146 
through tints which we cannot distinguish from it, we are 
likely to take a column of fluid too short. In my own 
case I have noticed this on several occasions ; for example, 
in trying to estimates particular colour the mean of seven 
trials made by pouring out from the cylinder gave 141 as 
the length of the column, the mean of seven trials made by 
pouring into the cylinder gave 13*7 as the length, and with 
discs inside the mean of six trials made by moving the disc 
from below upwards gave 12 5 as the length of column, 
while moving from above downwards the mean of six ex- 
periments gave 11*7. In trying to estimate a colour it 
seems to me that it would be well to approach the limit by 
both ways, and then take the mean of the results. 
In a previous part of this paper I stated that altering 
the position of the cylinders made a little difference in my 
perception of colour. I made some experiments to try this. 
I took a solution containing 2400 and poured from the 
cylinder on the right hand into the cylinder on the left hand- 
the columns ought to be equal. The mean of nine trials 
o-ave length of column on right hand 112 ; length of column 
on left hand 10-61. In these experiments, the judgment was 
made using both eyes. I next tried using one eye only — 
with the right eye the results were, length of column on 
right hand 1 0*86, length of column on left hand 1109; these 
numbers were the mean of nine trials. With the left eye 
alone, the results were, right hand cylinder 10-78, left hand 
1101, being the mean of nine trials; thus using one eye 
only, the results are nearly the same in both cases ; they 
also tend to make the right hand a little less, thus reversing 
the case of two eyes. These experiments were made in a 
room with a small window facing the south. I afterwards 
repeated the experiments, using two eyes, in another room 
having a window of larger dimensions and facing the north. 
A solution was used containing 1600, the mean of nine 
trials gave right hand cylinder 11*16, left hand 1055, nearly 
