94 Dr. W. F. G. Swann on 



an atom, Is a fact .sufficiently tempting, and suggestive, fco 

 encourage us to try to get over the difficulties associated 

 with the theory, but there seems little hope of being able to 

 account for the perfect radial distribution required. Con- 

 siderations of a statistical nature would suggest, that, if <f> is 

 the angle made by the axis of a doublet with the radial 

 direction, the mean value of cos <fi is of the order of the 

 ratio of the potential energy of a doublet in the radial con- 

 figuration, to the kinetic energy of a gas molecule at the 

 same temperature. All ordinary views as to the mean- by 

 which the potential energy of the configuration might arise, 

 however, lead to a value of cos (f> of far too small an order of 

 magnitude. When once a set of doublets got into radial 

 c mfiguration, they would have a fairly strong tendency to 

 retain that state, just as the molecular magnets which con- 

 stitute a magnetized ring of iron tend to hold themselves in 

 the magnetized configuration; though it must be admitted 

 that with our knowledge of the temperature conditions 

 inside the earth, and the absence of our knowlege of any 

 example of permanent electrostatic retentiveness, do not lend 

 support to the hypothesis. 



The form of the theory is such as to make the field at the 

 e [uator proportional to coa, which would mean that, apart 

 from temperature considerations, it would be quite easy i»> 

 produce a field equivalent to that of the earth, by rapidly 

 rotating a sphere in the laboratory. 



It may be shown from the expression on page 93, thai 

 the surface electric field required by the theory is about 

 10" E.S.U., and the difficulties in this respect would have t*> 

 l)> overcome as in the 1 last problem, by imagining a surface 

 distribution such as to just annul tin 4 electrostatic Held. 



The possibility of an explanation of the field as arising from 

 a directive <>r'/eitt<ifio>i of the molecular magnets which re* 

 present the atoms. 



Consider a sphere in which there are /> atoms per c.c. 



Let >n he the moment o{ the atomic niiignot to which the 

 resubmit atomic whirl in the atom at large distances IS 

 equivalent. Let (/> be the angle made by the axis oi the 

 atomic magnet with the axis of the earth's rotation, and lei 

 j be thi> average value of cos<i for all the atoms in a cubic 

 centimetre. Considering first a case where j is simply n 

 function of «k we find, for the horizontal magnetic held 



