Mechanism of Radiation. 425 



§ 7. We are therefore driven to the conclusion that the 

 vibrators by which radiation possessing a line-spectrum is 

 emitted, must behave like the molecules of a monatomic gas. 

 In other words, we are led to suppose that a line -spectrum is 

 emitted only by the dissociated atoms of a gas and not by 

 the complete molecules, and, further, that these dissociated 

 atoms must be spherically symmetrical. 



§ 8. On this view, we can see why it is that pure line- 

 spectra are emitted only by elements, and not by compounds. 

 We can also understand why a gas raised to incandescence 

 by mere heating emits a continuous and not a line-spectrum : 

 the vibrations must be those of complete molecules, not of 

 dissociated atoms. The same substance placed in a flame 

 emits its line-spectrum at a temperature lower than that at 

 which it first becomes luminous if merely heated. Hence the 

 dissociation and radiation which occur when a substance is 

 placed in a flame, cannot be merely the result of the tempera- 

 ture of the flame : they must be produced by chemical action, 

 a view originally put forward by Pringsheim*. 



The faint continuous spectrum which is often seen to 

 accompany a line-spectrum must be supposed to proceed 

 from undissociated molecules. 



We have incidentally found a factor tending to increase 

 the width of the lines, namely, the rotation of the atoms by 

 which these lines tire emitted. We must, however, consider 

 this question at greater length in a later section (§ 34). 



The Electrical Structure of an Atom. 



§ 9. Let us now leave this train of thought altogether, and 

 regard an atom as a collection of negative and positive ions, 

 the negative ions each carrying a charge of electricity of 

 amount — e, and the positive ions each carrying a charge +<?. 



The conception of an atom which we are tempted to adopt 

 at first sight, is that of a collection of point charges of positive 

 and negative electricity, these charges repelling or attracting 

 according to the law of the inverse square of the distance. 

 These point charges would be supposed to settle into a position 

 of stable equilibrium, and the oscillations about this position 

 might be supposed to be the vibrations of definite period 

 which result in the emission of a line-spectrum. 



Against this very simple conception of an atom there is 

 one fatal objection : Earnshaw's theorem shows that no position 



* Wied. Ann. xlviii. p. 428 ; xlix. p. 347. On the luminosity produced 

 by heating, see J. Evershed, Phil. Mag. xxxix. p. 460; A. Smithells, 

 Phil. Mag. xxxvii. p. 245, and xxxix. p. 122 ; A. Paschen, AVied. Ann. 

 1. p. 409/' 



