88 Proceedings of the Royal Society of Edinburgh. [Sess. 
imaginary green, but the red was a real colour corresponding closely to the 
red corner of my triangle. So the diagram of a completely red-blind man 
should have no full contour lines. All my diagrams have full contour 
lines. The fundamental characteristic of the system of Helmholtz and 
Maxwell was that the colour sensations of the normal could be represented 
on a plane, whereas the sensations of the colour blind required only a line 
or a point. All my cases require plane diagrams. Thus it follows that the 
colour blind are trichromatics. 
This result I was prepared for, as I have frequently inferred it on 
previous occasions from other experiments, though I was never able to 
demonstrate it directly before. But the two cases (6) and (7), whom I have 
named the colour different, came as a great surprise to me. 1 had no idea 
that trained observers with so high a power of discriminating colour could 
make such serious errors with the colours of everyday objects. The two 
converse cases (9) and (10), which I have arranged directly below (6) and 
(7) for purposes of comparison, are also striking: it is remarkable that 
observers with such a low power of discriminating colour should make so 
few mistakes. It is difference in the distribution of colour-discriminating 
power, not lack of it, that causes trouble ; also, difference in the relative 
intensity of the colours has a strong influence. 
{Issued separately May 1 , 1922 .) 
