396 Dr. W. Watson on @ Quarte-T'hread 
into the thread being accurately adjusted and read off. A 
knowledge of the amount of twist is of assistance when 
adjusting the temperature compensation. ; 
The fixed mirror D is supported on an L-shaped piece of 
brass, which is held in place by three studs fixed to the base 
which pass through loose-fitting holes. The L-shaped piece 
is held up by small spiral springs against three nuts 8, R, Q, 
which screw on these studs. By adjusting these three nuts 
the position of the spot of light reflected from the fixed mirror 
can be adjusted. 
The suspended system consists of two magnets 8 cms. long 
and 1 mm. in diameter attached by means of small platinum 
straps, p (fig. 4), to two rods of fused silica which form 
part of the plate of fused silica C forming the mirror. 
These rods are bent as shown, and are fused to the ends 
E and F of the suspension fibre. There is also a small 
vertical rod of fused silica, which serves to carry a small 
weight g used when adjusting the sensitiveness. The reason 
why this gravity-bob is carried by a rod of fused silica is that, 
owing to the exceedingly small coefficient of expansion of 
silica, the centre of gravity, and hence the sensitiveness, is not 
appreciably altered by changes of temperature. Since the 
weight g cannot be moved vertically, the sensitiveness is 
adjusted by filing the weight till the required sensitiveness is 
nearly attained. The final adjustment is made by means of 
a very small weight r which can be screwed up and down. 
The upper surface of the mirror is platinized*. It will be 
noticed that the movable system consists of very few parts, 
while the mirror being attached to the fibres by fusion, we 
are in no way dependent on cements. 
Since the tension put into the thread is such as to stretch 
the spring G through two or three millimetres, the variations 
in tension produced by changes in the length of the metal 
base are insignificant, and will certainly, over the compa- 
ratively small temperature-range likely to occur, be pro- 
portional to the first power of the temperature, and hence can 
be eliminated when making the temperature adjustment. A 
photograph of one of the suspensions is shown in fig. 6. 
The adjustment for temperature-compensation is made by 
means of a small weight W which is clamped on to one of 
the magnets. The weight of the moving system is about 
3 grams. 
{n order to damp the vibrations of the magnet four copper 
plates K are placed near the poles. These plates are attached 
* Watson, Phil. Mag. July 1903, p. 191; Proc. Phys, Soc. of London, 
xviii. p. 502 (1903). 
