mate coefficient of linear expansion of quartz per degree
between 80 deg. C. and 30 deg. C. is 0.0000017 (Threlfall = loc. cit.).

This must be regarded with some suspicion, as the data were not
concordant. There is no doubt, however, about the extreme
inexpansibility of quartz.

Temperature coefficient of modulus of torsional rigidity per degree
centigrade, 22 deg. to 98 deg. C, 0.000133

Ditto, absolute simple rigidity, 0.000128 (Threlfall).

Limit of allowable rate of twist in round numbers is, one-third turn
per centimetre, in a fibre 0.01 cm. diameter.

The limiting rate is probably roughly inversely as the diameter.

Attention must be called to the rapid increase in the torsional
rigidity of these threads as the temperature rises. A quartz spiral
spring-balance will be appreciably stronger in hot weather.

Sec. 89. In the majority of instances in which quartz threads are
applied in the laboratory, it is desirable to keep the coefficient of
torsion as small as possible, and hence threads are used as fine as
possible.

It is convenient to remember that a thread 0.0014 cm. or 0.0007 inch
in diameter breaks with a weight of about ten grammes, and may
conveniently be employed to carry, say, five grammes. With threads
three times finer the breaking strength per unit area increases, say,
50 per cent. In ordinary practice--galvanometric work for
instance--where it is desirable to use a thread as fine and short as
possible to sustain a weight up to, say, half a gramme, it will be
found that fibres five centimetres long or over give no trouble
through defect of elastic properties. A factor of safety of two is
a fair allowance when loading threads.

No difficulty will be experienced in mounting threads having a
diameter of 0.0002 inch or over. With finer threads it is necessary
to employ very dark backgrounds (Mr. Boys uses the darkness of a
slightly opened drawer), or the threads cannot be sufficiently well
seen.

In the case of instruments in which threads remain highly twisted for
long periods of time, the above rule as to the safe limit of twist
does not allow of a sufficient margin; it is only applicable to
galvanometric and similar purposes.

The cause of the increase in tenacity as the diameter diminishes is at
present unknown. It is due neither to an effect of annealing
(annealed threads are rotten), nor is it a skin effect, nor is it due
to the cooling of the thread under higher capillary pr