dry wood and many minerals; (2) pores through which gases do not pass under pressure, but pass by their proper molecular movement of diffusion as in artificial graphite; and (3) pores through which gases pass neither by capillary transpiration nor by their proper diffusive movement, but only after liquefaction, such as the pores of wrought metals and the finest pores of graphite." It is noteworthy that Graham considers that gases are liquefied in the pores of metals. But in the experiments here described this can hardly be the case. For at 270° and at higher temperatures hydrides of palladium do not exist, as is conclusively shown by Troost and Hautefeuille ; just as water does not exist in superheated steam. And yet it is at these temperatures that palladium is permeable to hydrogen, and not at temperatures at which hydride of palladium is stable. There are several facts which must be borne in mind in seeking for an interpretation for the phenomenon of the passage of hydrogen through palladium. First. The hydrogen in the act of passing is a reducing agent, as shown by its behaviour towards the oxides of nitrogen. At such temperatures as were here employed, hydrogen is without action on the lower oxides of nitrogen. Second. Bellati and Lussano have shown (Atti R. Ist. Ven. i. series vii. p. 1173) that hydrogen "diffuses" through an iron plate which is used as a negative electrode on electrolysing dilute sulphuric acid. Their observation has been confirmed and amplified by Shields (Chem. News, lxv. p. 195). Shields has shown that neither lead nor platinum nor palladium allow hydrogen to pass under similar circumstances, and experiments made by myself show that nickel does not allow carbon monoxide to pass at temperatures at which that compound is stable. Third. Iron and platinum, as shown by Deville and Debray (Compt. rend. lvii. p. 965), are permeable at a red heat to hydrogen. I think these considerations prove that it is necessary to add a fourth class to the three classes suggested by Graham. Graham's first class involves actual holes, that is, passages large in comparison with the molecular diameter; his second implies pores small compared with molecular diameter, but still greatly exceeding that diameter; his third class would be termed "solid solution" in the present state of our knowledge, i. e. when coal-gas passes through india-rubber, the latter dissolves the gas on the side exposed to it, while the gas evaporates from the other side, so as to render the pressure of the dissolved gas equal on both sides of the membrane. The case is precisely analogous to the passage of water through a semi-permeable diaphragm. But, in order that hydrogen may pass through iron, it must be in the state in which it is liberated by an electric current, or it must be hot. That hydrogen will not pass through palladium at the ordinary temperature appears to show that the compound of palladium and hydrogen has practically no dissociation pressure at ordinary temperature; otherwise the hydrogen would pass by solution, in the same manner as coal-gas passes through india-rubber, or water through a semi-permeable diaphragm. That it will not pass, even when liberated by an electric current on one side of the palladium membrane, shows that it at once enters into combination with the palladium, and is no longer in statu nascendi, to use a generally understood expression which is independent of theory. But that it passes through hot palladium appears to show that it is then in a state analogous to that of electrically liberated hydrogen. it It is known that electrified bodies are discharged if a flame burns in their vicinity. This may be attributed to the liberation of atomic oxygen in a kind of Grotthus chain. For may be imagined that when a molecule of oxygen encounters the oxidizable matter of a flame, it is dissociated: while one atom serves to oxidize the carbon, the other exchanges with a neighbouring molecule, and a succession of exchanges occur till the atomic oxygen near the electrified body receives or communicates a charge, and restores the potential of the charged body to that of surrounding objects. To ascertain whether a flame of oxygen burning in hydrogen would similarly cause the hydrogen to assume the atomic state, a piece of apparatus was contrived in which such a flame burned in close proximity to a very thin iron plate, on the other side of which a Torricellian vacuum was maintained. No hydrogen passed through the plate: hence either the hydrogen was not atomically transferred, as oxygen is supposed to be under similar circumstances, or such atoms were unable to pass through the plate as they would have done if liberated electrically. It must be noticed, however, that it is conceivable that the double atom of hydrogen which we term a molecule may have united directly with the oxygen, without separating into its two components. The result of this experiment is therefore inconclusive. It appears to me necessary to suppose that at a temperature far above that at which hydride of palladium is capable of existence, the palladium has still the power of so attracting the hydrogen that the molecule is split. This necessarily implies a gain of energy, so far as the splitting of the hydrogen molecule is concerned, for any energy lost by the temporary and transient union of hydrogen and palladium is at once gained during its escape on the other side of the partition. But the hydrogen in expanding, which it does on passing through the partition, loses energy, and hence, on the whole, energy will probably be lost during the process. It is to such theories, I think, that we must look to explain the passage of hydrogen through a palladium diaphragm. (2) In answering the question why the pressure raised by the entering hydrogen is never equal to that of the atmosphere, I think it must be admitted that the gas contained in the palladium vessel is not without influence on the passage of the hydrogen. A diminution of the pressure of the external hydrogen by the addition of nitrogen considerably increases the partial pressure of the internal hydrogen. Here the action of external nitrogen apparently neutralizes partially the effect of the internal nitrogen, and more hydrogen penetrates the metallic diaphragm. With gases other than nitrogen in the interior, the pressure of the hydrogen becomes more nearly equal to that on the exterior. The constancy of the results, however, proves that the deficiency is not due to experimental error. This whole subject is full of difficulty. Experiments are in progress on the absorption of gases by platinum, and on the passage of gases through other metallic diaphragms, which may ultimately render an explanation possible. But I have thought it desirable to place these experiments on record, incomplete as they are, rather than wait for a complete solution to the problem. I cannot conclude without acknowledging the able manner in which my late assistant, Mr. Percy Williams, has aided me in carrying out these experiments. XXIII. Note on the Rigidity of the Earth. ROF. NEWCOMB† has estimated the Rigidity of the Earth from the observed 427 days' ‡ period of the Variations of Latitude, and found it to be somewhat greater than that of steel. His estimation being a rough one, I have undertaken a more precise calculation with the help of the formulas of Thomson and Tait. * Communicated by the Author. † Monthly Notices Astron. Soc. 1892, pp. 336–341. ‡ The recent investigations of the astronomers of Poulkova confirm also this period. Prof. Newcomb begins with the remark that the condition for the possibility of the 427 days' period is, that the pole of the principal axis shall be at a distance from the North Pole [that is from the point which would be the pole of rotation and inertia if the perturbation did not exist] equal to of the distance between the pole of rotation and the North Pole. (See fig. 1.) Thus the pole of rotation R revolves in nearly 305 days about the pole of inertia P', but this latter and also R revolve about the North Pole in 427 days. The little circle is rolling on the greater. The precise statement of the problem is the following The equations of motion relatively to axes revolving with angular velocities P, q,'r are of course : dq =X. p[ d − 2r dy + 24 de − z(q2 + r2) +v (pq− dr) + 2(d; +pr)]=> P •[ dt2 dt dz dt y dt Suppose that the body is elastic, that xo, yo, zo are the coordinates of particles belonging to their undisturbed positions, suppose that §, n, are the displacements, suppose that the revolving axes are fixed relatively to the coordinates xo, yo, 20, then xo, yo, zo are constants, and as Further, neglecting §, n, in comparison with vo, Yo, Zo, and writing again for xo, Yo, zo, x, y, z, we have the equa The precise integration of these equations is impossible, but we may avail ourselves of the circumstance that in the present problem the quantities are quite insignificant in comparison to other terms. We being the components of a rotation superimposed on the strain, shall give no deformation of the body, and reduce the equations to the form: p[−x(q2+r2)+ypq+zpr]=X, But if a, b, c are the direction cosines of the axis R, then p=aw, c=rw, where is the angular velocity about the axis R, which in the present problem is evidently constant. Further, the forces X contain the attraction of the particles and the components of elastic forces. So we may write our equations in the form |