without being cleaned; it being only necessary to add a little sulphate of copper and water from time to time, to supply the necessary waste. The only power demanded of the electromagnets is the unlocking of the mechanism, which is driven by weight power.

In fig. 1, we have

a sectional view of the lower leg of the siphon, showing the principle on which this method is based. It may be necessary to remark, however, that the electro-magnets and battery do not occupy these positions in reality, but are placed here for convenience of illustration.

1

W = wheel with 40 teeth in which is inserted the screw S. n = a small steel wire passing through the brass cap c, to prevent the disk d from revolving.

e = an ivory disk inserted in the tube, to prevent the float b from rubbing against the sides of the tube.

Now suppose the mercury should rise in the short leg of the siphon, as represented in the figure. The float b will be raised, and cause the platinum disk d to come in contact with the point of the platinum wire p, closing the circuit through the electromagnet m; the armature of which being attracted, unlocks the

clock-work, and allows the wheel a to make a complete revolution. By this means the wheel W is advanced one tooth, which raises the screw S the of an inch, and consequently carries the point p that distance away from the disk d.

As long as the mercury rises, the magnet m will be operated, and the platinum point p will be kept the of an inch above

the disk d.

If, on the contrary, the mercury falls in the siphon, the under side of the platinum disk d will be brought in contact with the point of the wire p', thereby closing the circuit through the magnet m'; the armature of which allows the one tooth wheel a' to make a complete revolution, thereby causing the screw S to be depressed the of an inch, carrying of course, the platinum point p' with it.

It will now be readily seen how the platinum disk d, carried by the float b, may always be maintained midway between the two points p and p', and distant a little less than the ' of an iuch from each.

The barometer is of the siphon form; the inside diameter of the portions near the surface of the mercury is nearly one inch. The upper and lower portions were made from the same glass tube, the two being connected by a tube of smaller diameter. The experiments and observations, so far, indicate that there is no appreciable difference in the size of the two legs of the siphon.

The float b is of ivory; the form a paraboloid of revolution. The under side of this float is very slightly concave. The diameter is one-tenth of an inch less than the inside diameter of the tube, so that there is no friction between the sides of the float and glass. The platinum disk d is supported by a steel wire passing through a brass cap c fitted on the top of the tube, and an ivory disk e inserted at a distance of 2 inches above the float b. The ivory disk is connected with the brass cap by means of two wires, so that it can readily be removed. A light steel wire n passes through a hole in the cap, for the purpose of preventing the disk d from revolving. This is made sufficiently free to prevent any friction.

The disk d is made of brass one-half an inch in diameter, and is covered on both sides with platinum plates.

The wire p is attached to a fine screw, for adjusting the distance of the points p and p' from the surface of the disk d.

These wires p and p' are, of course, insulated by being attached to an ivory block, as shown in the figure. The wires from these points are led to the top of the screw S, where they are fastened to an ivory block, after which they are connected with the electro-magnets m, m'.

A platinum wire is inserted in the side of the barometer tube,

and passes down in the mercury on the side of the float b. This wire is also connected with one pole of the battery.

The principle employed for giving motion to the screw S, which follows the fluctuations of the mercurial column, has been taken from the stop work long used on clocks. The barrel of a clock on which the cord is wound usually has a one-tooth wheel on its axis; and at every revolution of the barrel, a cog wheel is made to advance one tooth. This cog wheel is, of course, always detached from the barrel tooth wheel, except when in the act of advancing the tooth. In fig. 2, we have a vertical view of a portion of the mechanism, showing the method of communicating motion to the screw S. The one tooth wheels, a a', when at rest occupy the positions as shown in the drawing; and being detached from the cog wheel W, it is free to move in either direction. The screw S, which is shown in fig. 1, is raised

or depressed by the revolution of the wheel W. The one-tooth wheels a and a', moving in the direction of the arrows, give opposite motions to the wheel W; the office of a being to elevate the screw, and of a' to depress it, corresponding to the fall and rise of the mercurial column.

The mechanism for giving motion to the wheels a and a' is ordinary clock work, each being directly acted on by the barrel wheel, which is driven by a weight. One revolution of the barrel corresponds to twelve of the wheels a and a'. The axles, to which are attached a, a', carry another wheel having a single half-tooth, as shown in the drawing, fig. 2, which, resting against a little projection on the armature of the magnet, holds the wheel in the position as shown in the figure.

In order that the wheels a and a' may not revolve with too great rapidity, a train of clock work is connected, consisting of two additional axles, a fan being attached to the latter, by means of which the motion can be regulated to any desirable velocity. Three axles would undoubtedly be sufficient, the barrel axle, the axles a, a', and an additional one for the fan. We adopted the present form, because we happened to have a couple of clock movements at hand, and used them just as they were.

In order to prevent the cogs of the wheels a, a', from coming to the circumference of W at the same time, during rapid oscillations of the barometrical column, two circuit-"breakers" were connected; so that, at every revolution the circuit is interrupted, and neither wheel can revolve until they both are

at rest.

Fig. 3 is a perspective view of the apparatus as it is when in operation. The frame work for supporting the barometer tube and other mechanism is of black walnut two inches thick, which is firmly fastened to the east wall of the west transit room. This wall is built of brick, and is two feet thick, so that the whole apparatus occupies a very firm position.

Having given a general idea of the mechanism for causing the screw S to follow the motions of the barometrical column, we will show how the curve of pressure is recorded, as well as the printed results.

The wheel W, fig. 2, which receives the impulses, has 40 teeth; and the screw S, having 50 threads to the inch, one tooth of the wheel W corresponds to the of an inch change in the barometrical column, or Too of an inch change of pressure. To the wheel W is attached another of nearly the same diameter, having 80 teeth; this wheel is geared into one of 40 teeth carrying an 80-tooth wheel on the same axle. This second 80-tooth wheel is geared into a 50-tooth wheel, which operates the screw S', fig. 3, of 26 threads to the inch. To this screw is attached an arm, carrying a pencil which traces the curve of pressure on the revolving cylinder o.

From this arrangement, the curve is magnified a little more than three times the barometrical pressure. It would have been an easy matter to adapt the second screw and cog wheel, so that the curve would be exactly an integer scale-say 1, 2, 3 or 4 times; but as our printed results may be obtained much more accurately, and as often as is necessary, it was not thought of sufficient importance to construct a screw especially for this

purpose.

We will now explain the mechanism for printing the results.

AM. JOUR. SCI.-SECOND SERIES, VOL. XLI, No. 121.-JAN., 1866.