442 SCIENCE PROGRESS 



emission of light is due. The theory was in agreement with 

 observation in so far as it required a displacement of the bright 

 emission lines towards the region of longer wave length to 

 follow the increased damping consequent upon the closer 

 packing of the molecules ; but for the absorption lines, it 

 required a widening unaccompanied by any displacement, 

 whereas experiment shows that both emission and absorption 

 lines suffer the same displacement. A modification of the 

 theory was attempted by Wilsing, 1 but was unsatisfactory, and 

 we must look to the electric rather than to the mechanical 

 properties of the medium to find an explanation. 



Such an explanation was offered by Fitzgerald, 2 who sup- 

 posed that when the pressure was increased the luminous 

 vibrations were slowed down owing to the increased specific 

 inductive capacity of the medium in which the vibrations take 

 place. In fact, if we imagine the vibrating systems as small 

 Hertzian oscillators, vibrating in a medium of specific inductive 

 capacity K, the frequency, N, of the vibrations emitted is such 

 that N~ 2 varies as K. Thus when K is large N is small. Now 

 an increase of pressure causes an increase in the specific induc- 

 tive capacity of a gas, and so it follows that there is a vera causa 

 for some shift towards the longer wave lengths of the emitted 

 vibrations. The same argument has been in a more general 

 form expressed by Sir Joseph Larmor 3 thus : " Each molecule 

 individually, through the agency of its plastic field of force or 

 aether strain, provides a yielding region in the aether in which 

 the effective stiffness is diminished. The elastic energy which 

 maintains the free vibrations of the radiator is located in the 

 field of force in the adjacent aether; and, by dynamical principles, 

 any loosening of the constraints in that field such as an adjacent 

 molecule would produce, which would itself be somewhat inten- 

 sified by equality of period, must in general tend towards 

 increasing the free period, involving displacement of the radia- 

 tion towards longer wave length." 



Humphreys, 4 using this theory, obtained a pressure shift 

 about three hundred times larger than the observed value. The 

 calculation was, however, implicitly based upon the assumption 

 that the surrounding medium was continuous right up to the 

 vibrator in question. This is not permissible ; since we are 



1 Astroph. Journ. 7, p. 317, 1898. * Ibid. 25, p. 120, 1907. 



' Ibid. 5, p. 210, 1,897. 4 Ibid. 26, p. 30, 1907. 



