ACTION OF HEAT ON MAGNECRYSTALS. 
169 
employed, was a copper vessel 1*15 inch broad in the direction of the magnetic axis, 
3'5 long and 7 deep. When in its place between the magnetic poles, they could 
either bear against its sides or against blocks placed between it and them, so that the 
pole distances should be unchangeable for the time. The upper part was clothed 
in flannel and was within the balance box (3368.) ; the lower part passed through a 
hole in the magnet table, and could be heated by a spirit-lamp applied below. Oil 
was most frequently employed in this bath for high temperatures, but sometimes 
water; and then its surface was covered with oil to prevent evaporation and diminish 
the production of currents. A thermometer was inserted in this bath at one end to 
indicate its temperature. 
3396. A copper cylinder, T1 inch in diameter, 3 inches in depth, closed at the 
bottom, and expanded at the upper edge, so as to rest on the side edges of the bath, 
was destined to hold the medium, either carnphine, water, or oil, which was immedi- 
ately around the crystal or other magnetic object. Currents were, of necessity, 
formed in the fluid, for it could neither be heated nor cooled without them ; but 
the point was to reduce them as much as possible about the object to be observed, 
and the arrangement described was found very useful for this purpose. It was 
necessary that the fluid used in this cylinder should be very clean and clear from 
any filaments or other matters, that might obstruct the motion of the immersed 
object. 
3397 . Because of the relative positions of the thermometer and the object to be 
observed, it is evident, that, with rising temperatures, the former will at the same 
moment be hotter than the latter, whilst with falling temperatures it will be cooler. 
The influence of this circumstance was observed in many of the experiments (3408.) ; 
but as the cooling was much slower and far more regular than the heating, the chief 
observations were made as the temperature fell. The greatest source of errors existed 
in the currents, and could only be overcome, and then only in part, by slow and 
numerous observations. These currents were often found to have prevalent sets, but 
these were, to a large extent, remedied by the observations in two positions, i.e. at 
the upsetting points. For low temperatures a smaller trough was employed, well 
clothed in flannel and filled with an excellent frigorific mixture. 
3398. Bismuth crystal.— The crystal before described (3373.) was placed on the 
torsion balance; its upsetting angle was then ascertained to be 105°; and being 
observed from time to time, it was found to remain the same both for low and high 
temperatures. The torsion force was measured, first at common temperatures, then 
as the temperature rose, and also as it fell ; it was observed to be greater at the same 
upper temperature when rising than when falling; an eflPect referred by particular 
examination to the fact that the lower the temperature of the bismuth the greater 
the torsion force; and that, as before said, as the bismuth gained its temperature 
later than the oil and thermometer in the bath, so it was cooler in the first case than 
in the second, for the same thermometer indication. As the cooling was purposely 
rendered slow, that the temperature of the bismuth might be near to that indicated 
