Table of contents

An account is given of the use of a Rayleigh interferometer in a series of subsidiary experiments conducted in connexion with a recent determination of the diffusivity of solutions of salts in water.

The theory of the instrument, as affected by the different dispersions of the various media traversed by the light-rays, is studied, and the observations are corrected so as to apply to the green mercury light λ = 5461Å, which was employed in the experiments on diffusion.

The Paper deals with an extension of the study of "flashing" in discharge tubes containing air instead of the usual mixture of neon and helium.

The volt-ampère characteristics for the air discharge tubes are shown to be of the general form, i = k(V - M), where i is the current through the tube, k the conductance of the tube, V the voltage across the electrodes, and M a constant, theoretically of the value of the kathode fall of potential V_A.

The relation for the time of "flashing" T is given theoretically by

T = CR . log_e (E - V_B)/(E - V_C) + C Δ . log_e (V_C - V_A)/(V_B - V_A)

where C is the capacity in microfarads across the tube, R is the resistance in megohms, in series with the tube, E, V_C, V_B, the charging, upper critical, and lower critical voltages respectively, and Δ a constant.

This relation is generally confirmed by experiment.

It is shown that, as in the case of the "osglim" lamp, there is a critical resistance R_c below which no "flashing" is possible, given by R_c = (E - V_B)/[k(V_B - V_A)].

The Paper describes an improved design of the modified Sprengel pump described by the author in the Proceedings of the Physical Society, Vol. 36, p. 291. The walls of the pump are freed from gas skins by electric means, and the shattering of the mercury pellets (which tends to liberate gas) is checked by a special construction of the fall tube.

The apparatus described by Nettleton in the Proceedings of the Physical Society, April 1922, has been employed for the study of the Thomson effect in copper, iron, and carbon steels. It was found that different results were obtained according as the electric current in the wire under test flowed with or against the temperature gradient. The results are compared with those obtained by other workers for the thermo-electric power of the materials investigated.

Convenient arrangements for the superposition of a telephonic and a radio-frequency displacement of the light spot of a Cathode-Ray Tube are described. The arrangements include the following: -

1. Circular or elliptical trace at a telephonic frequency to serve as a time axis. When great steadiness, and uniformity are desired, a valve-maintained tuning fork controls the rate of revolution of the light spot.

2. Radial, circular and straight-line displacements at low radio or high audio-frequencies are next described and examples of the stationary patterns produced when these are superposed on the circular or elliptical time trace are given.

3. A special application of the method to the delineation of wave forms consists in the superposition of a long narrow elliptical time displacement and a straight-line high-frequency displacement in a direction at right angles to the major axis of the ellipse. By this means a considerable portion of the time ellipse is nearly a straight line representing a uniform time axis. The wave shape at the high frequency then appears in normal form and may be made of quite considerable size.

Some examples of each of the foregoing methods are given in the form of reproductions from prints made directly on pieces of bromide paper held down on the top of the Cathode-Ray Tube for a few seconds (3 to 10).

A special case occurs when an oscillatory valve system is arranged with grid and anode circuits closely coupled and harmonically resonant to each other. Wave forms consisting of a fundamental and a resonant single harmonic can be produced in this manner. Examples of current wave shapes in the grid and anode circuits are given, together with the voltage wave shape induced in a loosely coupled untuned coil.

The variation of Young's Modulus with temperature is found for zinc, silver, phosphorbronze, lead, and soda glass by a static method, to within about 150° of the melting-point. For these substances (as for several others recorded) the modulus varies exponentially with temperature, so that q(Young's Modulus)=q₁e^-b₁^t (where q₁ and b₁ are constants) up to a temperature roughly half-way from absolute zero to the melting-point, and q=q₂e^-b₂^t for the remainder. The values of the constants are tabulated.

By observations of the "Æolian Tones" of vibrators of aerofoil section, critical values for flow past an object of aerofoil section have been obtained of a fluid incident at various angles. The minimum value of VL/ν for unstead flow is found to fall from 60 at 0° incidence to 45 at 20°, and then more rapidly.

The conditions of reflection of X-rays from a thin layer of crystal powder are discussed and it is pointed out that for any given angle of reflection a surface of double curvature can be found, such that it will reflect X-rays coming from one point, to any other definite point. For an element of this surface, situated so as to be distant from the two points by lengths a and b respectively, the relation sin α/sin β = a/b must be satisfied where α and β are the glancing angles of incidence and of emergence of the X-rays with respect to the surface.

An arrangement for crystal analysis based on this relation is discussed and expressions are given for its resolving power. An apparatus of this type for photographic work and a hot cathode X-ray bulb of simple design are described.

A new determination of the lattice constant of nickel oxide is given.

The various methods of studying the potential differences and electric charges which arise at the liquid-gas interface are reviewed and discussed under the following headings: -

1. The determination of P.D. at the interface of liquid-gas when one or both of these is at rest.

2. Cataphoresis of gas bubbles.

3. Passage of gas over a liquid, without rupture of the latter, or of a gas over a wet solid with the same proviso.

4. The fall of a liquid in an unbroken column through a gas.

5. Liquid jets.

6. Waterfall electricity.

7. Electrification produced by bubbling gases through liquids.

8. Electrification produced by shattering drops in an air stream.

9. Electrification produced by spraying a liquid.

An attempt is made to co-ordinate the material obtained from this vast area of research. Many of the results may be explained in terms of the modern theory of orientation and polarisation at the liquid-gas interface, but there are other facts which do not appear to come within the scope of any established principles. Suggestions are made as to possible developments in future research.

A simple explanation is proposed of absorption, residual charge, and allied phenomena, in solid dielectrics. In accordance with classical theory, the dielectric is assumed to possess a certain true capacity, defined by its dielectric constant, and a certain conductance, here considered as probably electrolytic in type. The so-called anomalous properties are considered to be due to the properties of the contact surfaces of the dielectric and its metal electrodes. These contact surfaces appear to offer great resistance to the passage of ions or electrons across them, so that when an E.M.F. is applied to a condenser there is an accumulation of charges at the surfaces. These charges constitute the absorbed and residual charges. The experimental evidence in favour of this hypothesis is reviewed.

The behaviour of each contact surface is such that it may be represented by a large capacity in parallel with a high resistance, and thus a capacity-resistance combination is suggested, which is equivalent to an actual condenser. The properties of this combination are considered mathematically with reference to the experimental laws of absorption, and power losses in alternating fields, and the extent to which the theory is capable of explaining the experimental results is pointed out. The bearing of the theory on dielectric measurements under various conditions is also discussed.

The author employs a permeameter resembling that of Iliovici, in which the currents in two coils providing the M.M.F. of a magnetic circuit containing the specimen are adjusted until no magnetic potential difference exists between a selected pair of fixed points on the specimen. In the present apparatus the required absence of magnetic potential difference is tested by bringing up a yoke till its ends abut upon the two points in question: the approach of the yoke should excite no current in a search coil wound on the specimen and connected in a low-resistance galvanometer circuit. Resistance is then added to the galvanometer, and the deflection caused by a reversal of the two magnetizing currents enables the flux density to be calculated. The hysteresis loss per c.c. per cycle in cobalt-chrome steel is found by this method to be approximately 0.056 B_max^1.7. ergs, though the index of B_max. actually decreases with increasing values of the flux density. The magnetization curve for this substance, above the point of inflexion near the origin, is found to be given by H - c/β=d+H/β_s, where β=B - H, β_s is the saturation value of β, and c, d are constants. To form a permanent magnet whose external energy per c.c. is within 5 per cent. of the maximum obtainable, a magnetizing force of upwards of 1,000 C.G.S. units must be applied to cobalt-chrome steel. The βH loop between the +β and - H axes is given by a formula similar to that for the magnetization curve, with different constants instead of c, d, β_s, the value corresponding to β_s being much less than the true saturation density.

The simple relation to average current density, previously relied upon to describe low-pressure broadening of the lines in a hydrogen discharge, is shown by examination over a finer range to represent a rough approximation only. Concentration of ions is the obvious controlling factor if the Stark hypothesis be adopted; recombination of ions on this hypothesis may explain the capacity and inductance curves obtained in these experiments, between 0.3 and 1.0 A.U. This view is further tested by controlling the broadening without altering the current in the tube or the period of the discharge; for this, a range of widening over which rectified and unrectified discharges show lines differing in width, is provided by a magnetic field. The several effects involved in this experiment are accounted for on the theory that line width depends on the number of charges which surround an emitting particle. A thermionic method is also employed for varying the line width, by drawing into the discharge ions additional to those produced by collision.

Continuing the previous work on absorption of light by thallium vapour, the author has now studied the absorption of T1 vapour from λ2400 to λ2000, and that of In from λ6000 to λ2000. The absorption tube was of steel, and provided with quartz windows at the ends, and absorption was studied with a quartz spectograph. Tables were given in each case of the wavelengths of the lines absorbed. The absorption spectra indicate marked similarities. The first members of the series, 1π₂ - mδ', of both T1 and In, are found to exhibit assymmetrical absorption. In both cases there appear at high temperatures and on the short wave side of 1π₂ - 2δ' a channelled space spectrum consisting of assymmetrical bands degraded towards the red.

None of the lines of the principal series appeared in absorption, even at the highest temperatures used by the author. One remarkable feature is the very marked absorption of the members of 1π₂ - mδ'.

It is assumed that variations in the potential energy of a body (gravitational or electrostatic) are always accompanied by proportionate changes in its mass. Combining this assumption with the theories of Newtonian dynamics and Maxwelian electrodynamics, it has been found possible to predict all those phenomena, which are usually regarded as the crucial tests of the theories of relativity, both "special" and "general."

Formulae are given to enable the electrostatic capacity of two unequal spheres in contact to be computed more easily. It is shown that the natural logarithm of any number n can always be expressed as the sum of the capacities of a limited number of pairs of spheres in contact, the number of pairs never being greater than n/2. For instance log 7 is the sum of the capacities of the pairs of spheres whose radii are (1, 1/6), (1/2, 1/5), and (1/3, 1/4) respectively. Several similar relations are pointed out. An exact formula is given for the capacity of the pair of spheres, whose radii are a and b respectively. The formulae given are a help when computing the capacity coefficients of unequal spheres when very close together.

It is shown that the production of condensation nuclei, previously known to take place on illumination of air saturated with iodine vapour, also takes place under similar conditions with bromine and chlorine; the usual tests indicate that the nuclei in all three cases are similar.

The effect occurs in two distinct stages:

1. A surface action dependent on the presence on the surface of traces of some oxidisable substance.

2. Formation of nuclei in the "activated" vapour by illumination.

   A parallel is shown to exist between the conditions under which nuclei are found, and those under which an induction period is observed in the hydrogen-chlorine reaction.

Continuing the work of S. W. J. Smith and H. Moss upon the relation between the length of a capillary jet and its velocity of efflux from a cylindrical orifice, further examination has been made of the causes to which the main features of the curves obtained by these authors are due. Such curves consist of two main branches. In the first, with increasing velocity, the jet length rises until a critical point is reached. In the second, which begins at this point, the jet length diminishes rapidly with further increase of velocity.

The results contained in the present Paper indicate that, while surface tension is of prime importance in the first parts of these curves, viscosity is the dominating factor in the second.

A general dimensional formula which covers both cases is derived.

The following Paper deals with the variation with temperature excess of the controlled heat loss from an electrically heated wire. The wire, whose temperature was measured by means of a Callendar-Griffiths bridge, was surrounded by air at a known pressure enclosed in a glass tube, the latter being immersed in a constant temperature bath. The rate of heat loss, therefore, could be calculated from a knowledge of the current used and the resistance per unit length of the wire. Total heat loss/temperature excess curves were plotted, these tending to become straight lines as the convection was reduced.

This Paper is a sequel to one by S. W. J. Smith and J. Guild (Proc. Phys. Soc., 1912, Vol. 24, p. 342) on the "Self-Demagnetisation of Annealed Steel Rods," in which it was shown that the residual moment of a magnetised rod of annealed steel (of particular dimensions) does not diminish continuously towards zero as the temperature is raised, but changes sign at about 180°, reaches a maximum negative value at about 220°C., and subsequently slowly falls until the temperature at which steel ceases to be ferromagnetic is attained.

Experimental evidence is adduced to show that the reversal is primarily due to the laminated structure of the iron-iron carbide "eutectoid" which the steel contains.

It is shown that the reversal disappears when this structure is destroyed, and an explanation of the fact that the effect is most pronounced in a steel containing the eutectoid percentage of carbon is given.

Measurements have also been made to find how the intensity of the residual moment at 220°C., depends upon (1) the dimension-ratio of the bar, (2) the intensity of the field by which it is magnetised. The results are described and an interpretation is given.

It is suggested that measurements of the kind described could be employed to obtain rapidly and in a simple manner useful information concerning particular samples of steel.

The atmospheric electricity elements dealt with include the potential gradient, the numbers of positive and negative ions and the charges carried by them, the conductivity of the air and the vertical air-earth current. For instantaneous values it suffices to know the potential gradient and either the conductivity or the current. But in the case of mean values, depending on a number of individual observations, a knowledge of all three elements is necessary. Particulars are given of the annual variations observed at various stations. Information as to diurnal variation is very scanty, except in the case of potential gradient. The monthly values in the annual variation, and the hourly values in the daily variation, are expressed as percentages of their arithmetic means. This is done partly because less uncertainty attaches to relative than to absolute values in the case of all the electrical elements. Reference is made to some conclusions as to the diurnal variation of potential gradient and the existence of a sun-spot period recently arrived at by Dr. Bauer and Dr. Mauchly, of the Carnegie Institution of Washington.

The phenomena of underblown pipes has been studied in the case of four wooden pipes ranging over an octave. It has been found that it presents many features of remarkable interest. The existing theories reading the phenomena have been discussed, and a new theory is proposed which attempts at once to explain both these phenomena.

The facts of short and long-distance transmission for long and short waves are summarised and the examples of variable signals attributed to the varying phase and amplitude of rays ionically refracted by the upper atmosphere. The diurnal variation of such signals is described, and the strong night signals attributed to a lower atmosphere devoid of excessive ionization. Explanations of the difference between the magnitudes of directional errors for overland and oversea transmission and of the marked success of nocturnal short-wave transmission are put forward.

The salient points of the evidence relating to the seasonal variations of signal strength and to the influence of the configuration and inhomogeneity of the earth's surface are also summarised.

The facts of wireless telegraphy, as set forth show that the atmosphere exerts a variable but usually favourable influence on wave propagation. This influence may be interpreted in terms of ionic refraction (Eccles) which, to prevent excessive dissipation, must take place at levels sufficiently high to make the mean free paths of the effective carriers larges (Larmor).

A careful re-examination of the problem of the temperature gradients through gases at various pressures has been attempted.

Sets of curves indicating these gradients at various pressures and other interesting relations were obtained. Amongst other features, they show quite definitely the existence in contact with the vessel walls - at low pressures - of very large "temperature jumps."

An attempt to explain the general character of the results is based on the assumption of the presence at the walls, of a gas film possessing certain properties due to the attracting forces existing amongst wall molecules and gas molecules.

Instead of using one thermocouple only to determine the temperature at any point in the enclosure, several thermocouples of differing thicknesses - and therefore differing curvature - were employed under identical conditions. Systematic differences amongst corresponding readings of the thermocouples were noticed. These may also be attributed to the presence round the various thermocouples of gas films whose thickness and character vary with the thickness of the couple.

The Paper presents a brief review of data, as to the nature and origin of atmospherics. obtained in the course of a fundamental investigation of atmospherics undertaken by the Radio Research Board. The determination of the wave form of atmospherics by cathode ray oscillograph is referred to, and typical constants of peak field strength and duration are quoted. The results of an examination of the meteorological conditions prevailing at apparent sources of atmospherics, as located by direction finding, are tabulated, and show that even with the limited data available relations with thunderstorm phenomena, precipitation, or barometric minima have been traced, in 87 per cent. of 490 cases, distributed over Europe, the Mediterranean, and North Africa. The tracing of a "cold front" for 40 hours and 2,000 k. across Western Europe, by directional recording of atmospherics, is reported. Reference is made to the fine structure of atmospherics in relation to their interferent properties. The present state of the investigation is summarised.

A thundercloud is an electric generator in which the separation of positive and negative charges occurs at a rate corresponding to a current which may amount to some amperes; the potential difference between its poles may reach values of the order of a million kilovolts. Attention is drawn to three effects of the electric field of thunderclouds or shower clouds (all of which may occur even when there is no thunder). (1) The electric field of the cloud may cause ionization at great heights, the result being continuous or discontinuous discharge between the cloud and the upper atmosphere. (2) Discharge from pointed earthed conductors is likely to constitute an important part of the current between the ground and the base of a thundercloud, and the resulting ionization near the ground may be large. (3) By its accelerating action on particles the electric field of a thundercloud may produce extremely penetrating corpuscular radiation.

Balfour Stewart's "atmospheric dynamo" theory of the daily magnetic variations is described; the earth's permanent magnetic field is the magnetic field of the dynamo; the atmosphere (undergoing small, but widespread, motions owing to thermal or tidal causes) is the moving armature; and particular ionized regions in the upper strata correspond to the "windings" in which the induced currents flow. The variations of ionization are the chief cause of the changes of intensity in the diurnal magnetic variations, as between day and night, winter and summer, and sunspot minimum and sunspot maximum. Hence these changes in the magnetic variations indicate the changes in the ionization of the atmospheric regions concerned. It is suggested that there are two modes of ionization, both due to the sun, and affecting different regions; one region, ionized by ultra-violet radiation, is a practically world-wide layer, possibly at about 50 kilometres above the surface; this is most intensely ionized directly under the sun in the sun-lit hemisphere, and the ionization diminishes during the night. The other region consists of the polar caps within the auroral zones, and especially these zones themselves; this region is ionized by the impact of corpuscles emitted from disturbed regions on the sun's surface.

Two methods used for determining the dielectric constant of thin sheets of mica are described and the errors discussed.

The results of measurements on over 170 sheets of mica are given. The samples tested included best clear mica, ruby, green and brown in colour, stained and spotted samples, and also samples with gaseous inclusions. There were also samples of amber mica and silver amber mica.

The results show that practically all samples of muscovite (potassium mica) have the same dielectric constant, the most probable value being 7.0. The presence of inclusions has no great effect on this value. The samples of phlogopite (magnesium mica) - i.e., the amber micas - have rather lower dielectric constants, about 6.0 for normal amber mica, and 5.0 for the silver amber variety.

Values of power factor are given for many of the samples. For the clear micas (muscovite) it may be as low as 0.0003, and possibly less. The effect of inclusions is to increase it considerably. The power factor of the amber micas is greater than that of the muscovite variety - about 0.02 for the clear amber and 0.05 for the silver amber variety.

The samples were all about 8 cm. square and 0.1 mm. thick, and the measurements were all made on the Carey Foster bridge at a frequency of 800 cycles per second.

From both theoretical and experimental viewpoints it is established that the kinetic energy of emission of electrons from a substance exposed to X-rays may be (½mv²)K, L, M, etc. = hv - hvK, L, M, etc. where hv is the energy of the incident quantum and hvK, L, M, etc., is the energy associated with one of the "quantum absorption limits" of the substance. As the K quantum limit has greater energy than the L.M., etc., limits, it follows that the K β-rays should have smallest energy. Hence arises the difficulty in giving a quantitative explanation of the selective increase in ionization by collision and absorption when v exceeds vK.

The problem has been approached experimentally by Barkla and Miss Dallas, who have shown how the total energy of the β-rays from silver (as measured by the ionization produced in H₂) varies with wave-length as the region of the K absorption limit is crossed. The author has now determined directly the aggregate number of β-rays of all types emitted over the same range, and shows thereby, using his own previous observations, and also those of the authors quoted, that no selective change in the average energy occurs when the K fluorescent spectrum is excited. It is suggested that this is brought about by a re-grouping amongst the numbers and energies of the β-ray constituents, together with the addition of the K group of minimum individual energy as is required by the above equation.

The experimental work has necessitated the separation of the β-rays from the secondary effects they produce, the chief of which is the production of δ-rays of very low energy. The method of this separation is dealt with and the origin of the δ-rays discussed from the points of view of (a) thermionics, (b) recoil electrons, recently brought into prominence by Compton and C. T. R. Wilson, (c) impact electrons. δ-rays are impact electrons whose most frequent energy is little greater than 3.5 volts, which seems to be independent of the wave-length producing them, and of the substance from which they come. There are indications that they are influenced, at the moment of their departure from the surface, by the scattered X-rays.

1. The experimental evidence for the general chemical constant C₀ possessing the theoretical value ( - 1.589) is collected.

2. Tables of chemical constants are given and data collected for the determination of characteristic temperatures of the elements.

3. A linear relation between logarithm of mass and characteristic temperature is indicated.

The Paper describes an investigation of the sensibility of circular diaphragms as receivers of submarine sounds. The diaphragms were mounted with one face immersed in water and to the other was attached a microphonic or electro-magnetic detector. The diaphragms were all designed to have a frequency of 850 cycles/s under these conditions.

The response of the diaphragms to sounds of a single definite frequency was measured and resonance curves were obtained from which the magnitude of the damping due to the detector was determined. A value was also obtained for the concentration of energy in the restricted volume of the experimental tank, and an investigation was made of the effect of the proximity of sound reflectors and of the vibrating diaphragm of the source of sound.

It was shown that when corrections were made for these various effects, the values for the resonance amplitude and for the persistence of vibration were consistent with those determined mathematically by Prof. H. Lamb for diaphragms under ideal conditions.

The investigation was extended to cover complex sounds or "noises" of no definite pitch, and the results were again consistent with theoretical anticipations.

A direct-reading frequency meter for audio frequencies is described. It is a null instrument, reading by single adjustment, the working system embodying a new method of balancing mutual inductance by resistance. The standard type has five ranges, covering from 180 to 4,000 cycles per second, with accuracy of the order of 1 in 1 000, and negligible temperature coefficient except at the lowest frequencies; an extra set of ranges from 18 cycles per second upwards can be easily added.