GRAVITATION AS AFFECTED BY TEMPERATURE. 
357 
This is the point established by C. Y. Boys that for a given convenient value of T, 
Ovaries as l/a and is independent of m. Hence we make the beam short for sensitive¬ 
ness. As to the value of T, we find the system is easily worked if the period does 
not exceed 5 mins. ; but when it approaches 30 mins., the control is insufficient. It 
is well to aim at a value of T about 4 mins. 
In my experience the best quartz fibres are of about 15/x diam. Taking this as 
the size of the fibre and knowing its length we arrive at m. Hence for a given value 
of T we obtain the length of the beam. Equation III shows that sensitiveness 
oc M 1/s oc B,D 1/5 , where It, D are the radius and density of the large sphere. 
We see, then, that radius rather than density of M should be large. Thus copper 
would do almost as well as lead, provided it can be obtained equally free from 
magnetic impurities. I used lead, partly for its greater effect, partly for purity, and 
partly for economy, but for high temperature work where lead would melt, copper or 
other material of high melting point would be used. 
IY. Precautions. 
1. Apart from gas effects, there are two kinds of force to be avoided, viz. :— 
(a) Electrostatic. —-Entirely surrounding the vacuum vessel, except at the window, 
are metal tubes through which tap water flows. This system forms a perfect earthed 
screen between any external field and the movable system. But it is possible that 
• in a vacuum charges may arise on gas particles leaving the solid surfaces after 
occlusion; or some internal charge may arise in some other way. Any such charge 
should be removed by the “earthed ” metal lining to the vacuum vessel. 
Again, charges arising from contact of different metals were avoided by having all 
the materials composing the -beam system (including the lining sheath of the tube 
mentioned above) made of the same metal. Thus, in the final arrangement, the balls, 
m, m, the wires carrying them, the beam frame, and the mirror case were of the 
purest silver. The only foreign materials in the whole system were the small beam 
mirror and a minute amount of Margot’s solder (see p. 364) used to fix the mirror 
case to the beam wire. 
( b ) Magnetic. —Impurities of iron, nickel and cobalt were avoided in the materials 
composing M, m and all parts connected to them. After working any of these 
with a tool, or after handling, the surfaces were dipped in nitric acid and well 
washed. 
Iron screws, clamps, &c., were not used on parts of the apparatus adjoining M, m. 
If the internal, moving, system were entirely unmagnetic then magnetism of the outer 
parts would not matter; but suppose the inner system is slightly magnetic, it would 
respond to any magnetic influence, say, from M. When temperature is changed this 
response would change also, for the permeability of M would change with temperature. 
Thus we should have an apparent change in gravitative attraction, and this spurious 
VOL. CCXYI.-A. 3 C 
