138 Mr Herschel on a Peculiarity in the extraordinary 
considering the subject more carefully, however, it struck me 
that, as the curves in question had been traced to the immediate 
vicinity of the axis, and, when lost sight of, were then approach- 
ing it more rapidly than in any other part of their course, it was 
highly probable that they would meet it in that part of the spec- 
trum out of the reach of observation, and if so, must cut it, and 
the portions corresponding to the two extremities of the spec- 
trum would thus lie on opposite sides, the absolute lengths of 
the ordinates given by the experiment remaining unaltered. 
To put this idea to the test was a matter of considerable de- 
licacy, as well from the imperfections of the specimen, as from 
its feeble polarizing energy, and the great difference of its action 
on the different colours. In ordinary cases, to determine whe- 
ther a crystal be attractive or repulsive, nothing more is requi- 
red than to place a plate of it between crossed tourmalines, so as 
to view the polarized rings, and then crossing it with a plate of 
mica or sulphate of lime, having its principal section 45° inclined 
to the plane of primitive polarization, to notice in which qua- 
drants of the rings the tints are raised, and in which depressed. 
If the plate of the substance examined be then removed (with- 
out altering the position of the mica plate), and replaced by a 
plate of carbonate of lime, tourmaline, or any other known sub- 
stance, (which it is convenient to keep as a standard of compari- 
son), it is immediately seen whether the crystal in question be 
of the same, or an opposite character with the standard ; the 
corresponding quadrants of the rings seen in the two substances 
being similarly affected in the one case, and the alternate ones 
in the other. So coarse a method proved, as might be expected, 
unavailing in the present instance, the order of the tints being 
so completely altered by a plate of mica of moderate thickness, 
as to be no longer recognised, and I was obliged to have re- 
course to measures taken in homogeneous light, and on a divided 
apparatus. The method I pursued was as follows : — Having 
enclosed the crystallized plate of the second variety described in 
my former paper, in castor oil, in a proper apparatus for vary- 
ing its inclination to the polarized beam, and adjusted it so as 
to revolve in a plane 45° inclined to that of primitive polariza- 
tion, I noticed the inclinations at which the first minimum of 
