On this account I confined myself at first to the investigation of iron, nickel, and copper.

In the experiments with these metals the results showed not the remotest approach to regularity in respect of the quantities disintegrated. I content myself with bringing forward a few experiments and remarking only that the electrodes formed cylinders rounded at the fore end, of at most 4 or 5 grammes weight. Experiment 1.

Hydrogen developed in the voltameter, 31 cub. centims. or 2.78 milligr.

The positive electrode of iron, lost 7 milligr. iron, and took up 3 milligr. copper.

The negative electrode of copper, lost 4 milligr. copper, took up 1 milligr. iron.

Experiment 2.

Hydrogen developed in the voltameter, 20 cub. centims. = 1.78 milligr.

Positive electrode iron, lost 5 milligr. iron, took up 4 milligr. copper.

Negative electrode copper, lost 5 milligr. copper, took up 0.5 milligr. iron.

Experiment 3.

Hydrogen developed in the voltameter, 65 cub. centims. = 5.82 milligr.

Positive electrode iron, lost 36 milligr. iron, took up no copper. Negative electrode copper, lost 10 milligr. copper, took up 3 milligr. iron.

Experiment 4.

Hydrogen developed in the voltameter, 48 cub. centims. = 4.3 milligr.

Positive electrode iron, lost 6 milligr. iron, took up no copper. Negative electrode copper, lost 9 milligr. copper, took up 1 milligr. iron.

Experiment 5.

Hydrogen developed in the voltameter, 89 cub. centims. = 7.97 milligr.

Positive electrode iron, lost 9 milligr. iron, took up a trace of copper.

Negative electrode copper, lost 18 milligr. copper, took up milligr. iron,

Experiment 6.

Hydrogen developed in the voltameter, 60 cub. centims. = 5.38 milligr.

Positive electrode copper, lost 8 milligr. copper, took up 4.5 milligr. iron.

Negative electrode iron, lost 5.5 milligr. iron, took up 1 milligr. copper.

The longer of these experiments had to be performed with short intermissions, in order not to heat the electrodes to fusion. As very small quantities were in question, the chemical analysis was of course cautiously effected, and was only applied to the superficial layers of the electrodes, these being the only ones concerned. Since, in spite of the extreme rarefaction of the air in the cylinder, a few small portions on the electrodes (especially the particles of other metals precipitated upon them) were sometimes found to be oxidized, this slight content of oxygen had to be taken into account in the analyses. Although under these circumstances the numbers given present little uncertainties, yet they cannot be affected with gross errors; and therefore we have in them a satisfactorily approximate representation of the proportions.

Therefrom these proportions are perceived to be absolutely irregular. Even when the same metals are employed respectively for the positive and for the negative electrodes in different experiments, they give sometimes equal loss at both electrodes, sometimes more at the positive, and sometimes more at the negative electrode. There is likewise visible no connexion with the amount of hydrogen simultaneously developed. The only regularly occurring circumstance in these, as in all my experiments, was, that the heating of the positive electrode was always the quickest, and it was the first to become red-hot; but this circumstance, in the case of two electrodes of the same kind, is universally known.

Now, it seems, a cause can at once be adduced which contains a partial explanation of this irregularity. The experiments above given show that, although a brisk scattering of the disintegrated metallic particles in all directions always takes place, and in consequence of this a more or less dense precipitate of powder is always seen on the neighbouring sides of the cylinder, yet a not inconsiderable portion of the metal lost by one electrode may often be found on the other. Such portions of the foreign metal are firmly melted to the electrode on which they are seated. Now, supposing an experiment in which this has taken place continued, doubtless one of those portions might again be subjected to disintegration, since it is situated in the front, the part most affected by it; and thus the act of disintegration would have been employed more than once upon the same metallic particles. That such a behaviour actually played a part in the experiments here discussed is shown by several observations. For example, in experiment 1 the 3 milligr. of copper taken up by the iron electrode was not on its surface, but under a thin layer of iron,

which had evidently, therefore, come later, doubtless thrown back again from the copper electrode, as the iron electrode itself was certainly not heated to fusion. It is thus seen that an electrode, under the circumstances present in these experiments, may take up again already-disintegrated metal and subject it to a fresh disintegration. Consequently the numerical results of all such experiments are completely illusory.

More success could perhaps be expected from an arrangement in which chemical alterations of the disintegrated metal are taken advantage of to avoid repeated disintegrations; and this comes nearer to Grove's experiment. If in experiments in air the powdered metal were completely oxidized and thus presented no mechanically conducting connexion with the rest of the electrode, this purpose would perhaps have been attained. But determinations according to this method have not only the difficulty experienced by Grove, namely that oxidation readily takes place also on the mass of the electrode, but also one of an exactly opposite nature; and even this makes them as ill adapted as the previous determinations to answer the question proposed. That is to say, in general the whole of the disintegrated substance of an oxidable electrode is not actually oxidated in air, but the oxidation extends more towards the circumference of the arc of light, while pure metal readily passes in its interior.

On this I have made various experiments, and will only cite one case, which speaks distinctly. I placed two electrodes of nickel opposite each other in air, and caused the arc to pass over, during which 60 cub. centims. of hydrogen were developed in the voltameter. Both electrodes obtained an oxidized layer, which with a little care could be very well separated clean from the metal. After this the positive electrode gave a loss of weight of 4 milligr., and the negative one of 13 milligr. But the positive electrode showed in front a projecting piece of pure metal parted off from the rest of the electrode by a fine fissure. I sprung it off at the fissure, and found that it had been seated on a fully oxidated base. This piece had therefore first arrived thither during the experiment, and, indeed (since neither electrode had once become red-hot), from the opposite electrode. It was, however, oxidated only on the outermost surface; throughout its interior was pure metal: it amounted to 6 milligr.

It is thus just as little possible to measure in this way. Also an artificial blast of air or oxygen into the arc would hardly give serviceable experiments.

It may therefore certainly be maintained generally that no experiments can answer our question in which both electrodes take part in the disintegration. We must accordingly endeavour to take away the cooperation of one of the electrodes. For this

I have tried the following means, and ascertained its perfect utility.

For one electrode I took a large, thick plate of copper; for the other, in order to have a metal not oxidable in air, a knob of pure silver. The latter was screwed on an iron rod, which, wrapped round with a spiral spring, was passed through a thick caoutchouc cork, which bore three insulated supports ruuning from the sides of the rod. If the copper plate was touched with the feet of these, the silver knob could, by a screw-arrangement on the iron rod, be brought as near to the plate as was desired. It was then only necessary to press the head of the iron rod (which at the same time contained the polar wire), in order by a very brief contact between the silver knob and the plate to introduce the luminous arc. The spiral spring immediately jerked the silver knob back to its previous position.

The arc thus produced, which was maintained very perfect with a battery of 36 Grove's elements, I now carried over continually new points, by holding the caoutchouc cork in my hand. and making the three insulated supports glide over the plate. In this manner the end of the arc was always formed by a fresh point; so that, on the part of the silver that had once gone over, repeated action was not possible. With respect to the copper itself, it was to be expected that nowhere would any material disintegration take place; for the heating was distributed more uniformly over the considerable mass of the entire plate, which never became so hot that the hand could not have been held for a long time in contact with it. Indeed the result was thisthat never was the slightest trace of copper found on the silver knob, and scarcely anywhere on the plate was a layer of cupreous oxide to be perceived, which otherwise very readily forms on heated copper. The copper, then, remained substantially intact; and thus the conditions prescribed for the experiment were fulfilled.

The one-sided disintegration of the silver electrode in these experiments manifested itself in a form a little complicated. First, following the course of the arc, a stripe of bright metallic silver was sharply impressed into the copper; but this was succeeded by a rather thin layer above and on each side of it, and more or less impressed, of black powder, which, cautiously removed from the plate and analyzed, proved to be a silver-compound, probably an oxide. This powder, it seems, forms at the periphery of the arc, and consequently where the air has access. It is therefore deposited on the plate sidewards of the proper course of the arc, and of course, moreover, where at each moment the arc itself has preceded it, consequently above the stripe of metallic silver. Where, however, the arc is interrupted by sud

denly moving the silver knob away from the plate, it does not make its appearance; there metallic silver alone is to be seen. Precisely the same powder is seated, at the commencement of an experiment, on the knob itself; but it vanishes there, and does not appear again, as soon as the knob approaches a red heat, by which it is reduced to the metallic condition. It is not an improbable conjecture that this oxidation of the silver is to be ascribed to the presence of ozone in the voltaic arc produced in air; and therein might be seen a further argument for the opinion entertained probably by the majority of physiciststhat in the voltaic arc the movement of the electricity is not brought about exclusively by the transport of the substance of the electrodes, but that a partial leap of the intervening layers of gas takes place, analogous to the induction-spark.

Now the quantity of the silver that goes over in this way (that is, the disintegrated powder) can be settled by simply weighing it, since no permanent oxidation of the knob itself takes place. I obtained thus, with two silver knobs of from 5 to 6 grammes weight, the following results in the order given :—

† Compare also Grove's account (Phil. Mag. S. 4. vol. iv. p. 498) of

similar phenomena of oxidation of silver in the induction-spark.