into collision with the attracting object, corresponding exactly to the case of the top when its velocity is not sufficient to prevent it from falling. But if the velocity is greater than that which is necessary to keep the nearest point of the ring at a uniform distance from P, then PD will be greater than PA, and the ring will become less and less eccentric, the circle, or rather the curve, ADEF gradually enlarging till it coincide with the ring, and the curve CC' gradually contracting till it disappear in the central point. The ring then becomes perfectly concentric with the planet, and the state of stable equilibrium is restored. The latter case corresponds again to the case of the top whose velocity is sufficient to cause it to rise towards a vertical position.*

Such was my theory, formed independently of experiment, but afterwards confirmed by it. After repeated trials, I succeeded, by means of the apparatus represented in the following drawing, in showing it.

M is a magnet supported on a stand, and R an iron ring capable of revolving rapidly. E is a wooden support to contain the ring. D is an appendage employed for the purpose of bringing the centre of gravity of the whole to the same level with the point of support, and so getting rid of any conical motion which the axis of the ring might have independently of the magnet. The ring R, its support E, and the appendage D, revolve together upon a hollow on the top of the stem C, and are set in motion before the magnet is introduced. It is then found that the ring, when revolving with sufficient rapidity, is not, as Laplace asserts, in instable equilibrium, but that the rotatory motion is able to preserve it from collision with the magnet. We find also precisely the same eccentric revolution which was anticipated by theory, and corresponding exactly to that which, as I have previously stated, is observed in Saturn's ring itself.

*If the top never reach a perfectly vertical position, neither will the ring ever become perfectly concentric with the planet. But it is sufficient if we establish that it will constantly tend towards that state, approaching indefinitely near to it.

The power of preserving the equilibrium, in the model, is so decided, that the whole apparatus may be turned consi

M

R

E

D

C

B

derably on one side, without derangement, the ring accompanying the magnet; and, if so turned before the introduction of the magnet, the magnet will bring the ring into a concentric position, permitting its introduction into it, while a non-magnetic bar cannot be so introduced. The magnet, instead of causing collision, prevents it.*

I have thus, both by theory and by experiment, attempted to explain that phenomenon, hitherto, as I think, not accounted for in any satisfactory manner, and to show that it rests on the same principle as the standing of a spinning-top, the precession of the equinoxes, the retrogradation of the moon's nodes, and the perturbation of the planes of the orbits of the planets. How far I have succeeded I leave others to decide.

* The loose structure, or fluidity, of Saturn's ring will not affect my theory, so long as there is sufficient cohesion, or mutual attraction, among its component particles, to keep them together as one body, and sufficient velocity of rotation to preserve the annular form.

On an Improved Safe Lock. By GEORGE MITCHELL, Letter-carrier, Edinburgh.

The following method of making a lock secure has been arrived at, owing to its peculiar (yet simple) mode of con

struction.

We shall at once refer the reader to the diagrams, where, with the following explanations, we shall endeavour to prove it safe to the extent that it is impossible to open the lock by any other means than with its own key.

A is an inside drawing of the lock; B, the bolt; C, the players; D, the protector; E, the revolver, with the revolv

ing plate, R, which is made fast to the revolver, E, at a distance equal to the thickness of the lock. FF' (of figures 1 and 2) are entrances to the lock from behind. G, entrance to the false chamber, O, which is made fast to the back of the lock. Fig. 3 is a drawing of the false chamber, which is

supplied with two sets of false players, pp'. The key, after having passed the false players of either side, and reached the top of the chamber, is brought forward into the lock by the entrance F or F', one of which is now exposed by the key's clearance in the revolving plate, R, having reached that point which is necessary for the key to enter the lock, either to lock or unlock as the case may be. But previous to the entrance, F or F', being exposed to admit the key, the entrance, G, is shut up, so that the entrance to the lock is at no time exposed.

The cut in the key, is made to clear the band on the top of the protector, which secures the revolver, while at the same time the band affords protection to the players of the lock.

We shall suppose, as an illustration, that by some means an impression of the lock is endeavoured to be obtained. In the first place, the impression from the false players (which, by the way, are different on either side) is obtained, the picklock passes on into the lock, and gets an impression such as can be got by obliterating the impression it had received

in the false chamber; returning, he gets the impression from the false players, which are no guide whatever to the work of the lock, which, as a matter of course, the term false implies. Under these circumstances, therefore, the lock may be considered secure.

New Designs for Iron Roofs of great clear Span. By ROBERT HENRY Bow, Civil-Engineer, Edinburgh.* (With a Plate.)

After making some introductory remarks, and insisting upon the propriety of employing roofs of great clear span for principal railway stations, the author institutes a comparison between the different classes of structures employed for the primary supporting frames, or principals, of roofs; and deduces that the triangular frame (in which the rafters constitute the main compressed member of the fabric) deserves to be preferred before all arched, compound, or other forms, when the nature of the covering to be used demands a considerably inclined surface, as for slates or glass. And he further shows that where untied or abutting principals can be used, rafters, when made straight, and trussed as for bridges, form principals of a very economical character; but that, for such a situation, rigid arched structures are quite imadmissible.

In the designs, Nos. 2 and 3, Plate XXVI., proposed by Mr Bow, each rafter is treated as a bridge. In order to test the merits of these, as suitable forms for long spans, they are compared with the design No. 1, which is the best at present in use.

In the calculations undertaken in order to make the comparisons, the weight of each part is represented by the product of the successive multiplication of its length by its strain, and the allowance of metal per ton of strain. For ties the sectional area of metal is estimated at one-eighth of a square inch; for rafters at quarter of a square inch; and for the

* Read 25th April 1853 and 9th January 1854.