1912-13.] The Absorption of Light by Inorganic Salts, No. XI. 163 
where the bands are sharp and the wave-lengths can be determined 
accurately. In other cases a spectrophotometer must be employed, but 
the spectrophotometry of solutions of the necessary dilution is very 
troublesome. The colour changes of the cobalt and copper salts have 
often been ascribed to ionic dissociation ; but, as has been decidedly shown 
in the tenth article, these changes occur only at great concentrations, and 
must be put down to hydration. 
There is one interesting point in connection with the colour of the 
ions which has not, I think, yet received attention. To explain an 
absorption band we must have vibrating charged particles. These particles 
must have centres to vibrate about. If, for example, we take the case 
of CoS0 4 , we can suppose that the Co has a positive charge, the S0 4 a 
negative charge, and that they vibrate towards each other under the 
influence of an electrical attraction. When the CoS0 4 is ionised, these 
vibrations cease. Consequently, the Co ion cannot have a spectrum with 
absorption bands unless we introduce vibrations inside the atom or provide 
it with a watery atmosphere. 
In 1877 Kundt stated the law that if one colourless solvent had a 
greater dispersive and refractive power than another, and if a colouring 
matter producing an absorption band was dissolved in both, then in the 
case of the first medium the absorption band would be nearer the red 
than in the case of the second. Kundt’s own measurements showed 
•exceptions to the rule, and it is as often disobeyed as obeyed ; yet it has 
persisted in the literature and has become a kind of definition of chemical 
change. When it is not obeyed, the solvent is supposed to enter into 
chemical combination with the colouring matter, and when it is obeyed 
there is no chemical combination. Kayser * thinks that, on the whole, 
there is something in the law, but that it is very often masked by 
disturbing factors. 
My own belief is, that when the disturbing factors are taken away 
there is nothing left. The idea of molecules of the dissolved substance 
moving about in the solvent without essential change is a very tempting 
one, and suggests analogies with Hertzian vibrators in media of different 
dielectric constants. The period of the Hertzian vibrator should, of 
course, be proportional to the square root of the dielectric constant of 
the medium. Also, we can think of the electric doublets associated with 
the absorption bands as being affected by weak electric forces from the 
doublets in the ultra-violet |which cause the dispersion and refraction 
of the solvent, and I developed a theory on these lines seven years 
* Spectroscopie, vol. iii. p. 85. 
