When they are sufficiently charged B is let down on the upper coating, which instantly charges the plate to be tried, and immediately the wire Gg is lifted up from Dd, but not high enough to touch Aa. Consequently, immediately after the plate is charged, the communication between Dd and the ground is taken away, and consequently as fast as any fluid flows from the uncoated part of the under surface of the glass to the lower coating, some fluid will flow into Dd and overcharge it, and consequently make the pith balls separate.

303] In order to prevent, if possible, the ill effects proceeding from this spreading of the electricity, I took some coated plates of glass, and covered all the uncoated part with cement to the thickness of or an inch, as in Fig. 22, which represents a section Fig. 22.

B

passing through the middle of the plate perpendicular to its plane, and in which the glass plate and coatings are represented by the same letters as before, and the dotted lines represent the cement*. Thinking that it would be impossible for the electricity to spread between the cement and the glass, in which case this method must have been perfectly effectual, as it would be necessary for the electricity to spread itself not only on the perpendicular surface ef, but also to some distance on the horizontal surface fg, before the quantity of redundant fluid lodged on the surface of the cement could bear any sensible proportion to that in the coating CD.

304] The result was that in dry weather the electricity seemed to spread as fast on those plates which were covered with cement as on the others, but in damp weather not so fast, the difference between dry and damp weather being less in those plates which were covered with cement than the others; and besides that there seemed as much difference between the swiftness with which it spread on the surface of the Nuremberg and English plates * [Art. 484.]

after they were covered with cement as before, which shows plainly that the electricity spread between the cement and the glass, and not on the surface or through the substance of the cement. It could not be owing, I think, to its passing through the substance of the glass, for if it was, there would hardly be much difference in the uncoated plates between damp and dry weather, whereas, in reality, there was a very great one.

I also tried what effect varnishing the glass plates would have, but I did not find that it did better, if as well, as covering them with cement.

305] As there seemed, therefore, to be very little advantage in covering the plates with cement or varnishing them, and as it was attended with a good deal of trouble, I did not make use of those methods, but trusted only to letting the wires down and up pretty quick, so as to allow very little time for the electricity to spread on the surface of the plates, and this I have reason to think was sufficiently effectual, as I never found much difference in the divergence of the pith balls, whether the wires were let down and up almost as quick as I could, or whether they were suffered to rest a second or two at bottom.

306] As the wire Cc is suffered to rest so short a time on Aa and Bb, it is plain that the lower coatings of the trial plate and plate to be tried must have a very free communication with the ground and the outside coating of the jars, or else there would not be time for them to receive their full charge. I accordingly took care that the wires which made the communication should be clean and should touch each other in as broad a surface as I could conveniently. As for the method I took to have a ready communication with the ground, it is described in [Art. 258].

307] Besides this gradual spreading of the electricity on the surface of the glass, there is another sort which is of much worse consequence, as I know no method of guarding against it, namely, the electricity always spreads instantaneously on the surface of the glass to a small distance from the edge of the coating, on the same principle as it flies through the air in the form of a spark. This is visible in a dark room, as one may see a faint light on the surface of the glass all round the edges of the coating, especially if the glass is thin, for if it is thick it is not so visible*.

[From a photograph taken in the Cavendish Laboratory of a plate of glass

with a circular tinfoil coating on one side, a larger coating being applied to the other side of the glass. The electrification of the coatings was produced by an induction coil.]

308] There is another circumstance which shows this instantaneous spreading of the electricity, namely, after having charged and discharged a coated plate of glass a great many times together without cleaning it, I have frequently seen a narrow fringed ring of dirt on the glass all round the coating, the space between the ring and the coating being clean, and in general about inch broad*. This must in all probability have proceeded from some dirt being driven off from the tinfoil by the explosions, and deposited on the glass about the extremity of that space over which the electricity spreads instantaneously, and therefore seems to show that the distance to which the electricity spreads instantaneously is not very different from of an inch.

309] From some experiments which will be mentioned by and byt, I am inclined to think that the distance to which the electricity spreads instantaneously is about of an inch when the thickness of the glass is about of an inch and about 8 of an inch when its thickness is about of an inch; or more properly the quantity of redundant fluid which spreads itself on the surface of the glass is the same that it would be if the distance to which it spread was so much, and that the glass in all parts of that space was as much charged as it is in the coated part.

310] If I charged and discharged a coated plate several times running, in the dark, with intervals of not many seconds between each time, I commonly observed that the flash of light round the edges of the coating was stronger the first or second time than the succeeding ones, which seems to shew that the electricity spreads further the first or second time than the succeeding ones. Accordingly I frequently found in trying the following experiments that the pith balls would separate rather differently the first or second time of trying any coated plate than the succeeding one. Observing that I now speak of the half dozen trials which, as I said in [Art. 299], I commonly took with the same plate immediately after one another.

311] Before I proceed to the experiments it may be proper to remind the reader‡ that if a plate of glass or other non-conducting substance, either flat or concave on one side and convex on the + [See below, Arts. 314 to 323.] [See Art. 166, Prop. xxxiv., Cor. vi.]

[See Art. 538, Feb. 13, 1773.]

+

other, provided its thickness is very small in respect of its least radius of curvature, is coated on each side with plates of metal of any shape, of the same size and placed opposite to each other, its charge ought by the theory to be equal to that of a globe whose diameter is equal to the square of the semidiameter of a circle whose area equals that of the coating divided by twice the thickness of the glass, supposing the coated plate and globe to be placed at an infinite distance from any over or undercharged body, and to be connected to the jar by which they are electrified by canals of incompressible fluid; provided also that the electricity does not penetrate to any sensible depth into the substance of the glass, and that the thickness of the glass bears so small a proportion to the size of the coating that the electricity may be considered as spread uniformly thereon.

312] It was before said that the electricity spreads instantaneously to a certain distance on the surface of the glass, so that the surface of the glass charged with electricity is in reality somewhat greater than the area of the coating. Therefore, if the plate is flat, let the area of the coating be increased by a quantity which bears the same proportion to the real coating as the quantity of redundant fluid spread on the surface of the glass beyond the extent of the coating does to that spread on the coated part of the glass. That is, let the area of the coating be so much increased as to allow for the instantaneous spreading of the electricity, and let a circle be taken whose area equals that of the coating thus increased. I call the square of the semidiameter of this circle, divided by twice the thickness of the glass expressed in inches, the computed charge of the plate, because, according to the abovementioned suppositions, its charge ought to be equal to that of a globe whose diameter equals that number of inches.

313] In like manner, in what may more properly be called a Leyden vial, that is, where the glass is not flat, but convex or concave, let a circle be taken whose area is a mean between that of the inside and outside coatings, allowance being made for the spreading of the electricity. I call the square of the semidiameter of this circle, divided by twice the thickness of the glass, the computed charge of the vial. In like manner, if the real charge of any plate is found to be equal to that of a globe of x inches in diameter, I shall call its real charge x.