Spectra and Planck's Law. 425 



shall consider the type of atom which is required to satisfy 

 the assumption we have made in the preceding investigation. 

 This atom consists of a field of electric force which may 

 be regarded as made up of a series of shells of attractive 

 and repulsive force following one another alternately, the 

 radii of the boundary of these shells, which are places where 

 an electron would be in equilibrium, being in harmonical 

 progression. Superposed on the field of electric force is a 

 field of magnetic force, also arranged in shells, the outer 

 boundary of the magnetic field coinciding with a place 

 where the electric force vanishes. I contemplate that when 

 the atom is exposed to such conditions as may arise in strong 

 electric discharges or other methods of producing spectra, 

 the outer layers of this magnetic field may get detached 

 and the boundary of the magnetic field come close up to the 

 centre, that, in fact, there is what might be called a magnetic 

 ionization of the atom, and that the atom resumes its normal 

 magnetic state when the electric discharge, &c, ceases. 

 The atoms in a luminous gas thus possess a double mani- 

 fold n ess, one arising from the different positions of the 

 electrons in the atom, the other from the variations in the 

 magnetic boundary of the atom. The first manifoldness 

 would give rise to different lines in the same series, the 

 second to a number of different series most of which would 

 only be emitted by the special type of atoms produced when 

 .an electric discharge passes through the gas. 



So far we have only considered the case when only one 

 electron was in the atom, so that a position of equilibrium 

 was a place where the force due to the positive charge 

 vanishes. If there are more electrons than one, the position 

 of equilibrium will not be where the force due to the positive 

 charge vanishes, but where this force at any electron balances 

 the repulsion due to -the other electrons. This will displace 

 the position of equilibrium, and instead of these being 

 given by sin cu = 0, or cu = mr, they will be given by 

 cu-=7r(n + S), where 8 is a quantity depending on the repul- 

 sion of the electrons and perhaps also on n. 



As the frequencies of the vibrations are proportional to 

 a 2 —r 2 they will now be proportional to 



1 1 



(m + 8'y (n + 8) 2 ' 



Let us now consider the various types of series that could 

 arise on this view. 



