pon the screen. Altering in no
particular the wedge-shaped vessel, but simply substituting for the
water the transparent bisulphide of carbon, you notice how much higher
the beam is thrown, and how much richer is the display of colour. To
augment the size of our spectrum we here employ (at L) a slit, instead
of a circular aperture.[6]

[Illustration: Fig. 9.]

The synthesis of white light may be effected in three ways, all of
which are worthy of attention: Here, in the first instance, we have a
rich spectrum produced by the decomposition of the beam (from L, fig.
9). One face of the prism (P) is protected by a diaphragm (not shown
in the figure), with a longitudinal slit, through which the beam
passes into the prism. It emerges decomposed at the other side. I
permit the colours to pass through a cylindrical lens (C), which so
squeezes them together as to produce upon the screen a sharply defined
rectangular image of the longitudinal slit. In that image the colours
are reblended, and it is perfectly white. Between the prism and the
cylindrical lens may be seen the colours, tracking themselves through
the dust of the room. Cutting off the more refrangible fringe by a
card, the rectangle is seen red: cutting off the less refrangible
fringe, the rectangle is seen blue. By means of a thin glass prism
(W), I deflect one portion of the colours, and leave the residual
portion. On the screen are now two coloured rectangles produced in
this way. These are _complementary_ colours--colours which, by their
union, produce white. Note, that by judicious management, one of these
colours is rendered _yellow_, and the other _blue_. I withdraw the
thin prism; yellow and blue immediately commingle, and we have _white_
as the result of their union. On our way, then, we remove the fallacy,
first exposed by Wuensch, and afterwards independently by Helmholtz,
that the mixture of blue and yellow lights produces green.

Restoring the circular aperture, we obtain once more a spectrum like
that of Newton. By means of a lens, we can gather up these colours,
and build them together, not to an image of the aperture, but to an
image of the carbon-points themselves.

Finally, by means of a rotating disk, on which are spread in sectors
the colours of the spectrum, we blend together the prismatic colours
in the eye itself, and thus produce the impression of whiteness.

Having unravelled the interwoven constituents of white light, we have
next to in