SPECTRAL LINES 445 



tions emitted by a source of light are changed under the action 

 of a magnetic field. Humphreys argued that this being so, the 

 luminous particles must have a magnetic field of their own, and 

 consequently, since they can be acted upon by an external 

 magnetic field, they must of necessity be acted upon by the 

 fields of the neighbouring particles. He took for his model 

 of the atom that pictured by Sir J. J. Thomson, in which a 

 number of coaxial rings of electrons rotate inside a sphere of 

 positive electricity uniformly distributed ; and with certain 

 assumptions as to the radius of the sphere and the number of 

 electrons contained in the atom he was able to calculate the 

 strength of the atomic magnetic field. In the case of the iron 

 atom, by assuming it to contain 5,000 electrons the strength of 

 the magnetic field at the centre of the atom was found to be 

 57rT0 r C.G.S. units. Humphreys l also found that the observed 

 displacement could be accounted for by means of a strength of 

 field of 45' io 7 units. This is a field ten thousand times as 

 strong as the field of the strongest electromagnet used in pro- 

 ducing the Zeeman effect, and it seems a priori improbable that 

 the atoms of all metals could have such enormously strong 

 magnetic fields without their existence being revealed in other 

 ways. Humphreys admitted that the result was, at first sight, 

 somewhat startling, but argued that the magnetic properties of 

 atoms when luminous might be vastly different from the 

 magnetic properties of cold masses of the pure elements. The 

 Zeeman effect, however, seems to disprove the existence of 

 atomic fields of such magnitude. In the elementary theory of 

 this effect, the assumption is made that the atomic magnetic 

 fields are small compared with the externally applied field, and 

 the agreement between the calculated separation of the com- 

 ponents into which the original line is resolved and the observed 

 separation is sufficient justification of the assumption. The 

 error in Humphreys' calculations appears to lie in the assump- 

 tion as to the number of electrons which are contained in 

 an atom. 



Modern researches in connection with radioactivity indicate 

 that this number is roughly equal to half the atomic weight. 

 This may be, and probably is, an under-estimate. A model of 

 the hydrogen atom which contains only one electron appears to 

 be too simple and too unstable to be the true one : yet it should 



1 Astroph. Journ. 27, p. 194, 1908. 



