fixed support so that the board can oscillate a little sideways. This carries the image of the sun backwards and forwards over the slit, so that the light admitted by the slit is alternately taken from a part of the solar disk near the following limb which is advancing towards the earth, and from a part near the preceding limb which is receding. Accordingly the solar lines in the spectrum are alternately shifted a little-perhaps about one-twentieth or one-thirtieth of a tenthet-metreto the right and left, while the earth-lines maintain their position unaltered. The eye readily detects this motion even when so small.

There is a solar line a little less refrangible than the eighth pair of double lines in the great B oxygen group, which, with the arrangement described above, is seen to approach and recede from the double line in sympathy with the motion of the pendulum. Another line well placed for the observation is the earth-line, which is a very little more refrangible than D, ; and another convenient group is where there is a strong iron line on the more refrangible side of D, about as far from D, on one side as D, is on the other. There is an earth-line in nearly the same position. The two appear as a single line when the light is taken from the preceding limb of the sun, and as a double line when it is taken from the following limb; and with the pendulum arrangement these appearances alternate.

Of these three the observation on the B line can be well made in the second spectrum of a Rowland's grating 1 inch long. The observations near D are best made in the fifth spectrum.

There are of course multitudes of other lines on which the observation can be made.

4. On the Cause of Double Lines in Spectra.

By G. JOHNSTONE STONEY, M.A., D.Sc., F.R.S.

The lines of the spectrum of a gas are due to some events which occur within the molecules, and which are able to affect the æther. These events may be Hertzian discharges between molecules that are differently electrified, or they may be the moving about of those irremovable electric charges, the supposition of which offers the simplest explanation of Faraday's law of electrolysis. The amount of the charge which is associated with each of the bonds, and of which two or more seem to be present in every chemical atom, is always the same quantity of electricity. In a communication made to the British Association in 1874 the author invited attention to this fixed quantity of electricity as one of three fundamental units presented to us by nature (see Phil. Mag.' for May 1881), and estimated its value, which is about the twentiethet (i.e. 1/100) of the electromagnetic unit of quantity in the ohm series.

Several considerations (of which perhaps the most decisive is the phenomenon of the reversal of lines) suggest that the source of the spectral lines is to be sought not in the Hertzian discharges, but in the carrying about of the fixed electric charges, which for convenience may be called the electrons. The present investigation however is not dependent on this or any other particular hypothesis, since it is with the laws of the events within the molecules that it is concerned, and the course of the investigation shows that these laws are the laws of the motion of separate elements of volume, which may be conveniently thought of as the motion of those parts of the molecule to which the electrons are to be regarded as indissolubly bound. An electron, if waved about in some particular way by the motions within the molecule, would occasion such electro-magnetic waves as are revealed to us by the spectroscope.

Now the irrotational motion of an element of volume consists in its traversing some orbit, accompanied perhaps by a simultaneous distortion of its form. We are only concerned with the orbital motion. This motion may be resolved by Fourier's Theorem into the superposition of partials, each of which is a simple pendulous motion in an ellipse, and each of these partials produces its own line in the spectrum. Seven constants are required for the full determination of each partial if the orbit of the electron is a curve of double curvature, or five if it is a plain curve. Now the observation of a line supplies only two equations between

these. The wave-length of the line, when corrected for the refraction of the air, gives the periodic time of the motion of the electron in the corresponding partial, and the brightness of the line gives a quantity proportional to a2+b2, a and b being the axes of the ellipse.

But there is one case, and fortunately a case which at all events frequently occurs, and that perhaps is universal, in which we receive a very interesting addition to our knowledge; explaining on the one hand the double lines that are so frequent in spectra, and on the other telling us the actual forms of the elliptic partials of the motion going on in the molecules. This important case occurs whenever some of the forces which determine the motion of the electron are feeble compared with the others, and are such as to produce that familiar form of perturbation which consists in an apsidal motion of the elliptic partials. When this perturbation prevails, the lines are rendered double by it, and an examination of the positions and intensities of the two constituents of a double line enables us to determine (a) the form of the elliptic partial to which they are due, (b) the time which the electron takes to travel round it, and (c) both the direction and speed of the apsidal perturbation. Thus the principal double line of sodium is found to have its source in a long elliptic partial, the ratio of the axes of which lies somewhere between 11:1 and 13:1. Round this partial the electron travels about 1,984 times during one revolution of its slow apsidal perturbation, and there is time for about 36 of these slow apsidal revolutions to take place during each flight of the molecule. Moreover in this case the apsidal motion takes place in the same direction as the motion of the electron round the ellipse. An equal amount of information can be obtained in the case of every other double line that can be adequately observed.

The author thought he had reason to suspect from observation that almost all spectral lines are double, and that they appear single only when our spectroscopes have insufficient revolving power, or when each of the constituents has so widened out as to obliterate the interval between them, or in the rare cases when the partial from which they arise being circular one of the two constituents of the double line is of cypher intensity. If this shall turn out to be the case, there must be some common cause for the apsidal perturbation, and the author ventured to suggest as the most probable cause the feeble reaction which the æther exerts on the electron, consequent on the energy which the molecule imparts to the æther when producing the electromagnetic waves. A fuller account of the investigation is being printed by the Royal Dublin Society in its Scientific Transactions.

5. Seventh Report of the Committee on Solar Radiation.
See Reports, p. 160.

6. Report of the Committee on Meteorological Photography.
See Reports, p. 130.

7. Report of the Committee on the Meteorological Observations on Ben Nevis. See Reports, p. 140.

8. Report of the Committee on the Reduction of Magnetic Observations. See Reports, p. 149.

1 It is shown in the investigation that each of the two constituents of a double line arises from a circular motion. Accordingly they would not suffer further duplication if an additional apsidal perturbation were introduced.

9. Report of the Committee on the Seasonal Variations in the Temperature of Lakes, Rivers, and Estuaries. See Reports, p. 454.

10. On the probable Nature of the Bright Streaks on the Moon. By Dr. RALPH COPELAND, F.R.A.S., F.R.S.E.

In this paper the author described the chief features of the bright lunar streaks, especially their invisibility when the shadows of the mountains are most conspicuous, and their great prominence when the lunar shadows are imperceptible. It was explained that the bright streaks demanded for their visibility not so much a high angle of illumination, as a front illumination. In other words, they become visible when the light falls more or less closely in the line of sight.

If this condition is fulfilled the streaks come prominently into view quite regardless of the inclination of the surfaces on which they occur. The surfaces indeed may make almost any angle, either with the line of sight or with the sun's rays, provided they are at all turned towards the common direction of the spectator and the sun.

An important deduction from this fundamental fact is that each elementary portion of the streak surface is of a form that is symmetrical to the spectator from whatever point it is seen. The sphere alone appears to fulfil this condition; hence it may be assumed that the surface of the streak material must be made up of a large number of more or less complete spherical surfaces. These minute surfaces may be either concave or convex. We may therefore regard the streaks as being produced by a material pitted with minute cavities of spherical figure, or strewn over with minute solid spheres. In the latter case it is probable that the material is more or less transparent, or at least translucent.

To test this hypothesis, a plaster model of the moon 22 inches in diameter was made, on which the bright streaks are represented by lines of minute spherules of transparent glass attached to the surface. These possess in a marked degree the desired property of remaining inconspicuous under cross light, while they flash out brilliantly when lit up from the front. Although the spherules are but 1-50th to 1-30th inch in diameter, they are still too large in proportion to the model, and therefore cast perceptible shadows when they would otherwise be invisible. This might have been largely avoided by the tedious process of partially imbedding them in the model. The corresponding diameter on the moon's surface would be from 2 to 3 miles.

When suitably illuminated the phases of the model were found, on photometric examination, to follow a law not very unlike that of the lunar phases as derived by Zöllner from his own observations, and those of Sir John Herschel, the light of the 'full moon' being nearly five times that at quadrature. Without streaks the model closely agreed with Lambert's formula for a non-reflecting sphere, for which the full disc is 3.1416 times as bright as the half disc illuminated from the side. The paper was illustrated by a diagram showing the relative brightness of the phases as well as by the model and photographs of the same. The model, suitably illuminated, was also exhibited at the evening conversazioni of the Association.

TUESDAY, AUGUST 25.

The following Report and Papers were read:

1. Report of the Committee on Electrical Standards.-See Reports, p. 152.

1

2. The Causes of Variation of Clark Standard Cells. By J. SWINBURNE. The various parts of the cell are examined separately. Any zinc will do, if amalgamated before use. The greatest variations are due to impurities, such as 1 Published in full in Electrical Review, August 28, 1891.

traces of iron, which are found in even the best purchased sulphate of zinc. The sulphate of zinc solution may also contain basic sulphates, and it is not homogeneous after any variations of temperature. This gives rise to variations of the electromotive force of the cell, and also to large variations of temperature-coefficient. The mercurous sulphate bought as pure nearly always contains a great deal of mercuric sulphate. The effects of these and other causes of variation are discussed, and an amalgam cell, preferably with a non-saturated solution, recommended.

3. A Joint Discussion with Section G. on Units and their Nomenclature was opened by Professor OLIVER J. LODGE, F.R.S., followed by W. H. PREECE, F.R.S.

The following Papers were read in connection with the discussion, viz. :-

Some Revolutionary Suggestions on the Nomenclature of Electrical and Mechanical Units. By Professor W. STROUD.

Present Practical System of Units.-1. The present practical system of units is very objectionable on three grounds

(a) There is no prima facie reason why the practical unit of current should be equal to 1-10th c.g.s. unit.

(3) The relation between the other practical electric units and the corresponding c.g.s. units is much more complex than need be.

(7) The units of work and power are far too small for practical requirements.

2. If we were starting to devise a practical system to-day, such a system could best be formed by taking 10 cm. as the unit of length, 10-2gm. as the unit of mass, and the second as unit of time.

3. That in the interests of the 'practical' men of the future and in the interests of the electrical students of both the present and the future, it is highly desirable to initiate a revolution with the object of dethroning the present practical system of units.

Nomenclature.-1. That the term Dyne to indicate 107 of our present (1891) dynes is objectionable, as custom has restricted the use of Greek derivatives entirely to c.g.s. units.

That 10' dynes, if required, might be called a Hebdomodyne, suitably contracted, of course, or preferably a joc (joule over centimetre).

2. That the classical languages are of little or no service for the provision of names for modern, more or less complex, physical conceptions, and therefore this method of coining words it is desirable to abandon.

3. That for c.g.s. units some system of automatic nomenclature in which every name shall be self-explanatory would prove a boon to the teachers and a blessing to the student, and that such a system is quite capable of being devised.

4. That the prefixes meizo to indicate 10°, and mei to indicate 10- may be found useful.

On a Table to Facilitate the Conversion of Electrostatic and Electromagnetic Measures into one another. By G. JOHNSONE STONEY, M.A., D.Sc., F.R.S. The fundamental equations of electricity are:

a2QQ'

F = a2?, for the repulsion between two quantities of electricity.

F = ¿¿PP'

1891.

for the repulsion between two quantities of magnetism, and

PP

F=

2PQ e
T'72

for the repulsion between a linear current and a magnet; a, b,

and c depending on the specific inductive capacities of the medium for electricity and magnetism.

The dimensions of the various units in electricity and magnetism, if written out in full, would contain these coefficients, which are so related that abje2 is a velocity. Our knowledge of the significance of this standard velocity is chiefly owing to Clerk Maxwell, and the author suggested that it shall be called the Maxwell. It appears from the electromagnetic theory of light that it is also the velocity of light.

The above relation may be written in either of the forms,

Accordingly, if we choose to confine our attention to the cases in which ca has its unit of value,

b= =v [multiplied by a coefficient of dimensions, c2/a, and of unit value]; and if, on the other hand, we confine our attention to the cases in which c2/b has its unit of value,

a = v [multiplied by a coefficient of dimensions, c2/b, and of unit value].

The first of these assumptions is that the specific inductive capacity for electricity of the medium (usually air) is taken as our unit of specific inductive capacity for electricity, and the second assumption is that the specific inductive capacity for magnetism of the medium is taken as the unit of specific inductive capacity for magnetism.

Electrical units consistent with the first assumption are called electrostatic units, those consistent with the second assumption are the electromagnetie units. In the dimensional equations of the electrostatic system ca disappears and is replaced by unity, in those of the second system it is cb which disappears. This makes the difference between the imperfect dimensional equations of the two systems, which is therefore only apparent; and the ratios between the units of each physical quantity, whether estimated electrostatically or electromagnetically, are essentially numerical.

Units consistent with both assumptions can only be obtained if we use the Maxwell velocity as our unit of velocity, in which case a, b, and c can all have unit values; and the central column of the following table is based on this assumption, and is introduced to afford a common ground up to which it is sufficient separately to trace from the right and left the numerical relations of the units of the systems in common use (by the help of the two systems of imperfect dimensional equations), in order to arrive at the numerical relations the whole way across. The Maxwell must be our unit of velocity in the central column; but we are at liberty to choose two other units arbitrarily, and they are so selected as to make the unit of time and the unit of LM the same in the central column as in the two adjoining ohm columns. This reduces the numerical relations to their simplest form. The table can easily be extended to include the units of every other electrical quantity.

The name potency is suggested for what is too often miscalled a force, or the intensity of the field. At every station in space there is potency over the magnetism that is there present, over the electricity, over the mass, and over the volume occupied (producing buoyancy), if the surrounding medium is excluded from it. There are, therefore, four potencies at each point of space, each being one factor of a force, the other factor being a quantity of magnetism, of electricity, of mass, or of volume, as the case may be. The author also expressed his hope that the phrases electromotive force and pressure may be discontinued, for what is in fact one factor of an energy, the other factor being a quantity of electricity. Voltage, which has in some degree come into use, was recommended instead.