. The
graduation makes it easy to re-establish any harmony after the weights have
been shifted.

Suspensions.--For a harmonograph to give satisfaction it is necessary
that very little friction should be set up at the point of suspension, so
that the pendulums may lose amplitude of swing very slowly.

One-way suspensions are easily made. Two types, the point and knife-edge
respectively, are shown in Fig. 168 and the top part of Fig. 172. The point
suspension is most suitable for small rods and moderate weights; the
knife-edge for large rods and heavy weights which would tend to crush a
fine point.

[Illustration: FIG. 171.--Gimbal giving universal movement: point
suspension.]

Points should rest in cup-shaped depressions in a metal plate; knife-edges
in V-shaped grooves in a metal ring.

[Illustration: FIG. 172.--Knife-edge universal-motion gimbal.]

Screws turned or filed to a sharp end make convenient points, as they can
be quickly adjusted so that a line joining the points lies exactly at right
angles to the pendulum. The cups to take the points should not be drilled
until the points have been thus adjusted. Make a punch mark on the
bedplate, and using this as centre for one of the points, describe an arc
of a circle with the other. This will give the exact centre for the other
cup. It is evident that if points and cup centres do not coincide exactly
there must be a certain amount of jamming and consequent friction.

In making a knife-edge, such as that shown in Fig. 172, put the finishing
touches on with a flat file drawn lengthwise to ensure the edge being
rectilinear. For the same reason the V slots in the ring support should be
worked out together. If they are formed separately, the chances are against
their being in line with one another.

Gimbals, or universal joints, giving motion in all directions, require the
employment of a ring which supports one pair of edges or points (Fig. 172),
and is itself supported on another pair of edges or points set at right
angles to the first. The cups or nicks in the ring should come halfway
through, so that all four points of suspension shall be in the same plane.
If they are not, the pendulum will not have the same swing-period in all
directions. If a gimbal does not work with equal freedom in all ways, there
will be a tendency for the pendulum to lose motion in the direction in
which most friction occurs.

By wedging up the ring of a gimbal the motion of the pendu