of light at the anterior surface of the glass must be considered: that plane is the reflecting plane which is equidistant from the anterior and posterior surfaces of the mirror. The mirror is fixed to that end of the magnet bar which is turned towards the telescope, and must form with it so solid a system that no reciprocal disarrangement of either may be feared during the experiment, although the magnet bar be taken out of the stirrup and replaced in it in a reversed position. Moreover, the mirror must have such a position relatively to the bar, that the normal of the mirror shall be quite, or very nearly, parallel to the magnetic axis of the bar. The mirror-holder represented at fig. 4. may serve both these purposes; its frame is attached by screws to the bar. The frame-work supporting the mirror may be turned by screw motion round two rectangular axes, by which it may be brought into the required position.
8. The suspender, elevating screw, and suspension thread.—It is very advantageous to fix to the ceiling the thread, which is to carry the magnet bar, as by this it is sufficiently insulated from the floor and protected from all shaking, and because a proper length may in this manner be given to the thread. If, for the support of the magnetometer, we employ, not a metal wire, (the elasticity of which for an equal tenacity is almost ten times greater than that of one formed of silk fibres,) but a thread composed of parallel fibres of raw silk, it lengthens greatly, especially at first; and it is therefore requisite from time to time to raise it, so that the magnet bar and the mirror fixed to it may regain their original height. In raising the thread it is necessary that it should not be displaced in a lateral direction. A screw may be employed for this purpose, in the grooves of which the thread lies, and upon which it can be wound up still further, while the end of the screw works into a fixed nut. The groove in which the thread places itself, by the turning forwards of the screw, takes then, of itself, (from the advance of the whole screw) the place which the vertically suspended line had before occupied. The fixed nut, with solid rest, through which the screw pin passes freely, is let into a wooden slider which is mortised into a large plank fixed to the ceiling, and can slide therein in a direction parallel with the north or south wall of the room. If the position of the magnetic meridian should change in the lapse of time in any considerable degree, this slider will serve to retain the magneto-
