THE MAKERSTOUN OBSERVATIONS. 11 
till each length of fibre bore nearly an equal strain. Third, That the thread 
should be as small as is consistent with durability. The number of fibres for 
a particular weight should be determined by experiment; sixteen fibres of the 
silk supplied to the Makerstoun Observatory, and bearing nearly a pound weight, 
were found insufficient several times, though that number was recommended in 
the Report of the Royal Society; this was probably due to some difference in the 
thread: a thread of 22 fibres has now performed well for seven years. 
I should notice that M. Nervanper has proposed to form the suspension 
thread, by moistening with hot water the fibre cut into lengths, and submitting 
each length in this state to a considerable tension, before combining them to form 
the thread. 
The Bifilar Magnetometer.—The chief source of error for this instrument is 
also to be found in its suspension; wires of silver or of gold have generally been 
adopted ; although threads like those for the declinometer have also been employed, 
as at Greenwich. I object to the use of skeins, not merely because the errors 
due to a silk suspension are probably greater than those due to wires, but chiefly 
because no correction can be applied for the errors due to the former, while those 
due to the latter can be wholly eliminated. One error common to every kind of 
suspension is due to the stretching of the thread or wires, which gives a false 
value of the secular change; it is curious that no attempt has been made to 
eliminate this error, which could easily be done by means of a small apparatus 
for measuring the distance between the suspended magnet and the base plate of 
instrument. Any twist in the wires will give a false value of the unit coefficient, 
if determined in the usual way by the torsion circle; I am afraid that there is an 
equal chance of a similar error with silk threads, besides the probable variation 
due to breaking of fibres, &c. The error arising from twist, however, does not 
appear, when the process, which I have proposed and used, of determining the 
unit coefficient by deflections is adopted. The error which is peculiar to the wire 
suspension is that already noticed of a variation due to temperature; this error, 
however, also disappears when the temperature coefficient is determined by my 
process, since then, the total effect of temperature upon the position of the magnet 
is at once obtained. There remains against the silk suspension the heavy and in- 
determinable errors due to humidity, and to the gradual breaking of fibres. 
By the use of metallic wires and an apparatus to determine the amount which 
they stretch, nothing is required to render this instrument as perfect as possible 
but a magnet with permanent magnetism, the proper processes being employed 
for the determinations of the unit and temperature coefficients. 
The Balance Magnetometer.—I conceive that with proper care this apparatus 
may give considerably better results than have ever yet been obtained from it, 
