e rays (scientifically called the angle of incidence),
the green of the leaf and the brown of the tree are modified. What has
to be studied therefore in these objects, if one wishes to recall their
colour to the beholder of a picture, is the composition of the
atmosphere which separates them from the eye. This atmosphere is the
real subject of the picture, and whatever is represented upon it only
exists through its medium.

[Illustration: DEGAS

THE GREEK DANCE--PASTEL.]

A second consequence of this analysis of light is, that shadow is not
absence of light, but light of _a different quality_ and of different
value. Shadow is not a part of the landscape, where light ceases, but
where it is subordinated to a light which appears to us more intense. In
the shadow the rays of the spectrum vibrate with different speed.
Painting should therefore try to discover here, as in the light parts,
the play of the atoms of solar light, instead of representing shadows
with ready-made tones composed of bitumen and black.

The third conclusion resulting from this: the colours in the shadow are
modified by _refraction_. That means, _f.i._ in a picture representing
an interior, the source of light (window) may not be indicated: the
light circling round the picture will then be composed of the
_reflections_ of rays whose source is invisible, and all the objects,
acting as mirrors for these reflections, will consequently influence
each other. Their colours will affect each other, even if the surfaces
be dull. A red vase placed upon a blue carpet will lead to a very
subtle, but mathematically exact, interchange between this blue and this
red, and this exchange of luminous waves will create between the two
colours a tone of reflections composed of both. These composite
reflections will form a scale of tones complementary of the two
principal colours. The science of optics can work out these
complementary colours with mathematical exactness. If _f.i._ a head
receives the orange rays of daylight from one side and the bluish light
of an interior from the other, green reflections will necessarily appear
on the nose and in the middle region of the face. The painter Besnard,
who has specially devoted himself to this minute study of complementary
colours, has given us some famous examples of it.

The last consequence of these propositions is that the blending of the
spectral tones is accomplished by a _parallel_ and _distinct_ projection
of the