THE ZEEMAN EFFECT AND DISPERSION. 425 



field and into a linear component in the direction of the 

 field, all these being of the same period. The effect of the 

 magnetic force in the simplest and apparently very common 

 case is to leave the component in the direction of the field 

 unchanged and to make one of the circular components 

 rotate more rapidly and the other less rapidly than when 

 there is no magnetic force acting. The changes of fre- 

 quency are directly proportional to the strength of the 

 magnetic field, but are independent of the intensity of the 

 vibration of the molecule. This simplest action is what we 

 would expect to take place when the magnetic force acts on 

 a moving electron which is equally free to vibrate in every 

 direction. On account of this important condition, however, 

 we need not be surprised to find more complicated effects 

 produced in the case of a large number of spectral lines. It 

 is very improbable that the molecules of most gases are 

 so symmetrical that all vibrations in every direction are 

 equally possible, and, as a matter of fact, very complicated 

 effects have been observed in the case of a large number ol 

 lines. The effect of magnetic force in this simplest case is 

 to make the axis of the orbit of an electron rotate round the 

 line of magnetic force. We might expect a disturbing force 

 to produce other changes in the orbit, in general ; such, for 

 example, as causing the inclination of the orbit and its ec- 

 centricity to change. Actions such as this would produce 

 complications in the spectra, as has some time ago been 

 pointed out by Dr. Stoney. 



These theories as to light vibrations being due to simple 

 harmonic vibrations of electrons are, however, almost cer- 

 tainly only provisional. They^ require the forces acting on 

 the electrons to be directly proportional to the distance from 

 their positions of equilibrium. It is unlikely that electrons 

 can be subject to such forces and their vibrations are much 

 more probably of the nature of perturbations of orbital 

 motions executed under quite other laws of force. For ex- 

 ample, the rotation of the lunar nodes is a vibration of, a 

 system, the earth, sun, and moon, which is controlled by 

 forces varying inversely as the square of the distance and is 

 one whose period is almost independent of the eccentricity of 



