92 An Account of Experiments for determining the Length of \ 
struction of the pendulum, it became of considerable importance 
to select a mode of suspension equally free from objection. Dia- 
mond points, spheres, and the knife edge, were each considered; 
but as it was found difficult to procure diamond points suffi- 
ciently well executed, the knife edge was preferred, after many 
experiments had been made with spheres, the result of which it 
may not be useless for a moment to dwell upon. 
It is known, that if two curved surfaces be ground together 
in every possible direction, they will become portions of spheres; 
and thus a perfect sphere may be formed by grinding a ball in a 
hemispherical cup. If a pendulum vibrate on such a sphere, 
working in a conical aperture, itis evident that the centre of the 
sphere will be accurately in the axis of vibration, In trying this — 
method, however, it was found that the friction was so consi- 
derable, as to bring the pendulum to a state of rest after a few vi- 
brations ; and when the friction was sufficiently dimimished, by a 
contrivance which it is unnecessary todescribe,the lateral force of 
the pendulum in an arc of two degrees and a half, was sufficiently 
powerful to carry the ball entirely out of the socke 3 and it was 
consequently evident, that though the are of vibration might not 
be large enough to effect this, it must necessarily cause the ball 
in some degree to ascend the inclined piane of the aperture ; 
and this consideration induced me to abandon at once a mode 
of suspension which I should otherwise have esteemed the best 
that could have been employed. 
The principal objections to the use of a knife edge, appeared 
to be, the difficulty of forming it perfectly straight, and the pos- 
sibility that it might suffer a change of figure from wear, during 
the experiments, which might introduce an error not to be de- 
tected. The first of these objections I found to be perfectly 
roundless, as a knife edge can be made so as not to deviate 
sensibly from a right line. The second objection would indeed 
be of weight, were the usual method of determining the time of 
vibration resorted to, by comparing the pendulum with a clock, 
at the distant intervals of 24 hours $ ; but it will hereafter appear, 
that should any alteration in the form of the knife edge take 
place, it must become perceptible every ninth minute ; in addi- 
tion to which, I proposed to measure the distance of the knife. 
edges both before and after the experiments, when any change 
would of course be immediately detected. 
Description of the Pendulum employed. 
The pendulum constructed upon these principles is formed of 
a bar of plate brass, one inch and a half wide, and one eighth 
of an inch thick. Through this bar, two triangular holes are 
made, at the distance of 39:4 inches from each other, ta admit 
the 
