If a very intense ray bundle (consisting of negative electrons) pass through the positively-charged layer, and if it keep its position for some short time, the gas on its way may for a moment become partly neutralized, and the N spectrum may partly turn into, the ordinary type corresponding to neutral nitrogen and which gives the red colour.

It is of interest to notice that the latter explanation of colour changes would require negatively-charged cosmic rays. In fact, the existence of a positively-charged upper layer of gas would involve that, at any rate, all aurore showing a ray structure must be produced by rays of negative electrons. For, as we saw, the electrical charge of the layer would increase the density of matter throughout the auroral region. The mass of gas in a cylinder of unit cross-section reaching from a height H to infinity will be

not hold for an electrically-charged layer. Integrating the equation (8), we get

This equation shows that the mass my which a ray has to penetrate to get down to a height H is now greater than that calculated for neutral nitrogen (Table IV.); and if the charge is to produce a sufficient density of nitrogen 600 km. above the ground, the mass m at the height of 100 km., say, must be very much greater than that found for neutral nitrogen.

Any electric ray with a carrier of atomic dimensions which had a penetrating power great enough to enable the ray to get down to a height of 100 km., would possess a magnetic deflectibility, which would be too small to explain the narrow ray streamers.

Only electron rays combine a sufficient penetrating power with a sufficient magnetic deflectibility to explain the height and structure of the aurora *.

I

Physical Institute, University,
Christiania,

March, 1st, 1923.

XX. The Theory of the Abnormal Cathode Fall.

To the Editors of the Philosophical Magazine. GENTLEMEN,

HAVE read with great interest a paper on the above subject published in this journal last month by the Research Staff of the General Electric Company whom I shall, for the convenience of this letter, identify with Mr. Ryde as his name is associated with the work. In consideration of the fact that the theory put forward is based almost entirely upon measurements made by me, I feel justified in making a few comments to caution the reader that its novelty of treatment and agreement with experiment are more apparent than real.

Under the more usual title of "Dark Space" I studied this fascinating phenomenon continuously for ten years, and I am entirely unable to accept Mr. Ryde's view that this simple theory is adequate even for a preliminary survey of the facts. To mention two minor points at the outset, I cannot see how a theory depending solely on the mass of the positive ions can give two entirely different results in the cases of N, and CO since we expect, and positive ray analysis proves, that the mean mass of the positive ions is practically identical in both. Again, with the definite stated conditions under which all my measurements were made, I never found the slightest evidence of a measurable potential

Note added to the Proof.-Since I wrote this paper observations have been made with regard to the auroral spectra emitted at various altitudes, and investigations have been continued regarding the constitution of the upper strata of the atmosphere.

These investigations have led to the view that a highly ionized upper layer cannot exist in the form of gas--but we must assume the charge mainly attached to clusters-or small crystals of nitrogen. These investigations will be dealt with in a second paper.

At present I should only like to point out that in an upper layer formed by dust particles more or less electrically charged we can no longer apply the gas equations (1), (7), and (8). The conclusions based on the assumption that these equations hold for the auroral region therefore ought to be reconsidered.

difference between the metal of the anode and the edge of the negative glow. If such exists, direct experiments show that it cannot be more than a few volts, which will not go far to explain the discrepancy of about 130 volts shown by Mr. Ryde's parallel curves.

In the first paper I published on the subject in 1907 I gave a simple theory, which Mr. Ryde appears to have overlooked, founded on precisely the same premises as his, namely, that the boundary of the negative glow could be regarded as a highly conducting plane source of positive ions and that the density of electrons in the dark space could be neglected in comparison with that of the positive ions. The only dif ference between my theory and the one put forward is that I took for the velocity of the positive ions the expression for their mobility at ordinary pressures, whereas Mr. Ryde takes that of the free fall in vacuo. When I formulated my theory I had the choice of either of these expressions, but considered the latter certain to give too high a result. In this choice I have since been justified, for the difficulty is in making the velocity low enough to fit the facts. My theory required id3V-2 to vary inversely as the pressure, which agreed very well with the numerical results obtained, but I abandoned it without any hesitation on ascertaining the law of distribution of potential in the dark space in 1911. The method by which the latter very simple but utterly baffling result was then demonstrated appears entirely free from objection. I have no more reason to doubt its substantial accuracy now than I had when I made the measurements, so that I find myself unable to consider very seriously any theory which does not attempt its explanation.

Equally formidable obstacles to the formulation of a workable theory are raised in my paper on the effect of the material of the cathode, where it is shown that a silver cathode gives twice the length of dark space given by a magnesium one and that with the same current and pressure the relation between V and d is accurately linear for six out of the nine metals tried. Further experiments with perforated cathodes carried out before the war but not published' till 1919, showed that the phenomenon just in front of the cathode, far from being a simple hail of positively charged particles, was inextricably complicated by intense local ionization. I was therefore forced to the conclusion, which has been further strengthened by my subsequent work on positive rays, that no simple theory can explain the mechanism of the discharge in a satisfactory quantitative manner.

It

seems in this way as intractible as an ordinary gas-flame, to which it bears many striking resemblances.

In conclusion, I may state that I now think that the true solution of all these perplexing discrepancies is probably to be sought for in the fact, now well established, that positive ions moving rapidly through a gas do not retain their identity but gain and lose electrons the whole time as they collide with other particles. The mechanism of these exchanges and its effect on the current carried is entirely unknown, but experiments of several different kinds are now focussed on this problem which should yield very valuable results in the near future.

June 7th, 1923.

F. W. ASTON.

XXI. On the Motion of Electrons in Gases. By V. A. BAILEY, M.A., D.Phil., Demonstrator, The Electrical Laboratory, Oxford, Lecturer of Queen's College, Oxford *.

1.

A

LARGE number of investigations have been made of the motion of electrons in gases, and it seems desirable to point out what reliable information may be obtained from the experiments, as the present generally accepted views of the behaviour of electrons in gases at the higher pressures and under low forces are to a great extent unsatisfactory. Too frequently the results of different experimenters do not agree with one another, or are obtained by a wrong interpretation of the observations; while in many cases the methods used are fundamentally unsound.

For values of the electric force X and gas-pressure p less than those for which ionization by collision occurs in notable quantities, the measurements are mainly concerned with W the velocity in the direction of the electric force, and K the coefficient of diffusion of charged particles in a gas. The particles may in general be either electrons or ions, but the motion of electrons is of special interest. The conditions under which ions are formed by electrons adhering to molecules have also been the subject of much discussion.

In some cases with ions the velocity W is found to be proportional to the ratio X/p. The constant of proportionality may then be termed the mobility of the ions, although the more common practice is to call W/X by that name, the gas being at atmospheric pressure.

* Communicated by Prof. J. S. Townsend, F.R.S.

No useful object is served by extending this definition to those cases where W is not proportional to X/p.

Some experimenters have made the mistake of basing their experiments on the assumption that W is always proportional to X/p, without verification. Considerable errors have arisen from this cause, for in the case of electrons the velocity W is generally not proportional to the ratio X/p.

2. Methods for measuring W have been devised by Rutherford, Langevin, Zeleny, Chattock, and Townsend. The methods of Zeleny and Chattock are restricted to those cases where W is proportional to X, and the same is true of that particular type of Rutherford's method in which an alternating e.m.f. is used. But in Lattey's form of Rutherford's method and in Langevin's method no such restriction occurs, and reliable results may be obtained by them, except in those cases where diffusion effects are comparable with those due to the motion W. The effects of diffusion may be avoided by Townsend's method which is particularly suitable for determining the values of W for electrons.

The only methods of determining the coefficient of diffusion K are those due to Townsend, which are described in his book on Electricity in Gases,' Chapter V.

3. When in 1908 † it was shown for the first time that in dry gases (or even in moist gases, provided X/p is sufficiently high) the ions exist as free electrons having an abnormal energy of agitation as a result of elastic collisions with the gas molecules, it became at once an important matter to determine the velocities W corresponding to this novel state of affairs. This was done by Lattey for air. and by Franck for argon and nitrogen, both using modifications of Rutherford's method.

In Lattey's form the charged particles coming through a gauze move under a constant electric force towards a parallel gauze for a known short interval of time t, and are then swept back again by reversing the field. No charged particles reach the second gauze until t is increased up to the value L/W, where L is the distance between the gauzes.

Franck made use of a sinusoidal alternating field in place of Lattey's constant field, so that his particles traversed the distance L with a variable velocity. He was thus compelled to assume that the velocity W was always proportional to X,

6

*Cf. J. S. Townsend's Electricity in Gases,' Chapter IV.

J. S. Townsend, Proc. Roy. Soc. A. lxxx. p. 207, and A. lxxxi. p. 464 (1908).

R. T. Lattey, Proc. Roy. Soc. A. lxxxiv. p. 173 (1910).

$ J. Franck, Verh. d. Deut. Phys. Ges. xii. pp. 291, 613 (1910).