354 Proceedings of the Poyal Society of Edinburgh. [Sess. 
poles pointing inwards ; and he has pointed out that if, in addition, the 
rotating unit has like poles pointing outwards — if, for instance, we sup- 
pose that the rotating unit has two external south poles, while the fixed 
magnets have their north poles pointing inwards, — such a system would 
have no magnetic polarity, and would he unaffected by an external 
magnetic field, approaching closely to the properties of the atoms of 
most of the elements which are only very slightly para- or diamagnetic. 
But if this system is approached by a small magnetic unit, the inner 
rotating unit might be swung from one position of stability to another 
with the absorption and dissipation of a definite amount of energy. 
Now it is evident, to return to the former discussion of the nature of 
the atom, that if we assume the atom to consist of electrons with magnetic 
properties and capable of rotation round a centre, such an atom will have 
properties corresponding to the atom postulated by Sir Alfred Ewing ; and 
that, while it need not have any magnetic polarity or be affected by a 
magnetic field, will, if approached sufficiently near by another atom to 
bring individual magnetic units into play, reproduce the phenomena which 
he has investigated. 
Sir Alfred Ewing in his investigations has assumed the rotating unit 
to be in the centre and the fixed magnets outside, but in order to bring 
his conception of the atom into line with the conception of the chemical 
atom, we shall suppose, in the first instance, the fixed magnets to radiate 
from the centre and the rotating units to be in the outer ring, and in 
order that the atom shall have no magnetic polarity, we shall suppose 
that we have like poles pointing outwards. Let us postulate, to avoid 
confusion, that the north poles are pointing inwards to the nucleus and 
the south poles pointing outwards. It is evident that if we assume the 
atoms of a given element to have the south poles of the fixed magnets 
pointing outwards, we must make the same assumption for the atoms of 
all the elements, otherwise we should have those with north poles and 
those with south poles respectively rushing together. 
We shall then picture the central stable group of electrons as consist- 
ing of magnets radiating from the centre, which can be diagrammatically 
illustrated in one plane, as shown in fig. 1. A monovalent element like 
potassium could then be represented diagrammatically with one magnetic 
electron in the outer shell, which we shall, for our present purpose, 
represent as a little magnet able to rotate on a central pivot. 
Let us now suppose a second atom to approach the first atom, and let 
us suppose, for simplicity, that the second atom consists of fixed magnets 
alone (fig. 2). As it approached it would be repelled by the south 
