Origin of Gravitation and Terrestrial Magnetism. 687 
Now G=666x10-! (Burgess, Physical Review, 1902), 
X10 c=3x10-, and a=2x10—-% (Hlectric Origin 
of Rigidity, section 6). This value of a is deduced from the 
velocity and energy of a negative electron in the cathode 
stream. Tor an electron forming part of an atom the value 
may be different. Provisionally assuming that a has the 
same value in an electron of matter as in a free electron we 
find from (6) that 
te Oe ee aes fate ee 
As we shall see that 6 and ¢ are of opposite sign and nearly 
equal, b+c¢ is the small difference of the numerical magni- 
tudes of b and c. Gravitation then can be traced to an 
electric origin, if between two electric doublets there is an 
outstanding electric attraction which is of the order 10-* 
times the force between two of the electrons, one in each 
doublet. Ordinary experimental evidence of a difference of 
this order of magnitude has hitherto been out of the question. 
We turn then to phenomena on the same large scale as 
universal gravitation in astronomy, to find whether it is pos- 
sible to get further evidence of this difference, terrestrial 
magnetism holding out the greatest promise. Now in “A 
Possible Cause of the Harth’s Magnetism and a Theory of its 
Variations,” and in “ The Cause of the Harth’s Magnetism ”’ 
(Terrestrial Magnetism and Atmospheric Electricity, v. 1900, 
and viii. 1903), I have shown that the Earth’s primary 
magnetic field can be accounted for by spherical distributions 
of positive and negative electricity within the Earth and 
revolving with it, the negative being on the whole nearer 
to the surface. In the second of these papers it was shown 
that if each two atoms of the Harth’s substance contain the 
electron pair %b in accordance with Helmholtz’s theory of 
chemical valence, and if the total electricity of each kind is 
supposed to form a uniform sphere of nearly the same radius 
as the Harth, the negative sphere need have a radius exceed- 
ing that of the positive sphere by only 10~* cm., that is by 
about the diameter of an ordinary molecule. In that paper 
the origin of this small difference in the radii was not inves- 
tigated, but was merely commented upon as implying a small 
tendency of each b ina molecular doublet to take up a position 
farther trom the centre of the Earth than . 
But on the present occasion we are not confining our 
attention to electrons only as they appear in electrolysis and 
the theory of chemical valence, but are prepared to take 
account of the much larger number of the electrons forming 
the substance of an atom. As the inertia of the negative 
3 B2 
