ON OUR KNOWLEDGE OF SPECTRUM ANALYSIS. 137 



be of the second order of magnitude. It is needless to enter farther into 

 the matter, but it is easily seen how fictitious the whole investigation of 

 Dr. Claes now becomes. 



Returning for a moment to the general law discovered by Kundt, that 

 the displacement of bands depends to a greater extent on the dispersive 

 power of the solvent than on any other of its properties, we may, perhaps, 

 suggest the way in which such a result might be brought about. The 

 refractive index of a liquid does not seem to be directly connected with 

 the vibrations of its molecules, for we speak of the refractive index of 

 waves which are infinitely long, and connect it with the inductive capacity 

 which has nothing to do with vibrations. It is otherwise with the dis- 

 persion ; we know that the very colouring matters which are most sensi- 

 tive as far as the displacement of bands is concerned, show the so-called 

 anomalous dispersion, which only means that the nearness of an absorp- 

 tion band causes an abnormally great dispersion at that point. If we are 

 allowed to reason backwards and connect a great dispersive power at one 

 place with a great coefficient of absorption at some point which is not too 

 far removed, we might easily understand how these vibrations would in 

 their turn affect the vibrations of the colouring matter and change their 

 period. For it is a perfectly general rule that any spectroscopic dis- 

 turbance, such as the widening of bands or appearance of high tempera- 

 ture lines, is more easily produced by molecules which vibrate in similar 

 jieriods than by others,' and a solvent the molecules of which can vibrate 

 in similar periods to that of the colouring matter will, no doubt, produce 

 a displacement of bands towards the less refrangible regions. It is true 

 that the high dispersion of some liquids — such as carbon bisulphide — 

 has not been traced to the influence of any specific absorption, but even 

 if, as is very likely, the above explanation is not strictly correct, we seem 

 to have in the vibrations of the solvent a connecting link between the dis- 

 persion of a liquid and the displacement of bands which it is capable of 

 producing. On the other hand it is easy to see why no perfectly general 

 rule can be given ; for the influence of the solvent will not only depend 

 on the vibration of its molecules but also on the closeness of the connec- 

 tion between them and the molecules of the colouring matter. If there 

 is any great affinity between the solvent and the colouring matter we 

 should expect a great influence, and if the two bodies simply mix without 

 troubling much about each other we should have no displacement at all, 

 or only a very small one. The displacement of bands according to this 

 view is due then, in the first instance, to the closeness of the chemical 

 relation between solvent and colouring matter ; and, secondly, to the 

 similarity of their vibrations. If the dispersive power of a liquid enters 

 into the question it can only be owing to the fact that the vibrations of 

 the Inminiferous medium and those of the colouring matter are similarly 

 affected by the periods in which the molecules of the solvent are capable 

 of vibrating. 



It is proved by some experiments made by Professor RusselP that 

 the shifting of bands can also be produced by solution in a solid body. 

 Professor Russell has proved that when chloride of cobalt is fused together 

 with potassic chloride it gives a certain absorption spectrum, which may 

 be obtained with small displacements of bands when the potassic chloride 



' Cf. a passage in a lecture by the author of this report before the Koyal 

 Institution. 



' Proc. Roy. Soc. xxxii. p. 258 (1881). 



