and discharging a coated plate of glass many times without cleaning it, a narrow fringed ring of dirt could be traced all round the coating, the space between this ring and the coating being clean, and in general about inch broad.

He also observed that the flash of light was stronger the first or second times of charging a plate than afterwards.

To determine how much the capacity of a coated plate was increased by this spreading of the electricity, he compared the capacity of a plate with a circular coating with that of the same plate with a new coating of nearly the same area, but cut into strips, so that its perimeter was very much greater than that of the circular coating.

In this way he found that if we suppose a strip of uniform breadth added to the coating all round its boundary, the capacity of this coating, supposing the electricity not to spread, will be equal to that of the actual coating as increased by the spreading of the electricity. The most probable breadth of this strip he found to be 0.07 inch for thick glass and 0·09 for thin.

When this correction was applied to the areas of the coatings of the different coated plates, the computed charges of plates made of the same kind of glass were found to be very nearly in the same ratio as their observed charges.

But the observed charges of coated plates were found to be always several times greater than the charges computed from their thickness and the area of their coatings, the ratio of the observed charge to the computed charge being for plate glass about 8.2, for crown glass about 8.5, for shellac about 447, and for bees' wax about 35. Thus Cavendish not only anticipated Faraday's discovery of the Specific Inductive Capacity of different substances, but measured its numerical value in several substances.

The values of the specific inductive capacity of various substances as determined by different modern observers are compared with those found by Cavendish in the table in Note 27.

To make it certain, however, that the difference between the observed and calculated capacities of coated plates really arose from the nature of the plate and not from some error in the theory,

Cavendish determined the capacity of a "plate of air," that is to say a condenser consisting of two circles of tinfoil on glass with air between them. The capacity of a plate of air was found to be much less than that of a plate of glass or of wax of the same dimensions, but it seemed to be about in excess of the calculated value. This discrepancy will be discussed in Note 17.

These may be considered the principal results of the investigations with coated plates, but the following list of collateral experimental researches will show how thoroughly Cavendish went to work.

A question of fundamental importance in the theory of dielectrics is whether the electric induction is strictly proportional to the electromotive force which produces it, or in other words, is the capacity of a condenser made of glass or any other dielectric the same for high and for low potentials?

The form in which Cavendish stated this question was as follows:-"Whether the charge of a coated plate bears the same proportion to that of a simple conductor, whether the electrification is strong or weak."

Cavendish, who explained the fact that the capacity of a glass plate is greater than that of an air plate, by supposing that the electricity is free to move within certain portions of the glass, supposed that when the plate was more strongly electrified the electricity would be able to penetrate further into the glass, and that therefore its charge would be greater in proportion to that of a simple conductor or a plate of air the stronger the degree of electrification.

But according to the experiments he made to answer this question† a coated plate and a simple conductor whose charges were equal for the usual degree of electrification remained sensibly equal for higher and lower degrees, and if, as appeared probable from the experiments on the spreading of electricity at the edge of the coating, this spreading extended further for high degrees of electrification than for low, it would be necessary to admit that the charge of a glass plate became less in proportion to that of a simple conductor as the degree of electrification increased. Cavendish, however, concluded that the experiments were hardly accurate

enough to warrant the deduction from them of so improbable a conclusion,

He also found that the result of the comparison of a coated plate and a simple conductor was the same whether they were charged positively or negatively.

He tried whether the capacity of a plate of rosin altered with the temperature, but he could not find that it did*. In glass he found that the capacity increased as the temperature rose, but the most decided increase did not occur till the glass began to conduct somewhat freely. Cavendish therefore does not consider the experiment quite decisivet.

He found that the apparent capacity of a Florence flask was greater when it continued charged a good while than when it was charged and discharged immediately, and he found that the same was the case with a coated globe of glass. This phenomenon, which Faraday called "electric absorption," has recently been carefully studied in different kinds of glass by Dr Hopkinson§. It is connected with the long-known phenomenon of the "residual charge," and the existence of such phenomena in many dielectrics renders it difficult to obtain consistent values of their inductive capacities; for the more rapidly the charging and discharging is effected the lower is the apparent value of the capacity. It is for this reason that condensers of glass cannot be used as standards of capacity when accurate measurements are desired.

Franklin had shown || that the charge of a glass condenser resides in the glass and not in the coatings, for when the coatings were removed they were found to be without charge, and when new coatings were put in their place the condenser thus reconstructed was found to be charged.

Cavendish tried whether this was the case with a chargea plate of air, by lifting one of the electrodes and changing the air between them and then replacing the electrode. He found that the charge was not altered during these operations, and concluded that the charge resides, not in the air, but in the metal plates.

* Art. 523.

+ Art. 366.

§ Phil. Trans. 1877, p. 599.

Franklin's Works, ed. Sparks, Vol. v., p. 201.

Art. 523.

In Arts. 336 to 339 we find a most ingenious method of determining by experiment the effect of the floor, walls and ceiling of a room, and of other surrounding objects, in increasing the apparent capacity of a conductor placed in a given position in the room. The method consists in measuring the capacities of two conductors of the same shape but of different dimensions, the centre of each being at the given point in the room. If the experiment had been made with the conductors at an infinite distance from all other bodies their capacities would have been in the ratio of their corresponding dimensions, but the effect of surrounding objects is to make their capacities vary in a higher ratio than that of their dimensions, and from the measured ratio of the two capacities, the correction for the effect of surrounding objects on the capacity of any small body may be calculated.

Cavendish also verified by experiment what he had already proved theoretically, that the capacity of two condensers is not sensibly altered when they are placed near to each other or far apart.

But besides this series of experiments on electric capacity, another course of experiments on electric resistance was going on between 1773 and 1781, the knowledge of which seems never to have been communicated to the world.

In his paper on the Torpedo in the Philosophical Transactions for 1776 (Art. 398) he alludes to "some experiments of which I "propose shortly to lay an account before this Society," but he never followed up this proposal by divulging the method by which he obtained the results which he proceeds to state-"that iron "wire conducts about 400 million times better than rain or dis"tilled water*," and that "sea water, or a solution of one part of "salt in 30 of water conducts 100 times, and a saturated solution "of sea-salt about 720 times better than rain water."

Such was the reputation of Cavendish for scientific accuracy, that these bare statements seem to have been accepted at once,

* This is equivalent to saying that iron wire conducts 555,555 times better than saturated solution of sea salt. A comparison of the experiments of Matthiessen on iron with those of Kohlrausch on solutions of sodium chloride at 18oC. would make the ratio 451,390. The resistance of iron increases and that of the solution diminishes as the temperature rises, and at a temperature of about 11°C. the ratio of the resistances would agree with that given by Cavendish.

and soon found their way into the general stock of scientific information, although no one, as far as I can make out, has ever conjectured by what method Cavendish actually obtained them, more than forty years before the invention of the galvanometer, the only instrument by which any one else has ever been able to compare electric resistances,

We learn from the manuscripts now first published, that Cavendish was his own galvanometer. In order to compare the intensity of currents he caused them to pass through his own body, and by comparing the intensity of the sensations he felt in his wrist and elbows, he estimated which of the two shocks was the more powerful.

As Cavendish does not appear to have prepared an account of these experiments in the manner in which he usually wrote out what he intended to publish, it may be well to describe them here, as we collect them from different parts of his Journals.

The conductors to be compared were for the most part solutions of common_salt of known strength or of other substances. These solutions were placed in glass tubes, more than a yard long, bent near one end. The tubes had been previously calibrated by means of mercury.

Two wires were run into the tube, probably through holes in corks at each end, to serve as electrodes. The length of the effective column of the liquid could be altered by sliding the wire in the straight part of the tube.

In order to send electric discharges of equal quantity and equal electromotive force through two different tubes Cavendish chose six jars of nearly equal capacity from "Nairne's last battery." The two tubes to be compared were placed so that the wires run into their bent ends communicated with the outside of this battery of six jars. The wires run into the straight ends of the tubes were fastened to two separately insulated pieces of tinfoil. The six jars were then all charged at once by the same conductor till the gauge electrometer indicated the proper degree of electrification. The conductor was then removed, so that the six jars remained with their inside coatings insulated from each other and equally charged.