406 
2. The Bifilar Magnetometer : its Errors and Corrections. 
By John Allan Broun, F.R.S., Director of the Trevandrum 
Observatory. Communicated by Professor Tait. 
The bifilar magnetometer has been employed in all the magnetical 
observations, with an exception or two ; its theory has been stated 
generally, but no careful examination of the errors of the instrument, 
as deduced from experience, has been published. The author has 
attempted to supply this deficiency. 
In the theory of the instrument, the proper elasticity of the wires 
has been neglected, as of little importance. The author shows that 
the wires receive a twist in the adjustment, which puts the upper 
suspension-points into a different plane from the lower ones ; that 
the force so produced enters into the equation of equilibrium in a 
marked manner. The theoretical conclusion is verified by experi- 
ments on the torsion force of the wires. But as a proper twist 
generally exists, the angle of torsion, as well as the time of vibra- 
tion, are unfitted for an accurate determination of the coefficient 
which relates the same divisions to the unit. Different methods of 
determining the unit coefficient are noted, and the results of experi- 
ments are given. 
The greatest source of error in connection with the observations 
is that due to varying temperature. It has been usual to suppose 
that the temperature affected only the magnetic moment of the 
magnet, and the length of the wires and the interval separating 
them. It is shown that it affects also the coefficient of elasticity of 
the wires. Experiments made hourly during many days, with an 
unmagnetic weight suspended by two wires, are cited, to show that 
the varying temperature produces effects on the position which might 
be expected from theoretical considerations. 
A method proposed by the author in 1842, and described in the 
“ Edinburgh Transactions” in 1845 (vol. xvi. p. 74), by which a 
temperature coefficient is found, including all the effects, is then 
noticed, and the results obtained from discussions of all the obser- 
vations of the bifilars in the colonial observatories (including the 
observations of nearly twenty-eight years), are given. From these 
it appears that the coefficients usually employed are on the average 
nearly one-half too great, and in some instances they are the 
double of the true coefficients. 
