essure. It is,
however, possible that it may be associated with some kind of
permanent set taken by the fibres during the stage of passage from the
liquid to the solid state.

Sec. 90. On the Attachment of Quartz Fibres.

For many purposes it is sufficient to cement the fibres in position by
means of ordinary yellow shellac, but where very great accuracy is
aimed at, the shellac (being itself imperfectly elastic and exposed to
shearing stress) imposes its imperfections on the whole system. This
source of error can be got over by soldering the threads in position.
Attempts were made by the writer in this direction, with fair success,
in 1889, but as Mr. Boys has carried the art to a high degree of
perfection, I will suppress the description of my own method and
describe his in preference. It has, of course, been frequently
repeated in my laboratory.

In many cases, however, if not in all, it may be replaced by Margot
soldering, as already described, a note on the application of which to
this purpose will follow.

A thread of the proper diameter having been selected, it is cut to the
right length. With fine threads this is not always a perfectly easy
matter. The best way is for the operator to station himself facing a
good light, not sunlight, which is too tiring to the eye, but bright
diffused light. The thread will be furnished with bits of paper stuck
on with paraffin at both ends, as already described.

A rough sketch of the apparatus--or, at all events, two lines showing
the exact length which the free part of the thread must have--are
marked on a smooth board, and this is supported with its plane
vertical. The thread is held against the board, and the upper piece
of paper is stuck lightly to the board with a trace of soft wax, so
that the lower edge of the paper is at any desired height above the
upper mark. This distance is measured, and forms the length of thread
allowed to overlap the support. A second bit of paper is attached
below the lower mark, a margin for the attachment of the lower end
being measured and left as before. The thread will be most easily
seen if the board is painted a dead black.

If it is desired to attach the thread to its supports merely by
shellac, this is practically all that needs to be done. The supports
should resemble large pins. The upper support will be a brass wire in
most cases, and will require to be filed away as shown in the sketch
(Fig. 71). It is then co