326 M. L. Lorenz on the Determination of the Direction 
The experiments that have hitherto been tried leave this 
question still undecided; for while Stokes, from experiments 
made with glass gratings, found that the plane of polarization is 
rendered more horizontal, Holtzmann, from experiments made 
with a smoke-grating, came to the opposite conclusion. In order 
finally to settle this question, I have imstituted a course of expe- 
riments with gratings of various kinds. 
By means of a heliostat and collecting lens, I introduce a por- 
tion of the sun’s rays into a chamber. At some distance from 
the focus of the first lens a smaller lens receives the rays, and 
transmits them, almost parallel, through a polarizing Nicol’s 
prism, which is fastened in a tube to a vertical circular are. An 
index with a vernier gives the angle which the plane of polariza- 
tion of the transmitted light makes with the vertical lime. At 
the distance of about 7 metres the light falls on a vertical gra- 
ting, which is fastened to a small plate in the middle of a hori- 
zontal circular arc, which is provided with a moveable horizontal 
telescope. Before the object-glass is placed a doubly-refracting 
prism of rock-crystal, which divides the polarized pencil into 
two, polarized perpendicularly to each other, This prism can 
be turned about the axis of the telescope. In general, therefore, 
two horizontal bands of light of unequal brightness will be seen 
in the telescope ;, but by turning the Nicol’s prism or the doubly- 
refracting crystal about its axis, the intensity of these two 
images can be rendered equal. 
The experiments were generally conducted as follows :—The 
Nicol’s prism was turned in such a direction that the plane of 
polarization made an angle of 45° with the vertical line, and the 
telescope was so placed that its vertical thread passed through 
the two illuminating points, while the horizontal thread lay be- 
tween the two horizontal bands of diffracted hight. When the 
telescope was turned through the angle 8, both bands, owing to 
the position of the rock-crystal, were of equal intensity. The 
telescope was then put back into its original position, and the 
Nicol’s prism was turned into such a position that one of the 
images entirely disappeared. 
If the Nicol’s prism required to be turned through the angle 
5 (or 5+ 90° or 6+180°), the plane of polarization must have 
been turned through the same angle, provided that the light has 
not been elliptically polarized by diffraction ; and if 6 be positive, 
the plane of polarization has been rendered more horizontal. 
Sometimes I first determined 6, and then the angle of diffrac- 
tion 8, for which the two bands are equally bright. 
There is, however, a source of error in these experiments, of 
which I only became aware after some time. If, for example, 
the upper half of the grating produce a more brilliant diffracted 
