141 
value got with discs inside is not so good as with discs 
outside. Also the theoretical depth 14*1 cm. when tried 
seemed to be slightly too great. From the foregoing experi- 
ments it would appear that with considerable additions of 
ammonia the results with discs outside were much improved. 
Why the results were not equally good with discs inside 
may, I think, be accounted for, and will be considered 
further on. I also compared the solution containing 4000 
with the one containing 1600. Standard solution 1600 in 
500 cub. c., length of column 21 2. 
ABC 
4000 8 4240 
The number under B was the mean of eight trials. The 
theoretical length was 8 ’5 ; with the standard solution on 
the left hand this length seemed to give a similar tint, but 
with standard solution on right it seemed a little darker ; 
and now I remarked for the first time, or, if I had previously 
remarked it, had not thought it worthy of notice, that even 
an apparently so trivial circumstance as altering the posi- 
tions of the cylinders from right to left had a perceptible 
influence in the determination of colour. I afterwards 
made some experiments to test this. Next I compared 
solutions containing 1600 and 6000. Standard solution 
1600 in 500 cub. c., length of column 21*2. 
ABC 
6000 5T 6682 
The number under B was the mean of eight trials. The 
result is not so satisfactory as the others. I also tried these 
solutions with white surfaces inside. Standard solution 
same as last : 
A B C 
6000 3-9 8697 
A widely remote result, and much further from the rea] 
value than with discs outside, also a disc placed in the solu- 
tion at the depth assigned by calculation seemed too dark. 
It seemed possible to me that a more satisfactory result 
