THE ZEEMAN EFFECT AND DISPERSION. 421 



the matter has on the ether. Hence there are two ways of 

 altering- the rate of wave propagation, one by altering- the 

 amount of connection between the ether and matter, and 

 the other by altering the frequency of vibration of the 

 matter. For example, by increasing the density of a gas 

 we must reduce the velocity of propagation of light through 

 it both because there is more matter acting on the ether 

 and because, as has been shown, the rate of vibration of 

 the molecules is reduced and thus the frequencies of vibra- 

 tion of the ultra-violet vibrations that control dispersion are 

 brought into closer coincidence with the visible radiations 

 and this makes the rate of propagation of these latter 

 slower. This latter effect has not been observed owing to 

 its extreme minuteness. There may also be other effects 

 on the waves due to other changes in the vibrations pro- 

 duced by the increased density. For example, it has been 

 shown that there are certain absorption bands in oxygen 

 whose intensity depends on the square of the density of 

 the gas. They are consequently, in all probability, due to 

 vibrations which are only executed during the collisions of 

 the molecules, and if the vibrations that control dispersion 

 are in any sensible degree of this character, there may be a 

 corresponding reduction of velocity of propagation of light 

 through oxygen depending on the square of its density. 



One of the best known methods of observing these 

 changes in velocity of propagation due to changes of 

 structure is by observing with polarised light the crystalline 

 structure produced by straining transparent solids. In this 

 case, as in that of crystalline structure in general, the rates 

 of propagation of vibrations executed in different directions 

 are different. This may be due either to the interconnection 

 of the matter and ether in different directions differing, or 

 to the frequencies of vibration of the molecules in different 

 directions differing, or to both these causes. On account 

 of the want of sharpness of the absorption bands in solids 

 it would not be easy to observe the probably very small 

 change in their position due to straining a solid, so that it 

 does not seem probable that we can in the near future dis- 

 tinguish between the two causes in the case of solids as we 



