beyond that the density will be perfectly uniform, ẞde being a sphere concentric to BDE, and of such a size, that if the matter in BDE was spread uniformly all over the sphere ẞde, its density would be the same as beyond it.

For any quantity of matter spread uniformly over the globe Boe or BDE affects a particle of matter placed without that sphere just in the same manner as if the whole fluid was collected in the center of the sphere, so that any particle of matter placed without the sphere ẞde will be in perfect equilibrio.

In like manner if the fluid within BDE is rarer than without, there will be a certain space surrounding the globe, as that between BDE and ẞde, in which the density will be infinite, or in which the particles will be pressed close together, and if the repulsion of the particles is inversely as the square of the distance, the density of the fluid beyond that will be uniform: the diameter of ẞde being such that if all the matter within it was spread uniformly, its density would be the same as without.

Let a fluid of the above-mentioned kind be spread uniformly through infinite space except in the canal acdef of any shape whatsoever, except that the ends aghb and mden are straight canals of an equal diameter, and of such a length that a particle placed at a or d shall not be sensibly affected by the repulsion of the matter in the part gcmnfh, and let there be a greater quantity of the fluid in this canal than in an equal space without.

Then the density of the fluid in different parts of the canal will be very different, but I imagine the density will be just the same at a as at d. For suppose ab and de to be joined, as in the figure, by a canal of an uniform diameter and regular shape, nowhere approaching near enough to gcmnfh to be affected by the repulsion of the particles within it. If the matter was not of the same density [at a and d] the matter therein could not be at rest, but there would be a continual current through the canal, which seems highly improbable.

COR. Let C be a conductor of electricity of any shape, em and fn wires extending from thence to a great distance. Let a and b be two

equal bodies placed on those wires at such a distance from C as not to be sensibly affected by the electricity thereof, and let the conductor or wires be electrified by any part: the quantity of electric fluid in the bodies a and b will not be sensibly different, or they will appear equally electrified.

A

B

C-
D

a

Case 1. Let the parallel planes Aa, Bb, &c., be continued infinitely. Let all infinite space except the space contained between Aa and Cc, and between Ee and Hh, be filled uniformly with particles repelling inversely as the square of their distance; let the space between Ee and Hh be filled with fluid of the same density, the particles of which can move from one part to another; and let the space between Aa and Cc be filled with matter whose density is to [that in] the rest of space as AD to AC.

H

k

Take EF = CD, and GH such that the matter between Ee and Ff when pressed close together, so that the particles touch each other, shall оссиру the space between Gg and Hh.

The space between Ee and Ff will be a vacuum, that between Ff and Gg of the same density as the rest of space; and between Gg and Hh the particles will touch one another.

Case 2. Let everything be as in case the first, except that there is a canal opening into the plane Hh, by which the matter in the space EH is at liberty to escape; part of the matter will then run out, and the density therein will be everywhere the same as without, except in the space EF, which will be a vacuum, EF being equal to CD.

Case 3. Suppose now that a canal opens into the plane Aa by which the fluid in the space AC may escape. It will have no tendency to do so, for the repulsion of the redundant fluid in AC on a particle at a will be exactly equal to [the] want of repulsion of space EH.

the

Case 4. Let now the space between Aa and Cc be filled with matter whose density is to the rest of space as AB to AC.

Then the space between Hh and Gg will be a vacuum, GH being equal to BC. In the space EF the particles of matter will be pressed together so as to touch each other, the quantity of matter therein exceeding what is naturally contained in that space by as much as is driven out of the space GH; and in the space between Ff and Gg the matter will be of the same density as without.

Case 5. Suppose now that a canal opens into the plane Hh as in Case 2, then will matter run into the space EII, and the density will be everywhere the same as without, except in the space EF, where the particles will be pressed close together, the quantity of matter therein.

exceeding the natural quantity by as much as is naturally contained in the space BC.

Case 6. Suppose now that a canal opens into the plane Aa, the fluid will have no tendency to run out thereat.

Case 7. Let us now consider what will be the result if the repulsion of the particles is inversely as some other power of the distance between that of the square and the cube; and first let us suppose matters as in the first case. There will be a certain space, as EF, which will be a vacuum, and a certain space, as FG, in which the particles will be pressed close together, for if the matter is uniform in EH, all the particles will be repelled towards H if there is not a vacuum at E, nor the particles pressed close together at G, but only the density less at E than at H, then the repulsion of space EH at E will be less on [a] particle at E and greater on a particle at H than if the density was uniform therein, consequently on that account as well as on account of the repulsion of AC a particle at E or H will be repelled towards H, but if the space EF is a vacuum and the particles in GH pressed close together, then if the spaces EF and GH are of a proper size, a particle at F or G may be in equilibrio.

Case 8. If you now suppose a canal to open into the plane Hh as in the 3rd case, some of the matter will run out thereat, so that the whole quantity of matter in the space EH will be less than natural. For if not, it has already been shown that a particle at H will be repelled from 4, but the quantity of matter which runs out will not be so much as the redundant matter in AC, for if there was, the want of repulsion of the space EH on a particle at h would be greater than the excess of repulsion of the space AC.

Case 9. Suppose now that a canal opens into the plane Aa as in Case 3; a particle at a will be repelled from Dd, but not with so much force as if there had been the natural quantity of fluid in the space EH, so that some of the fluid will run out at the canal, but not with so much force, nor will so much of the fluid run out as if there had been the natural quantity of fluid in EH.

Case 10. If you suppose matters to be as in the 4th case, then there must be a certain space adjacent to Ee, in which the particles will be pressed close together, and a certain space adjacent to Hh in which there must be a vacuum.

Case 11. If you suppose a canal to open into the plane Hh, some matter will run into the space EH thereby, so that the whole quantity of matter therein will be greater than natural.

The proof of these two cases is exactly similar to that of the two former.

Case 12. If you now suppose a canal to open into Aa, some fluid will run into it, but not with so much force nor in so great quantity as if the natural quantity of fluid had been contained in the space Hh.

I have supposed the planes Aa, &c. to be extended infinitely, because by that means I was enabled to solve the question accurately

in the cases where the repulsion is supposed inversely as the square of the distance, which I could not have done otherwise, but it is evident that the phenomena will be nearly of the same kind if the planes are not infinitely extended.

For if the distance ag be small in respect of the length and breadth of the plane Aa, a particle placed at a will be repelled by the plane Aa with very nearly the same force as if the plane was infinitely extended.

It is plain that these 6 last cases agree very exactly with the laws of electricity laid down in the 3rd and 4th hypotheses [Thoughts... Art. 202].

If the lines Bb and Dd touch one another so that

[Here the MS. ends. ED.]

NOTE 19, ART. 234.

Cavendish's Experiment on the Charge of a Globe between two
Hemispheres.

This experiment has recently been repeated at the Cavendish Laboratory in a somewhat different manner.

The hemispheres were fixed on an insulating stand, so as to form a spherical shell concentric with the globe, which stood inside the shell upon a short piece of a wide ebonite tube.

By this arrangement, since during the whole experiment the potentials of the globe and sphere remained sensibly equal, the insulating support of the globe was never exposed to the action of any sensible electromotive force, and therefore had no tendency to become charged.

If the other end of the insulator supporting the globe had been connected to earth, then, when the potential of the globe was high, electricity would have crept from it along the insulator, and would have crept back again when, in the second part of the experiment, the potential of the globe was sensibly zero. In fact this was the chief source of disturbance in Cavendish's experiment. See Art. 512.

Instead of removing the hemispheres before testing the potential of the globe, they were left in their position, but discharged to earth. The effect on the electrometer of a given charge of the globe was less than if the hemispheres had been removed, but this disadvantage was more than compensated by the perfect security from all external electric disturbances afforded by the conducting shell.

The short wire which formed the communication between the shell and the globe was fastened to a small metal disk hinged to the shell, and acting as a lid to a small hole in it, so that when the lid and its wire were lifted up by means of a silk string, the electrode of the

electrometer could be made to dip into the hole in the shell and rest on the globe within.

The electrometer was Thomson's Quadrant Electrometer.

The case of the electrometer, and one of the electrodes, were permanently connected to earth, and the testing electrode was also kept connected to earth, except when used to test the potential of the globe.

To estimate the original charge of the shell, a small brass ball was placed on an insulating stand at a distance of about 60 cm. from the centre of the shell.

The operations were conducted as follows:

The lid was closed, so that the shell communicated with the globe by the short wire.

A Leyden jar was charged from a machine in another room, the shell was charged from the jar, and the jar was taken out of the room again.

The small brass ball was then connected to earth for an instant, so as to give it a negative charge by induction, and was then left insulated.

The lid was then lifted up by means of the silk string, so as to take away the communication between the shell and the globe.

The shell was then discharged and kept connected to earth.

The testing electrode of the electrometer was then disconnected from earth, and made to pass through the hole in the shell so as to touch the globe within without touching the shell.

Not the slightest deflexion of the electrometer could be observed.

To test the sensitiveness of the apparatus, the shell was disconnected from earth and connected to the electrometer. The small brass ball was then discharged to earth.

This produced a large positive deflexion of the electrometer.

Now in the first part of the experiment, when the brass ball was connected to earth, it became charged negatively, the charge being about of the original positive charge of the shell.

When the shell was afterwards connected to earth the small ball induced on it a positive charge equal to about one-ninth of its own negative charge. When at the end of the experiment the small ball was discharged to earth, this charge remained on the shell, being about of its original charge.

Let us suppose that this produces a deflexion D of the electrometer, and let d be the largest deflexion which could escape observation in the first part of the experiment.