160 Proceedings of the Royal Society of Edinburgh. [Sess. 
crystallisation, striking colour changes were passed through. I therefore 
came to the conclusion that the three bands common to the six nickel 
salts were due not to the nickel atom but to its watery atmosphere, to the 
bonds connecting it to the water molecules round about it. Similarly the 
two bands common to the six cobalt salts were characteristic, not of the 
cobalt atom, but of the bonds connecting it to its watery atmosphere. My 
view is that when any of these twelve salts is dissolved in water, one 
watery atmosphere forms round the base and another round the acid 
radical. The bonds connecting the acid radical and base to one another 
are much weaker than those connecting each to its own watery atmosphere,, 
and do not affect the spectrum much, even at great concentrations. Only 
when there is not sufficient water left for the base to form its proper watery 
atmosphere is the absorption spectrum much affected. 
These watery atmospheres I believe to be the same as surround the 
ions in electrolysis. My view thus explains why the colour of cobalt,, 
nickel, copper salts, permanganates, etc., is not altered when they are 
ionised. The colour is characteristic mainly of the system ion -{-watery: 
atmosphere , and in the neutral molecule the anion and kation systems are 
but slightly bound together. 
The application of the formula for pe/m to the cobalt and nickel salts 
(fourth article) did not give any definite result. It was hopeless to try 
and construct a dynamical system representing a watery atmosphere. 
Having failed to make anything mathematically of the cobalt and nickel 
salts, I turned my attention to the salts of the alkali metals (fifth article)- 
As the alkali metals are monovalent, conditions might have been simpler 
there. But in their case the absorption was almost all off the range of 
the apparatus. Also in the case of the iron and copper salts it proved im- 
possible to connect their chemical constitution with their absorption spectra,, 
owing, in this case also I think, to the occurrence of watery atmospheres. 
Conditions in aqueous solutions having proved so complicated, I turned 
my attention to other solvents. The eighth and ninth articles give some- 
results for ethyl alcohol and acetone. The absorption in these solvents 
is more intense and the additive character of the results breaks down; 
they become more characteristic and irregular. Nothing could be done 
with them from the standpoint of the theory of dispersion, although the 
law of mass action was applied successfully to the colour change which 
occurs when water is added to an alcoholic solution of cobalt chloride or 
bromide. 
As a result of my own experimental work I have not been able to 
construct any dynamical models giving absorption spectra. The only case 
