J. A BROUN’S NOTE ON THE BIFILAR MAGNETOMETER. 43 
The quantity 2pv does not affect the unit coefficient only ; it has a much 
more weighty influence on the temperature coefficient. This was proved experi- 
mentally in the paper cited, where, however, it was only shown that the direc- 
tion of the temperature action was such as had been supposed by me previously ; 
but as the quantities obtained prove also that the amount of the action was such 
as to explain the differences found by the two methods for the determination of 
the temperature coefficient, I shall examine the results here. 
An unmagnetic weight was suspended by two silver wires ; the weight was 
turned round a vertical axis by the horizontal pull of a small weight hanging 
at the end of a silk fibre, acting at right angles to the line joining the wires, and 
passing over a delicate friction wheel.* As the action of currents of air within 
the box containing the weight produced a regular diurnal movement, the pull 
was made in the first series of experiments to one side, and in the second series 
to the opposite side. In the first series, which (from causes indicated) was the 
least consistent of the two, it was found that an increase of the daily mean 
temperature within the box (derived from twenty-four hourly observations) cor- 
responded to a movement of the weight in such a direction as to increase the 
bifilar torsion angle, so that the unifilar torsion force had diminished. In this 
case the mean reading changed by + 4:0 scale divisions for + 1° Fahr. When, 
as in the second series, the pull of the small weight was in the opposite direc- 
tion, an increase of temperature was found as before to coincide with a diminu- 
tion of the unifilar torsion force, the motion of the weight being also in the 
opposite direction, the daily mean reading changing by = 2°5 scale divisions for 
+1° Fahr. Both results confirmed the hypothesis, that the action of an in- 
crease of temperature was to diminish the force due to the torsion of the 
separate wires. It should be remarked, that the total effect of temperature in 
expanding the wires, and the metal which separates them, is to increase the 
bifilar torsion force, but to diminish the bifilar torsion angle, which is just the 
opposite of the result shown by these experiments. 
Taking the result of the second series of experiments as most free from 
error, we find that one scale division = 0327, the angle v = 65°, whence em- 
ploying the approximate coefficient from the bifilar torsion 65°, 
q = cot 65° (arc 0327) 2°5 
= 0°00011. 
Let us apply this result to the case of the bifilar magnetometer. We know 
that the effect of increase of temperature on the magnet is to diminish the 
magnetic moment ; the magnet then moves /rom the north; the result is as if 
the earth’s magnetism had diminished : the action of increase of temperature 
on the unifilar torsion of the wires corresponds to a diminution of the force 
* See Edinburgh Royal Society Transactions, xxii. p. 472, art. 15. 
