GRAVITATION AS AFFECTED BY TEMPERATURE. 
367 
form 3 (fig. 12). When a horizontal vibration reaches the torsion head of the 
suspending fibre, it passes down the fibre and reaches the top of the beam system, 
which at first experiences a simple horizontal vibration. Next, the vibration is carried 
down to the wires, m, m, each of which is set vibrating on a horizontal axis about its 
mass-centre. Soon we have a certain amount of pendular motion of two unequal periods. 
When at last the high frequency tremors of the fine wires and fibre have died out, 
there remain, very persistent in a high vacuum, the low frequency movements of m, m. 
All the unsymmetrical systems, 3, 4, 5, and 6, are troublesome for these reasons, 
whereas forms 1 and 2, give relatively little trouble. But the unsymmetrical form is, 
for a short beam, as shown by C. V. Boys, indispensable for sensitiveness. Working 
in the heart of a large city, I failed with Nos. 3, 4, 5, and 6 after long trial, and only 
succeeded with No. 7 (see fig. 10) by allowing in the vacuum chamber a small 
amount of air for damping purposes. 
It might be thought that some form of damping would be possible so as to render 
any beam system workable. Of the known damping methods (l) gas-friction and 
(2) liquid-friction are inadmissible in a high vacuum, but (3) electro-magnetic 
damping which I tried in several ways, failed always. It is impossible to have any 
magnetic material on the beam, and it is risky to have, say, a closed copper wire 
circuit carried on the beam. But there is one other damping method possible, viz., 
(4) rolling friction. I have used chains extensively as in No. 2 (fig. 12) and have 
found that the rolling of one link on the next brings in rolling friction to damp out 
tremors. Such a chain system acts very well in high vacuum, even in a very 
disturbed laboratory. (See results in Table I.) 
8. Sealing Materials Used. —The use of waxes in vacuum vessels is now well 
understood. For joining the optical window and the top glass plate to the vacuum 
tube (see figs. 9 and 10) I use (a) Faraday cement, or, better (b) a white “vapour-free” 
wax (supplied by Lilliendahl, Neudietendorf) ; these both seem much tougher than 
sealing wax or shellac. Then there is (c) a soft red sealing wax sold commercially, 
I never trust any ground glass joint, or any mercury-trap joint, or any platinum seal, 
but in all cases melt some of this wax outside the junction, and also on the top and 
bottom of all ground glass taps to ensure against leakage. ( d ) Ramsay’s tap grease 
is used to lubricate the taps. At moderate temperature, say 60° C., vapour comes 
off freely from (a) and (6), so such high temperatures have to be avoided for these 
seals, (c) and (d) are by no means vapour-free at ordinary temperatures, but the only 
way in which they come in contact with the vacuum is on the vacuum taps; and the 
small amount here used soon becomes vapour-free in the vacuum, without serious 
detriment to the latter. 
There are many joints in a Gaede mercury pump which become leaky periodically. 
It is well to serve all of them with some melted wax (c). 
9. Preparation of the Apparatus for an Experiment. —Suppose in this instance a 
copper internal system is to be used. The long glass tube (fig. 6) as received from 
3 d 2 
