MR. W. CROOKES OK REPULSION RESULTING FROM RADIATION. 
277 
The difference in this case is on the other side. It would now seem that there was 
more repulsion when the polarised ray passed through the crystal. There was, how- 
ever, great difficulty in bringing the beam and index ray of light back to zero ; the 
platinum grating was successful in preventing the tourmaline from flying to one or 
other side of the glass tube and there sticking, but there seemed to be a residual action 
which kept the suspending fibre in a state of torsion and interfered with the free 
movement of the crystal. 
Another apparatus was therefore made, introducing every refinement which expe- 
rience suggested, and employing a still finer torsion fibre. A minute portion of a 
magnetised sewing needle was attached to the beam at the centre, so as to allow 
the beam each time to be brought to zero by a control magnet outside. The 
natural tourmaline was discarded, and a large and very perfect crystal of artificial 
tourmaline suspended in its place. The standard candle was kept two inches from 
the polarising tourmaline, and sheets of glass enclosing an air space between, and a 
water cell 5 millims. thick, were interposed to cut off radiant heat. The apparatus 
was well packed with cotton wool, except in the actual paths of the light. 
The means of several observations are collected below into groups. The apparatus 
was allowed to come to perfect rest between each observation, and the lighted candle 
was only brought near it just before the screen was removed to take each observation. 
Artificial and natural 
tourmalines. 
Parallel. 
Crossed. 
0 
O 
Group 1 
15- 
15-5 
„ 2 
16-5 
15* 
„ 3 
15-5 
16-5 
„ 4 
16-5 
16- 
„ 5 .... 
IS- 
15 - 5 
„ 6 
IS'S 
15-5 
„ 7 .... . 
16- 
16-5 
„ 8 .... . 
16- 
16'5 
„ 9 .... 1 
15-5 
Mean . . 
15-7 
15-9 
Here the difference has vanished to within the limits of accuracy of which the 
apparatus is capable. 
261. Finally, the control magnet was removed further off so as to get greater ampli- 
tude of swing. The sensitiveness of the apparatus was thereby considerably increased. 
The following are the means of three groups of observations taken with it so 
arranged : — 
