193 
nal white surface, then the lengths of the columns will fulfil 
the condition Q^=Q/. 
The carbon that I used was lamp-black recalcined in a 
covered platinum crucible. After being so treated it 
seemed blacker than it did before. Of this 0 4170 grams 
were ground up in a mortar with 10 drops of a solu- 
tion of gum. The contents of the mortar were then rinsed 
out with water, and diffused by shaking through 500 cub. c. 
of water. This was poured into a cylinder ; the length of 
the column was 22*5 c.m. After the lapse of 24 hours I 
drew off by a pipette the upper portion ; the column so 
removed was 15 ’8 c.m. long. This carbon diffusion contained 
so much solid matter that I found it inconveniently strong 
for experiments, so as occasion required I made weaker 
diffusions, containing in 250 cub. c. 5, 10, 20, or 40 of the 
stronger diffusion. Of these weaker liquids portions were 
taken and mixed with water so as to yield a bulk of 500 
cub. c. I stated above that the carbon was ground up 
with 10 drops of gum, so as to yield a smooth, thick con- 
sistence. The amount of dry gum would be very small, 
nevertheless it had a remarkable effect in increasing the 
adhesion of the carbon to the water. When I had shaken 
up carbon alone with water it had sensibly subsided after 
the lapse of a few hours. After the addition of so small a 
quantity of gum the tendency to deposit is much 
diminished. The following are the details of some experi- 
ments. The strength of the carbon diffusion is ghmn in 
terms of the number of cub. c. of the strong carbon diffu- 
sions in 500 cub. c. of water. The external white surface 
was a sheet of white paper. In all cases I have used the 
right eye only. The number under A denotes the mean of 
two trials got by pouring into the cylinder, and therefore 
likely to yield too low results. B denotes the mean of 
two trials got by pouring out of the cylinder, and there- 
fore likely to give too high results. C denotes the mean 
of A and B. D denotes tlie length required by theory. 
