bliquely at the beam,
observed the colours fading until they disappeared. Augmenting the
obliquity the colours appeared once more, but they were now
complementary to the former ones.

Hence this beam, like the sky, exhibited a neutral point, on opposite
sides of which the light was polarised in planes at right angles to
each other.

Thinking that the action observed in the laboratory might be caused,
in some way, by the vaporous fumes diffused in its air, I had the
light removed to a room at the top of the Royal Institution. The
track of the beam was seen very finely in the air of this room, a
length of 14 or 15 feet being attainable. This beam exhibited all the
effects observed with the beam in the laboratory. Even the
uncondensed electric light falling on the floating matter showed,
though faintly, the effects of polarisation.

When the air was so sifted as to entirely remove the visible floating
matter, it no longer exerted any sensible action upon the light, but
behaved like a vacuum. The light is scattered and polarised by
_particles_, not by molecules or atoms.

By operating upon the fumes of chloride of ammonium, the smoke of
brown paper, and tobacco-smoke, I had varied and confirmed in many
ways those experiments on neutral points, when my attention was drawn
by Sir Charles Wheatstone to an important observation communicated to
the Paris Academy in 1860 by Professor Govi, of Turin. [Footnote:
Comptes Rendus,' tome li, pp. 360 and 669.] M. Govi had been led to
examine a beam of light sent through a room in which were successively
diffused the smoke of incense, and tobacco-smoke. His first brief
communication stated the fact of polarisation by such smoke; but in
his second communication he announced the discovery of a neutral point
in the beam, at the opposite sides of which the light was polarised in
planes at right angles to each other.

But unlike my observations on the laboratory air, and unlike the
action of the sky, the direction of maximum polarisation in M. Govi's
experiments enclosed a very small angle with the axis of the
illuminating beam. The question was left in this condition, and I am
not aware that M. Govi or any other investigator has pursued it
further.

I had noticed, as before stated, that as the clouds formed in the
experimental tube became denser, the polarisation of the light
discharged at right angles to the beam became weaker, the direction of
maximum polarisation becoming