DR. JAMES BOTTOMLEY ON COLORIMETRY. 179 



Also with a less proportion of cobalt to nickel, namely 20 

 cub. c. of cobalt solution to 50 cub. c. of nickel solution, 

 I still obtained a tint in which yellow seemed to predomi- 

 nate. Had I employed solutions so dilute that no colour 

 was perceptible in the mixture, this would not strictly 

 imply that the colours were complementary, but that the 

 resulting tint was too feeble to produce the impression of 

 colour ; and if we filled a long tube with such a dilute so- 

 lution the colour might again become manifest. More- 

 over my aim was not to mix two coloured solutions so as 

 to obtain a fluid which exercised no perceptible absorption 

 of light, but to obtain a fluid which would exercise a con- 

 siderable absorption subject to a certain condition. The 

 following consideration seemed to me to render it hopeless 

 to obtain a soluble black by nickel and cobalt only. A 

 solution of cobalt when dilute is pink; but if we look through 

 a considerable thickness or through a concentrated solution, 

 the pink shows a tendency to pass into a scarlet. This 

 shows that as the quantity of the salt increases, the ratio 

 of the yellow to the red increases. The colour of the 

 undissolved salt is brownish red; and the colour of the 

 solution seems to approximate towards this as the concen- 

 tration increases. Hence the colour of a solution of cobalt 

 alters not only in intensity, but also in kind, as the amount 

 of the salt is increased. On the other hand, the green of 

 a solution of nickel varies in intensity, but does not seem 

 to vary in character, at least in any marked manner, as the 

 quantity of the salt increases. In order that it should be 

 generally complementary in character to cobalt, any in- 

 constancy in the ratio of the red to the yellow of the latter 

 would require a corresponding variation in the ratio of 

 the yellow to the blue in the former, and the tint ought 

 to pass from an emerald-green to a bluish green. As this 

 does not seem to be the case, it would follow that even if 



n2 



