472 MR JOHN ALLAN BROUN ON THE BIFILAR MAGNETOMETER, 
certain that the varying temperature of the copper pillars supporting the torsion 
circle has no effect on the position of the magnet. 
14. The following experiments will prove that we cannot at present determine 
the correction for the suspension of the magnet by mere theoretical considerations. 
Considering that the elasticity of the wires diminishes with increase of tempera- 
ture, and that the twist of the wires in turning the torsion circle acts with the 
torsion force, an increase of temperature should be equivalent to a diminution of 
the total torsion moment, and the north end of the bifilar magnet should approach 
the north. The effect of temperature on the magnet itself being to diminish its 
moment, is to cause the north end to move from the north. The temperature 
coefficient should therefore be diminished by this action on the wires (excepting 
the case where the proper twist of the wires is greater than 7, and in an opposite 
direction). 
15. Trevandrum Observatory, July 1858.—Having suspended a lead cylinder, 
carrying a mirror (the whole weight being about 5000 grains), by the silver wires 
of Grubb’s bifilar, a fine silk fibre from the cocoon was attached to one of the 
wires near the lower end, and the fibre was passed over a light copper friction 
wheel (whose axle was taken from a dipping needle) ; forty grains were suspended 
at the other end of the fibre. The torsion circle was then turned till the angle 
was about 65°, the pull of the fibre over the friction roller being at right angles to 
the plane, passing through the lower ends of the two wires. The movements of 
the weight round the vertical axis were observed by means of a telescope and a 
glass scale at a distance of seven feet; the scale readings increased when the move- 
ment was direct, and one scale division was equal to 0°327. According to hypo- 
thesis, if the torsion was direct, an increase of temperature should correspond to 
a retrograde movement, or a diminution of scale readings, and vice versa. Hourly 
observations were made during the months of June and July 1858; but as the 
slightest shocks touching the pillar of the instrument changed the position of the 
weight, the series was broken from day to day. In the first series all the obser- 
vations comparable were obtained from July 21 to August 5, 1858. The tem- 
perature was obtained from a thermometer within the box, having its bulb resting 
on a brass cup. 
16. A change of the daily mean scale reading from day to day throughout the 
period is obvious (see second column of Table II.). On the assumption that this 
change was regular, a correction of + 2-0 Sc. div. a day from July 21 has been 
applied; the reduced quantities are given in the third column. The individual 
results are by no means satisfactory. They show generally, however, an increase 
of reading for an increase of temperature; and the final result is, that + 1°:0 Fahr. 
is equivalent to + 40 Sc. div., thus satisfying the hypothesis. 

