In trying the experiments, the jars, and consequently the wire Cc, are charged, the wire Cc being all that time lifted up as high as it will go by means of the counterpoise. When the jars are charged to the proper degree as shown by the electrometer, the wire Cc is let down on the wires Aa and Bb by lifting up the counterpoise. This instantly charges both the coated plates, for when Cc rests on Aa and Bb, and consequently Ff and Gg rest on Ee and Dd, the lower coatings of both plates communicate with the ground, and their upper coatings with Cc.

Immediately after this the counterpoise is let go, by which means Cc is lifted up, and Gg and Ff along with it, till the two last mentioned wires bear against Aa and Bb, so that immediately after the coated plates are charged, the communication between them and the wire Cc, by which they were electrified, is taken away, and at the same time the communication between the lower coating of the trial plate and the ground is taken away, and immediately after that a communication is made between the upper coating of the plate B and the lower coating of the trial plate, and also a communication is made between the upper surface of the trial plate and the ground, so that the upper coating of the trial plate and the lower coating of the plate B both communicate with the ground, and the upper coating of B and the lower coating of the trial plate communicate with each other and the wire Dd.

Consequently, if the quantity of redundant fluid communicated to the wires Bb and Mm and the upper side of the plate B together is equal to the deficient fluid on the under side of the trial plate, they and the wire Dd will be neither over nor undercharged after the operation is completed; but if the redundant fluid in them exceeds the deficient fluid on the lower side of the trial plate, Dd will be overcharged, and the pith balls will separate positively. On the other hand, if it is less than the deficient fluid, the pith balls will separate negatively.

297] The trial plate consisted of a flat plate of glass, or other electric substance, the lower surface of which was coated all over with tinfoil, but on the upper side there was only a small coating of tinfoil. I had also flat plates of brass of different sizes which I could lay on the upper surface, and slip backwards and forwards, and thereby increase or diminish the size of the upper coating at pleasure, for the area of the upper coating is equal to the area of

the plate of brass added to that of so much of the tinfoil as is left uncovered by the brass *.

By this means I could increase or diminish the quantity of deficient fluid on the lower side of the trial plate at pleasure, for I could alter the size of the upper coating at pleasure, and the quantity of deficient fluid on the under side of the plate is not much greater than it would be if the lower coating was no greater than the upper, and consequently depends on the size of that upper coating.

As it is necessary that the trial plate should be insulated, it was not laid immediately on the bars Nn and Pp, but was supported by sticks of waxed glass fastened to those bars.

Having by these means found what size it was necessary to give to the upper coating of the trial plate in order that the pith balls should separate positively just sensibly, and what size it was necessary to give to it that they might separate as much negatively, I removed the plate B and placed the plate or plates which I intended to compare with it (or the plate b as I shall call it) in its room and repeated the experiment in just the same manner as before. Then, if I found that the size which it was necessary to give to the upper coating of the trial plate in order to exhibit the same phenomena was the same as before, I concluded that the charge of the plate b was the same as that of B. If, on the other hand, I found that it was necessary to make the area of the upper coating of the trial plate greater or less than before in any ratio, I concluded that the charge of b was greater or less than that of B in the same ratio, for the quantity of deficient fluid on the lower side of the trial plate will be pretty nearly in proportion to the area of the upper coating.

N.B. In the following experiments it was always contrived so that the charges of the plates to be compared together should be pretty nearly alike, so that if the quantity of deficient fluid on the lower surface of the trial plate was not exactly in this proportion, it would make very little error in the proportion of the charges.

298] The method above described is that which I made use of in my first experiment, but I afterwards made use of another

N.B. In order to estimate how much of the tinfoil was left uncovered, I drew parallel lines upon it at small equal intervals from each other, and took notice which of these lines the edge of the brass plate stood at. [Arts. 442, 488.]

method a little different from this, and which I found more exact, though rather more complicated, namely, for each set of plates that I wanted to compare together I prepared two trial plates, which I shall call Z and l, not coated as that above described, but in the usual way, namely, with the coatings of the same size on both sides*.

The first of these plates, or L, was of such a size that when used as a trial plate with the plate B or b on the other side, the quantity of deficient fluid in it was rather more than ought to be in order that the pith balls should just separate negatively, and the second plate was rather greater than it ought to be in order that they should just separate positively.

I also prepared a sliding plate of the same kind as the trial plate used in the former method, but whose charge was many times less than that of the plate B or b. This sliding plate I placed along with the plate B or b on the side N, and on the other side I placed the trial plate L and found what size it was necessary to give to the coating of the sliding plate in order that the balls should just separate negatively. I then removed the plate B and put b in its room, and found what sized coating it was necessary to give to the sliding plate in order that the balls should separate the same as before. Having done this, I removed the trial plate L and put in its room, and tried each of the plates B and b as before, finding what coating it was necessary to give to the sliding plate that the balls might just separate positively.

Having done this, if I found that it required the coating of the sliding plate to be of the same size in order to exhibit the same phenomena in trying the plate B as in trying b, it is plain that the charges of B and b must be both alike, but if I found that it was necessary to give less surface, one square inch for instance, to the coating of the sliding plate in trying B than in trying b, then it is plain that the charge of B exceeds that of b by a quantity equal to that of the charge of the sliding plate when its surface is one square inch, supposing, as is very nearly the case, that the charge of the sliding plate is in proportion to the surface of its upper coating.

In this way of trying the experiment, it is plain that, in order to determine the proportion which the charges of B and b bear to each other, we must first know what proportion the charge of the

[Art. 457.]

sliding plate, when its coating is of a given size, bears to that of B. This I found by finding what sized coating must be given to the sliding plate that its charge should be equal to that of another plate, the proportion of whose charge to that of B I was acquainted with.

It is plain that, if it is necessary to give one inch less surface to the coating of the sliding plate in trying B than in trying b when the trial plate L is made use of, it will be necessary to make the same difference in the surface of the sliding plate when the trial plate l is made use of, so that I might have saved the trouble of making two trial plates. However, for the sake of more accuracy, I always chose to make two trial plates and to take the mean of the results obtained by means of each trial plate for the true result.

299] One reason why this method of trying the experiment is more exact than the former, or that by means of a sliding plate only, is that in the former method I was liable to some error from inaccuracy in judging how much of the tinfoil coating of the trial plate was left uncovered by the sliding brass plate, whereas in this method, as the charge of the sliding plate is but small in respect of that of B, it was not necessary to be accurate in estimating its surface. But I believe the principal reason is that an error which will be taken notice of by and by, and which proceeds from the spreading of the electricity on the surface of the glass, is greater in a sliding plate than in one coated in the usual manner.

In general I think it required scarcely so great an increase of the charge of the trial plate to make a sensible alteration in the degree of separation of the pith balls in the following experiments as in the preceding, and therefore it should seem as if these experiments were capable of rather more exactness than the former, but this was not the case, as the different trials were found not to agree together with quite so much exactness in these experiments as the preceding. For this reason, and also because they were attended with less trouble, I repeated the experiments oftener, as I not only compared each plate with the trial plate for more times together as I did in the preceding experiments, but in general I repeated the experiment on several different days.

300] The circumstance which gave me the most trouble in these experiments was the spreading of the electricity on the sur

face of the glass. To understand this, let A Bab, Fig. 21, be a flat plate of coated glass, cd and CD being the two coatings, and let CD be positively electrified, and let cd communicate with the ground.

Fig. 21.

D

It is plain that the electric fluid will flow gradually from CD and spread itself all round on the surface of the glass, and nearly the same quantity of fluid will flow from the opposite side of the glass into cd, so that those parts of the glass which are not coated gradually become charged, those parts becoming so soonest which are nearest the edge of the glass.

On discharging the plate the uncoated part of the glass gradually discharges itself, as on the side AB the fluid will flow gradually from the uncoated part of the glass into CD, and on the opposite side it will flow into the uncoated part of the glass from cd.

301] There is a great deal of difference in this respect between different kinds of glass, as on some kinds it spreads many times faster than on others. The glass on which it spreads the fastest of any I have tried is a thin kind of plate-glass, of a greenish colour, much like that of crown-glass, and which I have been told is brought from Nuremberg*. On the English plate-glass it does not spread near so fast, but there is a great deal of difference in that respect between different pieces. On the crown-glass it spreads not so fast as on the Nuremberg, but I think faster than on the generality of English plate-glass. On white glass I think it spreads as slowly as any.

302] The way in which I compared the velocity with which it spread on different plates was as followst. I took away the wire Ff (Fig. 20) and placed the plate which I wanted to try where the plates L or I used to be placed, the lower coating communicating as usual with Dd by the wire 8, but the wire ß being drawn up by a silk string so as not to touch the upper coating. The wire Cc is suffered to rest on Aa and the jars electrified.

*

[Art. 497.]

+ [See Arts. 485, 486, 487. Also 494 to 499.]