DR. JAMES BOTTOMLEY ON COLORIMETRY. 71 



cases where the relation 5'/= constant fails, it may be 

 traced to some decomposition having taken place, or to 

 some change effected by light. For example, I commenced 

 some experiments with ferricyanide of potassium; but as it 

 did not prove a suitable salt for making experiments 

 without some special precautions with regard to the action 

 of light, I discontinued them. As I am not aware that 

 any one has particularly noticed this darkening, a few re- , 

 marks may be interesting. A standard solution was pre- 

 pared containing 0"8 gram in 500 cub. c. In the afternoon 

 having occasion to use this standard for comparison with 

 another, the result was not satisfactory, owing to its trans- 

 parency not being so perfect as when freshly made. On 

 the following morning I made a fresh standard solution 

 of the same composition ; it differed from the old in being 

 more transparent, and I thought that it had more of a 

 greenish tint. This new solution being left on the table 

 before the window, after a time became of diminished 

 transparency ; also on looking down into the cylinder a 

 very faint red cloudiness was perceptible. I also compared 

 a solution containing 3*2 grams in 500 cub. c. which had 

 been freshly prepared with a solution containing 6*4 grams 

 in 500 cub. c; this solution had been prepared on the 

 previous day and had been exposed to light during that 

 interval. I found the length of the column indicated by 

 theory decidedly too great ; it occurred to me that the 

 discrepancy was due to some action of light on the ferri- 

 cyanide. About six o'clock in the afternoon I again com- 

 pared these solutions ; the theoretical length gave a colour 

 which w as still too dark, but the disparity of colour was 

 not so marked as at first. The comparison was also dis- 

 turbed a little by the slightly diminished transparency of 

 the weaker solution. 



I now prepared a fresh solution, containing 6 '4 grams 



