oblique to the beam. Experiments on the
fumes of chloride of ammonium gave me also reason to suspect that the
position of the neutral point was not constant, but that it varied
with the density of the illuminated fumes.

The examination of these questions led to the following new and
remarkable results: The laboratory being well filled with the fumes of
incense, and sufficient time being allowed for their uniform
diffusion, the electric beam was sent through the smoke. From the
track of the beam polarised light was discharged; but the direction of
maximum polarisation, instead of being perpendicular, now enclosed an
angle of only 12 deg. or 13 deg. with the axis of the beam.

A neutral point, with complementary effects at opposite sides of it,
was also exhibited by the beam. The angle enclosed by the axis of the
beam, and a line drawn from the neutral point to the observer's eye,
measured in the first instance 66 deg..

The windows of the laboratory were now opened for some minutes, a
portion of the incense-smoke being permitted to escape. On again
darkening the room and turning on the light, the line of vision to the
neutral point was found to enclose, with the axis of the beam, an
angle of 63 deg..

The windows were again opened for a few minutes, more of the smoke
being permitted to escape. Measured as before, the angle referred to
was found to be 54 deg..

This process was repeated three additional times the neutral point was
found to recede lower and lower down the beam, the angle between a
line drawn from the eye to the neutral point and the axis of the beam
falling successively from 54 deg. to 49 deg., 43 deg. and 33 deg..

The distances, roughly measured, of the neutral point from the lamp,
corresponding to the foregoing series of observations, were these:

1st observation 2 feet 2 inches.

2nd observation 2 feet 6 inches.

3rd observation 2 feet 10 inches.

4th observation 3 feet 2 inches.

5th observation 3 feet 7 inches.

6th observation 4 feet 6 inches.

At the end of this series of experiments the direction of maximum
polarisation had again become normal to the beam.

The laboratory was next filled with the fumes of gunpowder. In five
successive experiments, corresponding to five different densities of
the gunpowder-smoke, the angles enclosed between the line of vision to
the neutral point and the axis of the beam, were 63 degrees, 50 deg., 47 deg.,
42 deg., and 38 deg. r