The end of the iron or nickle wire nearer to the magnetometer was 
gripped vice-wise by a cleft metal plug of conical shape, which, when 
pushed into the end of the magnetising coil, established good pressure 
contacts between the end of the wire and the metal wall. The 
other end of the wire projected backwards out of the magnetis- 
ing coil sufficiently to enable it to be clamped to a twisting 
gear. From this end, and from the neighbouring end of the tube 
wall, wires well insulated and well twisted together were led to the 
commutator connected with the battery. 
It was only after a series of experiments, in which the effect 
mentioned in my earlier note was observed to take place sometimes 
in one way and sometimes in the other, that I came to the conclusion 
that there must be an original permanent twist in the wire. Thick 
wires showed comparatively small effects, steel wire showed the 
effect only when it was drawn thin, and so on. Nickel wires be- 
haved in an especially confusing manner, even after the greatest 
care was taken to insert them untwisted into the magnetising coil. 
The twisting gear mentioned above was added to the apparatus so 
as to make a direct experiment upon the effect of a small voluntarily 
applied twist. That large twists might reasonably be expected to 
produce peculiar disturbances in the magnetic distribution under the 
combined influence of longitudinal and circular magnetisations will 
be at once admitted when it is remembered to what an extent twist- 
ing affects the result of either taken alone. I was not prepared, 
however, for the pronounced influence exerted by even a small twist 
previously existing in the wire upon subsequently applied longi- 
tudinal and circular magnetisations. 
To prevent the possibility of such twists being inadvertently 
applied, the wire was, in the later experiments, annealed with a 
weight hanging free at the one end. This left the wire permanently 
magnetised under the influence of the earth’s vertical field ; but in 
the stronger cyclic field to which it became subjected the wire soon 
lost all trace of this original polarity. After being annealed the 
