discoverer. Hittorf measured the fall of potential in a vacuum tube, not merely between the electrodes but also along the whole length of tube, and found the very remarkable and interesting result that by far the greater part of the total fall in potential occurred close to the cathode. Thus in one of his experiments on hydrogen at a pressure of about 1th of a millimetre of mercury he found that, of the total fall in potential, about occurred close to the cathode, the potential gradient in the rest of the tube only amounting to two or three volts per centimetre. The question. immediately suggests itself whether, if we got rid of the electrodes altogether, we should reduce the potential difference to one-sixth or so of the value required when electrodes are used. I have here an experiment which is intended to settle this point; the apparatus consists of two bulbs connected together by an open tube, so that the bulbs are always filled with the same kind of gas at the same pressure. One bulb is without electrodes ; the other, whose diameter is approximately equal to the circumference of the first, is provided with electrodes which are placed at the opposite ends of a diameter; these electrodes are connected with a wire which makes one turn round the coil which connects the outsides of the two Leyden jars (fig. 1); the bulb without electrodes is placed inside this coil; the total electro-motive force acting round the bulb without electrodes is thus approximately the same as that acting between the electrodes of the other bulb. Setting the Wimshurst machine in action, and gradually increasing the length of the spark until a spark passes, it is found that the discharge begins to appear at about the same time in each of the bulbs, showing that the total electro-motive force required to produce discharge is not very different in the two cases, and that the potential difference required to start a discharge through a given length of gas is not very greatly increased by the presence of electrodes. We may, therefore, conclude that the potential differences measured in the tube with electrodes are primarily connected with work required to split up the gas through which the discharge passes.

RESISTANCE OF RAREFIED GASES.

Rarefied gases are exceedingly good conductors of electricity when they are acted upon by electro-motive forces sufficiently intense to produce discharges. This is clearly shown by the following experiment. A and B (fig. 2) are two coils in series placed in circuit with the outer coatings of two Leyden jars. In coil A an exhausted bulb is placed; this bulb serves as a kind of galvanometer, the brightness of the ring in it giving an indication of the current passing through the coil A. The substance whose resistance is to be tested is placed in a bulb inside the other coil; the currents induced in this bulb will, by their inductive action, exert on the primary coil an electro-motive force in the opposite direction to the current in the coil; this will tend to stop the current, and we shall detect its effect by the diminution in brightness of the discharge in the bulb inside A. The extent of this diminution will give us a clue to the magnitude of the currents induced in the bulb B (fig. 2). I place inside the coil B a bulb containing gas at a low pressure. You notice that the discharge in A is quite extinguished. I now replace this bulb by one of the same size filled with sulphuric acid and water in the proportions for which they conduct electricity best. You observe that the sulphuric acid in B does not diminish the brilliancy of the discharge in A to anything like the extent the exhausted gas did; thus the currents passing through the gas are

larger than those through the acid. If we compare the number of molecules of the gas with the number of molecules of sulphuric acid in the same volume, we find that the molecular conductivity' of the gas must be many million times that of the sulphuric acid (see J. J. Thomson, 'Recent Researches on Electricity and Magnetism,' p. 101). A calculation of the intensity of the current through the gas shows that some hundreds of amperes must be passing through a square centimetre of the gas, a greater current than is allowed by the Board of Trade rules to pass through the best conducting electric light leads.

The presence of a very small number of charged ions in a gas will impart to it a conductivity large enough to be detected by the method just described. As this method of detecting the existence of free ions may perhaps be of some service to Chemistry, it may be worth while to calculate from the principles of the Kinetic Theory of Gases the conductivity of a mixture of free ions and undissociated gas. For the sake of simplicity I will take the case when the ions form but a small fraction of the undissociated gas. Let e be the charge of electricity on the positive atom, m, the mass of this ion, -e the charge on a negative ion, m, its mass, N the number of positive or of negative ions per unit volume, X the electric intensity parallel to the axis of x, u, u, the mean translatory velocity parallel to

the axis of x of the positive and negative ions respectively. We shall suppose that the undissociated gas has no mean movement. Thus (Art. Diffusion, Encyclopædia Britannica,' or Maxwell's Collected Papers,' vol. ii. p. 629) we have, if p is the density of the undissociated gas,

where G2, G3 are constants depending on the size of the molecules and the temperature, p, is the pressure due to the positive ions.

ρ

If p is large compared with Nm, and if the state of the gas keeps uniform as we move parallel to the axis of x, this equation reduces to

We shall first investigate the case when the electric intensity is constant; then when things are in a steady state du,/dt vanishes, and we have

Xe 1
m Gap

Now, if D12 is the coefficient of interdiffusion between the positive ions and the undissociated gas, then (Maxwell, l.c., p. 631)

12

k k、 D12 = G2P

When P is the total pressure due to the ions and undissociated gas, k and k, are respectively the quotients of the pressure by the density for the undissociated gas and the positive ions. Now, since the ions form but a small part of the gas, the total pressure is practically equal to the pressure of the undissociated gas; hence we may put p=kp, so that

and therefore

D12 = G2p
k1

Let us take the case of hydrogen ions, for which e/mk1 =10-6 approximately. We do not know the coefficient of interdiffusion between hydrogen atoms and molecules; it will, however, be greater than that between oxygen and hydrogen, which was found by Loschmidt to be equal at atmospheric pressure to 72. Hence, if in the equation for u we put D12=7, we shall get a value of u less than the true one. Substituting this value for D12, we find

u1=7x10~7X

If the electric intensity is a volt per centimetre, X=10%. In this case u1=70; hence for each volt per centimetre we get a velocity of hydrogen ions equal to 70 cm/sec. With the same electrical intensity the velocity will be inversely proportional to the pressure of the undissociated gas, so that when the pressure of this is Too of the atmospheric pressure the velocity of hydrogen ions moving through it will be 70,000 cm/sec. The current carried by the positive ions is

or at atmospheric pressure

1000

Neu

7 x 10-1NeX

Since e/m=10' for hydrogen, we may put for the positive current

7 x 10-9Nm, X

To get an idea of the magnitude of the resistance, let us assume that the

current carried by the negative ions is equal to that carried by the positive. Thus, if q be the current, we have

q=1.4 × 10-2Nm, X

Thus the specific resistance of the gas is

102
1.4 x Nm,

Suppose that the hydrogen ions gave rise to a pressure of x atmospheres, then

Nm, xx 10

=

so that the specific resistance is

approximately,

10°/1·4 × x

Now, from experiments with electrolytes we find that we can easily detect by this method the conductivity of substances whose specific resistance is 1010; hence we could detect the conductivity of the gas even though a were as small as 1/7000; that is, we could easily detect the presence of free ions though they only amount to one part in 7000 of the total gas. It is important to notice that, inasmuch as the conductivity varies inversely as the pressure of the undissociated gas, we should be able to detect the existence of the same percentage of free ions at all pressures. Let us now consider the case when the electric intensity is variable. Suppose that it is proportional to cos pt, say X=X, cos pt; then we have

When the alterations in the electric intensity are so slow that p is small compared with Gap, the solution is practically of the same form as when the electric intensity is steady. But when the oscillations are so rapid that p is large compared with G2p, then approximately

Xo sin pt Р

and the maximum velocity is independent of the pressure. In this case the direction of the electric intensity gets reversed many times in the interval between two collisions of the ion; thus the ions, when they have acquired a high velocity under the electric intensity, do not, as in the case when the electric intensity is steady, lose their energy by impact against other molecules, and so raise the temperature of the surrounding gas; when p is very large, the force is reversed before the ions collide, and the velocity of the ion gets reduced by the action of the electric force. There is in this case very little heat-production; the effect of the free ions is rather to alter the self-induction of the circuit than its resistance. Thus, if a light

wave were passing through a medium with a small number of free ions, the effect of these ions would be rather to affect the velocity of propagation than to produce any great absorption. In the case of hydrogen at atmospheric pressures we have seen that Go is of the order 1010; in this case p would have to be larger than 100 to make the effects depending on du, [dt large compared with those depending on Gapu. We could not by discharging Leyden jars get electrical vibrations of this rapidity, but at the pressure of Too of an atmosphere G20 would only be of the order 10%, and we could easily get electrical vibrations sufficiently rapid to make p large compared with this quantity, and thus to make the effects depend chiefly upon the term du/dt, that is, upon the inertia of the ions.

The preceding experiments are, I think, sufficient to show the close analogies existing between the phenomena of chemical combination and of the electric discharge, and give hopes that the study of the passage of electricity through gases may be the means of throwing light on the mechanism of chemical combination. The work of chemists and physicists may be compared to that of two sets of engineers boring a tunnel from opposite ends - they have not met yet, but they have got so near together that they can hear the sounds of each other's works and appreciate the importance of each other's advances.

On the Electrification of Molecules and Chemical Change.
By H. BRERETON Baker.

[Ordered by the General Committee to be printed in extenso.]

MORE than twenty years ago a striking fact was discovered by Dr. Wanklyn, that dried sodium could be melted in dried chlorine without the production of the bright flame usual under the circumstances. The action of chlorine on other metals in absence of moisture was investigated by Dr. Cowper in 1876. He showed that in many cases the same result was obtained as that of Dr. Wanklyn in the case of sodium. About this time Professor Dixon, who was working on the rate of chemical change in a mixture of carbon monoxide, hydrogen, and oxygen, was led to suspect the great influence of the presence of moisture on the combustion of the former gas, and he succeeded, by drying a mixture of carbon monoxide and oxygen as completely as possible, in passing a stream of electric sparks in the mixture without any explosion taking place. It was this experiment which first led to the great interest taken by chemists in the influence of moisture on chemical action. Many chemists have investigated different chemical actions, and a large number of changes have been shown to be dependent on the presence of moisture. A list of them will be found in a paper on this subject in the 'Chemical Society's Journal' of July last. It seemed at one time to a chemist who was studying these actions, that no chemical action could take place without the presence of moisture. As action after action was investigated, and as new methods of purification were introduced, further additions could be made to the list. I have recently been engaged in studying several decompositions, however, and I believe that although, in some cases, no breaking up of the molecules takes place, as in the very interesting case of the action of heat on dried ammonium chloride, in which no dissociation occurs, yet in some cases action does take place. Potassium chlorate and silver oxide do decompose, and give not atomic but molecular oxygen. Carbon bisulphide burns