56 Proceedings of the Eoyal Society of Edinburgh. [Sess. 
lines in fig. 1 are such curves for solutions I, la, lb, I c, Id. It is seen from 
them that in the pure alcoholic solution I there is a very pronounced 
hand with its maximum at X = 675 /ul/ul. As water is added the height of the 
band decreases in approximately the same ratio all over, and there is a slight 
displacement of the band as a whole toward the red. This is shown by the 
position of the dotted curves, which are curve I reduced in various ratios. 
Finally, in the case of le, in which there is about 28 per cent, of water, the 
maximum of the band has been reduced to less than 4 ; and solutions Ig and 
111 show no trace of the band, the spectrum being that of the dilute aqueous 
solution. Results of a similar nature were obtained with the sets of curves 
derived from the alcoholic solutions II and III. 
It is obvious therefore that in these solutions the cobalt chloride is 
in two different states, one of which is characterised by this band in the 
red. This state I shall call the blue phase. As water is added the salt 
passes into that state in which it exists in dilute aqueous solution. This 
second state I shall call the red phase. It is characterised by a band in the 
green at A. = 510 fifi, but as the maximum value of A for this band in any 
aqueous solution not approaching saturation is only about 6, and the 
maximum for the band characteristic of the blue phase is about 232, the 
band in the green is obscured by the side of the other band if more than 
one-tenth of the salt is in the blue phase. 
In order to find out how the distribution of the salt between the two 
phases varied with the quantity of water added, the maximum value of A 
for the band in the red was plotted as a function of the proportion of water. 
The results are shown by the curves in fig. 2, each solution being represented 
by one point. The concentration of the water in the solution in grm.-mols. 
per litre is denoted as before by c 2 . Results are not shown for higher 
concentrations than c 2 = 21, as beyond this point practically all the chloride 
is in the red phase. 
The maximum values of A for the three pure alcoholic solutions do not 
agree. This is probably due to the presence of water in the absolute 
alcohol employed in making up the solution, as absolute alcohol generally 
has from | to 1 per cent of water. The curves joining the points for each 
series of solutions I, la, .... II, Ila, .... Ill, Ilia, .... become 
approximately straight lines when the quantity of water present in solu- 
tion is small. These lines meet at a point for which A = 232 and c— — '6. 
It is very probable that for pure nonaqueous solutions A varies only 
slightly with the concentration, there being nothing to indicate that the 
anhydrous chloride enters into any combination with the alcohol at room 
temperature. Hence from the curves it was assumed that the maximum 
