1888.] Mr G. N. Stewart on Intermittent Light. 451 
Take now another shorter time of stimulation, 0^5^. The colour 
of the resultant impression will be determined by the proportion 
^ut : A2B2 : A3B3 : : ap-^ : ap^ • <^3^3 5 • '• 
the colour will be the same as it was before with time O^B^. As 
the curves are drawn, then, in fig. 2, we cannot have a change in 
the colour with change in the time of stimulation. 
Suppose now that Oj , O2 , O3 do not lie on the same vertical line, 
i.e., that stimuli may be used of sufficient shortness to excite only 
one or two of the three groups (fig. 3). 
Then we have ^1^1 = 
A2B2 = O2B2 tan fS, 
~ ^3^3 y? 
where a, fB, y are the angles which the curves make with the 
abscissa axis. (The curves are still supposed to be straight lines.) 
Similarly ap^ = tan a 
ap2 = ^p2 P 
ap^ = O3&3 tan y 
.'. Aj^Bj^: A2B2: A3B3: : ap^: ap 2 '- ^Pz • • • (1) 
if and only if O^Bj, tana : O2B2 tan /5 : O3B3 tan y : : tan a: 
0p2 tan p : O353 tan y, (2) 
All the terms in the first member of (2) are constants. 
In the second member tan a, tan tan y are constants, and the 
proportion (2) cannot hold unless : Op ^ : O3&3 for any position 
of ^2? ^3 be constant. This is evidently not the case, and .*. 
changes of colour may occur with changes in the time of stimulation. 
M^e have no warrant, however, for supposing that, within our 
