roller should be movable (on a screw thread), but so arranged
that it can be fixed at any point, say by a lock nut, screw, clamp, or
other simple means. It will be evident that, by shifting the roller, a
greater or less speed of the cone can be effected, and as to the end of
the cone's axis an index hand sweeping an ordinary clock face is attached,
the speed of this index hand can be regulated to a nicety, in proportion
to that of the drum. Of course, before fixing the size and proportion of
the disk and cone, the number of revolutions of the drum in a given time
must be ascertained by experiment. For instance, the drum being found to
make 15 revolutions in 12 hours, the proportions would be:

Circumference of roller = 12 units.
Circumference of middle part of cone = 15 units.

Or, the drum making 21/2 revolutions in 3 hours, equal to 9 revolutions in
12 hours:

Circumference of roller = 12 units.
Circumference of middle part of cone = 9 units.

Any slight inaccuracy can be compensated by the cone and disk device.

The drum, or cylinder, is caused to gradually revolve by a weight attached
to an endless cord passing once around the drum. The latter might be
varnished to prevent slipping. The weight should be provided with an
automatic wedge, allowing it to be slipped along the cord in an upward
direction, but preventing its descent. The weight is represented partly in
section in the engraving. This weight should not be quite sufficient to
revolve the drum, it being counterbalanced by the liquid raised in the
chambers of the drum. The liquid, however, following its tendency to seek
the lowest level, gradually runs back through the small hole, D, in the
partitions, but is continually raised again, with the chamber it has just
entered, by the weight slightly turning the cylinder as it (the weight)
gradually gains advantage over the as gradually diminishing weight of each
chamber raised.

As to the drum, the same might be constructed as follows, viz.: First
solder the partitions into the cylinder, making them slanting or having
the direction of chords of a circle (see Fig. 2). The end disks should be
dish shaped, as shown. Place them on a level surface, apply heat, and melt
some mastic or good sealing wax in the same. Then adjust the cylinder
part, with its partitions, allowing it to sink into the slight depth of
molten matter. In this way, or perhaps by employing a solution of rubber
instead of the sealing wax, the ch