Refr activity and Atomic Interaction. 97 



there are crystalline compounds, having two or even three 

 different principal refractivities, and (although the regular 

 arrangement o£ molecules is an indispensable factor) the 

 optical anisotropy of a crystal is certainly to be attributed 

 in its essence to the anisotropy of the molecule. Now, 

 the constituent atoms, or rather their refractivities, as the 

 C or 0' or even f= of the above Table, being essentially 

 isotropic, an additive combination of these is utterly im- 

 potent to give an optically anisotropic molecule. This seems 

 to me a grave objection, notwithstanding that the anisotropy 

 of N is, in general, rather small. But I perceive a more 

 serious objection to the additive law in the following cir- 

 cumstance. In many cases the compound has an entirely 

 different distribution of absorption bands from that of the 

 constituents (elements or not), i. e. new free frequencies are 

 produced and the old ones are shifted by combination or 

 dissociation *. The resultant dispersion- and absorption- 

 curves are not obtainable by a superposition of the component 

 ones. In plain language, colour is produced, or destroyed. 

 by combination or by dissociation of substances which may 

 themselves be colourless, or nearly so, to the human eye or 

 to the spectrograph. It is hard to understand why this 

 circumstance has been entirely kept out of account in the 

 chemical studies of molecular refractivity. It is true that 

 there is a great chapter in chemistry entitled " Colour and 

 Constitution," but, as far as I can gather, this is treated 

 without any connexion with " Molecular and Atomic Re- 

 fraction and Dispersion." The existing so-called "theories 

 of colour," due to Witt, Armstrong, Kauffmann, and Baly, 

 have nothing in common with the rich material collected 

 by the refractivitists, the late J. W. Briihl and his followers f . 

 Enough has now been said to justify an attempt to amplify 

 Lorentz's additive law by taking explicitly into account the 

 electro-optical $ interaction of the atoms constituting a 

 molecule. This will occupy our attention in the following 

 sections. The need of a similar investigation has, of course, 

 been felt by physicists, and Lorentz himself says in con- 

 nexion with Bruin" s work (loc. cit. p. 150) : — " There is at 



* And this is closely connected with all the stated facts of huge 

 exaltation of molecular dispersion. 



t An account of " Colour and Constitution " is given in Enc. Brit. 

 11th ed.vol. vi. (1910) pp. 70-72. See also Smiles's < Chemical 

 Constitution and Physical Properties/ Longmans (1910), Chaps. X., 

 XI. 



\ I. e. so far as the dispersive electrified particles (electrons, say), and 

 these only, are concerned. 



Phil Mag. S. 6. Vol. 33. No. 193. Jan. 1917. H 



