In a paper on “The Bifilar Magnetometer, its Errors and Corrections” &c., which appeared in the Transactions of the Royal Society of Edinburgh, vol. xxii. p. 467; it was shown that, in turning the apparatus which carries the upper extremities of the two wires through an angle exceeding by ν that through which the lower extremities and the magnet are moved, the upper end of each wire is also turned through an angle greater by ν than the lower end. The whole force acting on the magnet, to turn it from the meridian, is compounded of the bifilar and the unifilar torsion—only the former of which had been considered in the method issued by the Committee on Physics of the Royal Society of London for the instruction of directors of observatories.
It was sought in the paper cited to show the change that the neglected forces would introduce into the expression for the unit coefficient, which was found to be
where Δν is the arc value ot one scale division of the instrument, and β is the variation of ν due to the unifilar torsion. There were errors of copying, including omissions in the investigation given. I also remarked, after the paper was printed, that the value of k found was less than when the unifilar torsions were neglected (since cot (ν + β) < cot ν); whereas all the determinations by other methods, in which the unifilar torsions were included, had shown a greater coefficient.