OH THE ELECTRICAL PROPERTIES OF PITRE SULPHUR. 
87 
made to screen half the magnetic system, while the other half was being magnetized, 
failed, even when the screening was attained by pushing half the system into a 
cylindrical hole bored in a thick soft iron bar. During the process of mounting the 
discs on aluminium wire, they generally got magnetized, and hence required to 
be demagnetized before the final magnetizing process was carried out. The 
demagnetizing was carried out by means of the arrangement of four similar electro¬ 
magnets. which were supplied with an alternating current. If we had foreseen the 
necessity, we should have provided these magnets with laminated cores ; as it was, 
care had to be exercised to prevent them becoming so hot as to melt the paraffin 
used as a cement for the disc magnets. The final adjustments for astaticism were 
made by twisting the aluminium wire. This wire was only ‘35 millim. in diameter, 
and so the difficulty of obtaining a perfectly straight system of magnets of about 
20 centims. long, and so astatic that the natural period of vibration was from 
l'b" to 2", will be understood by anyone engaged in similar work. The natural period 
of vibration of two magnetized discs was a fraction of a second. We had hoped that 
it would turn out more easy to secure good astaticism with four-magnet systems than 
with two, but this hope was not justified by the event. The adjusting and mounting 
of the astatic system is the most difficult part of the manipulation ; the final adjust¬ 
ment must be made by hand twisting, as described in the ‘Phil. Mag.’ (Joe. citX 
(6.) The fibre must have a minimum torsional coefficient. In plain English, the 
fibre must be as thin and long as possible ; and it is, of course, much more important 
to have the fibre thin than long. The proper thickness was calculated from the results 
obtained by one of us in a paper on “Quartz Threads” (‘Phil. Mag.’ [5], vol. 30, 
p. 99, 1890), and the thread was then picked out by measurement with a microscope ; 
a factor of safety of about 2 being generally allowed. As a matter of fact, the thread 
required is very fine, but not unmanageably so : pieces about 25 centims. long were 
generally used. We find that the thread can be fastened more securely by a little 
melted shellac than by hard paraffin ; we have had several accidents from the thread 
slipping out of the paraffin. Of course shellac that has not been overheated or 
dissolved in alcohol, must be used. Care was always taken to use the thread in an 
untwisted state. 
On the General Design ofi the Galvanometer for High Degrees of Sensitiveness. 
The general principles had been arrived at in 1889 from experience with the old 
galvanometer, from which fair results had been obtained. The following is the result 
of our general experience. 
(l.) It is of the first importance that the coils be adjustable to the suspended part, 
as well as the suspended parts to the coils. Unless provision be made for this it is 
practically beyond the powers of even the most skilful manipulator to make use of 
magnets properly filling up the coil space, and yet free to turn. This condition 
