674 
NATURE 
[May 19, 1923 

The difficulties arising from the large electric field, 
and from the inconsistency in sign between the hori- 
zontal and vertical components may be avoided, at first 
sight, by supposing, with Sutherland, that the earth 
may be regarded as two superposed spheres of positive 
and negative electricity, the diameter of the negative 
sphere being greater than that of the positive. The 
electric field at external points would be zero, but the 
magnetic field would not be zero. The actual density 
of positive and negative electricity in the earth is so 
great that, if all the positive and negative electricity in 
a cubic centimetre could be concentrated at two points 
one centimetre apart, they would attract each other 
with a force of the order 107° tons ; and, on account of 
this, it is only necessary for the radii of the two spheres 
of the size of the earth to differ by 2x 10~§ cm. (the 
diameter of a single molecule), in order to ensure that 
the two, rotating together, would give rise to a field of 
the order of magnitude of the earth’s field. Unfortun- 
ately, however, we find that the electrostatic forces 
opposing even this small separation are enormous, 
amounting to more than one thousand million volts per 
centimetre at the surface of the inner sphere. 
Regarding forces which suggest themselves as possibly 
available for causing electrostatic separations of the 
above or allied kinds, we have, in the first place, gravity, 
tending to pull the free electrons towards the earth’s 
centre, then centrifugal force tending to make them fly 
to the surface. Another possibility arises from an 
action analogous to the Thomson effect, by which the 
electronic density tends to decrease as we descend 
towards the earth’s centre, on account of the increase of 
temperature. These effects have been submitted to 
calculation by the writer, and it appears that the first 
gives rise to a field only 10 of the earth’s field, and in 
the wrong direction, the second to a field about ro~ 8 of 
the earth’s field, but of a type widely different from that 
of the earth, and the third to a field in the right direction 
but amounting to only 10-1 of the earth’s field. 
As a general rule, we may say that it is practically 
hopeless to seek an explanation of the earth’s magnetic 
field on the basis of the rotation of charges which have 
been separated against electrostatic attraction, since 
the mechanical forces necessary to produce the requisite 
separation must be, in all cases, enormous. 
If the earth were made mainly of iron, its rotation 
would, by gyroscopic action, bring about a partial 
orientation of the molecular magnets ; and it has been 
experimentally demonstrated by S. J. Barnett that iron 
can be magnetised in this way. The effect in the case 
of the earth is, however, extremely small, and is only 
sufficient to account for a magnetic field 2 x 107! times 
that of the earth. 
The suggestion has been made that the interior of the 
earth may be endowed with enormously high perme- 
ability, and that, in consequence, a very weak force 
would be sufficient to cause strong magnetisation 
therein. We must remember, however, that the very 
creation of a state of magnetisation within a sphere 
brings about an internal demagnetising field which is, 
as a matter of fact, equal to the external field at the 
equator. Hence any primary magnetising agent which 
is to be ultimately responsible for the earth’s field must 
be of such intensity that it will produce, on the mole- 
cular magnets, forces at least equal to the forces which 
NO. 2794, VOL. 111 | 

would be produced on them by a magnetic field equal 
in intensity to the earth’s magnetic field at the 
equator. 
Any theory attempting to account for the earth’s 
magnetic field on the basis of currents circulating within 
the earth, calls for some explanation of the electro- 
motive force wherewith to produce the currents. In 
this connexion it is of interest to recall a calculation by 
A: Lamb, to the effect that if currents were caused to 
circulate in a copper sphere of the earth’s size, and the 
electromotive forces which caused them were removed, 
the currents would take ten million years to decay to 
one-third of their initial values. Attempts to account 
for the earth’s field in this way have met criticism on the 
basis of the enormous currents which would be calculated 
by extrapolation back, even to epochs not more remote 
than those during which the earth’s crust has been 
solid, so that, unless there is some reason for supposing 
that the conductivity is, or has been, in the past, even 
greater than copper, we are confronted with accounting 
for the enormous amounts of energy necessary to have 
produced the field initially. 
The actual current density within the earth necessary 
to account for the earth’s field is very small, being, for 
example, of the order ro~® ampere per square centi- 
metre on the surface at the equator for the case where 
the current density is proportional to the distance from 
the axis of rotation. If, taking a sphere of iron, we 
assume about 10% free electrons per c.c., it is only 
necessary to suppose that the mean velocity of the 
electrons at the earth’s surface, relative to the centre, 
differs from that of the periphery by one part in 7 x ro!® 
in order to account for this current. It is, perhaps, not 
too much to hope that a fuller knowledge of the 
mechanism of conduction in solids than we have at 
present may lead to an explanation of such a small 
difference as arising directly on account of the earth’s 
rotation. 
There is always the chance that the origin of the 
earth’s field may have to be sought in some funda- 
mental but small departure from the ordinary electro- 
dynamic laws. In this connexion we may recall 
Lorentz’s theory of gravitation, according to which 
gravitational forces may be accounted for by supposing 
that the attraction between two unlike units of charge is 
different from the repulsion between two like units. 
Paying due regard to the care necessary in defining 
electrical neutrality in this case, the theory may be 
shown to lead to the conclusion that, in order that the 
free electrons in a body shall be in equilibrium, the body 
must acquire a charge density to an extent not wholly 
determined by the weight of the electrons. Schuster 
has discussed the possibilities in this regard, but it 
would appear that, under the most favourable assump- 
tions, the density would be insignificant as regards 
producing, by its rotation, a magnetic field comparable 
with the earth’s field. 
A greater measure of success is attained by making a 
somewhat similar assumption concerning the magnetic 
field produced by a moving charge. We first observe 
that a magnetic field is ultimately measured in terms of 
the force which it exerts on a moving electron; for 
even a material magnet which may be used in the 
measurement derives its properties from electrons, 
rotating within it. In analogy with the case of electro- 
