144 Royal Society :— 
compartment of the instrument, and is reflected at a surface of 
blackened glass, so as to be seen by the observer in immediate contact 
with the compound light which enters the slits and is refracted by 
the prisms. 
Each experiment consists in altering the breadth of the slits till 
the two lights seen by the observer agree both in colour and brightness, 
the eye being allowed time to rest before making any final decision. 
In this way the relative places of sixteen kinds of light were found by 
two observers. Both agree in finding the positions of the colours to 
lie very close to two sides of a triangle, the extreme colours of the 
spectrum forming doubtful fragments of the third side. They differ, 
however, in the intensity with which certain colours affect them, 
especially the greenish blue near the line F, which to one observer is 
remarkably feeble, both when seen singly, and when part of a mixture; 
while to the other, though less intense than the colours in the 
neighbourhood, it is still sufficiently powerful to act its part in com- 
binations. One result of this is, that a combination of this colour 
with red may be made, which appears red to the first observer and 
green to the second, though both have normal eyes as far as ordinary 
colours are concerned; and this blindness of the first has reference 
only to rays of a definite refrangibility, other rays near them, though 
similar in colour, not being deficient in intensity. Foran account of 
this peculiarity of the author’s eye, see the Report of the British 
Association for 1856, p. 12. 
By the operator attending to the proper illumination of the paper 
by the sun, and the observer taking care of his eyes, and completing 
an observation only when they are fresh, very good results can be 
obtained. The compound colour is then seen in contact with the 
white reflected light, and is not distinguishable from it, either in hue 
or brilliancy ; and the average difference of the observed breadth of a 
slit from the mean of the observations does not exceed =), of the 
breadth of the slit if the observer is careful. It is found, however, 
that the errors in the value of the sum of the three slits are greater 
than they would have been by theory, if the errors of each were 
independent ; and if the sums and differences of the breadth of two 
slits be taken, the errors of the sums are always found greater than 
those of the differences. This indicates that the human eye has a 
more accurate perception of differences of hue than of differences of 
illumination. 
Having ascertained the chromatic relations between sixteen colours 
selected from the spectrum, the next step is to ascertain the positions 
of these colours with reference to Fraunhofer’s lines. This is done 
by admitting light into the shorter arm of the instrument through 
the slit which forms the eyehole in the former experiments. A pure 
spectrum is then seen at the other end, and the position of the fixed 
lines read off on the graduated scale. In order to determine the wave- 
lengths of each kind of light, the incident light was first reflected 
from a stratum of air too thick to exhibit the colours of Newton’s 
rings. The spectrum then exhibited a series of dark bands, at 
intervals increasing from the red to the violet. The wave-lengths 
