more intense in the light reflected by the crystals, and the higher bands were more distinctly seen. The surface-colour of potassium permanganate, and the position and intensity of the bands in the spectrum of the reflected light, are independent of the relative position of the plane of incidence to the long axis of the crystal, or to the striæ produced by rubbing, when the powdered substance burnished on glass is used. 1. Surface-Colours.-Freshly prepared surfaces of potassium permanganate appear of a pale yellow when light, either unpolarized or polarized in any plane, is incident upon them at low angles. But with ordinary light, and with light polarized in the plane of incidence, the amount of white light reflected is so great at high angles that the surface-colour, if any, is completely masked. When the incident light is polarized perpendicularly to the plane of incidence, or when unpolarized light falls on the surface and a Nicol is placed between the eye and the permanganate, with its principal section in the plane of incidence, the surface-colour is seen to change as the angle increases, becoming successively green and blue, and finally white and metallic. The surface-colours alter with the surrounding medium. The following Table gives, approximately, the colour at various incidences, (A) when the light is either unpolarized or polarized in the plane of incidence, and (B) when it is polarized perpendicularly to that plane, for potassium permanganate in air, tetrachloride and bisulphide of carbon. 2. Reflection Spectra.-With unpolarized light, and still more with light polarized in the plane of incidence, the dark bands in the spectrum of the reflected light are never very distinct. I was not able to observe whether the bands shifted or not as the angle of incidence increased, as the amount of white light reflected at angles of 55° and upwards was so great as to render the bands invisible. They appeared, however, as long as they were visible, to coincide exactly with the bright spaces in the absorption spectrum of a dilute solution of potassium permanganate, which was thrown into the field by means of the reflecting prism. When the incident light is polarized perpendicularly to the plane of incidence, the dark bands are far more distinctly seen. At angles of less than 40° there are four bands, and the blue end of the spectrum is very weak. As the angle of incidence increases, the intensity of the blue rays diminishes; and then the amount of light in the red decreases; and at about 55° nearly the whole of the light comes from the bright bands. As the angle of incidence increases beyond this amount, the dark bands gradually move towards the blue end of the spectrum; and at about 60° a new band appears near D. With any further increase of the angle more of the blue rays are reflected; and the bands fade away, those in the more refrangible part of the spectrum disappearing first. The relative intensity of the dark bands varies with the angle of incidence. When this is small, the third and fourth bands, counting from the red end, are darkest; with the increase of the angle the second, the first, and finally the new band, become successively darkest. I have not been able to obtain any satisfactory measurements of the amount of the displacement of the bands, as, when a spectroscope of sufficient power to render it an easily measurable quantity is used, the bands become so ill-defined that it is impossible to measure them. Approximately the displacement amounts to about 006 in "tenth-metres ;" and the bands tend to coincide with the dark bands of the absorption spectrum, instead of with the bright bands as they do when the angle of incidence is about 550 or less. The shaded portions of the diagrams are intended to give the relative amount of light, as determined by eye estimations, in the different portions of the absorption and reflection spectra of potassium permanganate-the ordinates being taken to represent the intensity, and the abscissæ wave-lengths. The curved line gives the intensity of the light in the different portions of the normal spectrum, as determined by Mossotti from Fraunhofer's measurements, neglecting the minor irregularities in the curve as given by him. * Pogg. Ann. lxxii. p. 509. Fig. 1 is the absorption spectrum of a solution of potassium permanganate in water. Figs. 2, 3, and 4 the reflection spectra, when the incident light is polarized perpendicularly to the plane of incidence, and falls on the surface at angles of 50°, 60°, and 70°. As has already been announced by Wiedemann, the position of the bands in the reflected light depends on the nature of the surrounding medium. From the experiments I have made, it appears that, with unpolarized light, the first dark band of the reflection spectrum corresponds in position with the first bright band of the absorption spectrum, whether the permanganate is in air, benzene, or either bisulphide or tetrachloride of carbon; these liquids, however, act on the permanganate, and after a short time the surface becomes altered, and then the bands correspond with the dark bands of the absorption spec trum. Figs. 4 and 5 represent the distribution of light in the spectrum, with fresh surfaces of potassium permanganate in bisulphide and tetrachloride of carbon, when unpolarized light is incident upon them at an angle of about 55°: in both cases the bands are wider apart than in air. 650 C LIX. On a possible Cause of the Formation of Comets' Tails. By A. S. DAVIS, M.A.* IT T is well known that the phenomena observed during the formation of a comet's tail point to the conclusion that the material which forms that appendage is being continually emitted from the head of the comet with great velocity by some force acting in a direction directly away from the sun. The material appears in most cases to be first ejected from that side of the comet's nucleus which is turned towards the sun, and afterwards, under the influence of this force, to be turned backwards to form a tail. It is the object of this paper to suggest an explanation of this force. The remarkable identity which has been found to exist between the orbits of certain comets and the orbits of certain meteoric clouds renders it little short of certain that comets are themselves masses of solid or liquid bodies separated from one another by great intervals, except perhaps at the nucleus, where they may be closer together, and may even contain a solid core. The spectroscope indicates the presence of gas in a state of incandescence in the comet's head and nucleus; but how this incandescent gas is produced is not known. The violent action which is observed to take place as a comet approaches the sun, on that side of its nucleus which is exposed to the solar radiation, appears to indicate that the comet consists largely of matter which is rapidly volatilized under the influence of the sun's rays. Let us assume that such is the case, and let us consider what will be the effect of evaporation on the motion of one of the bodies undergoing it. In the first place the mass of a comet is so small, that the force of gravitation towards the centre on any of the bodies at some distance from the nucleus must be so small that it may be left out of consideration. We know that a molecule of matter in the gaseous condition has at ordinary and high temperatures a very quick motion of translation. A molecule, as it evaporates from the surface of one of the bodies composing the comet, must acquire a velocity relative to the body of several hundred yards per second. The body must in consequence suffer a recoil in an opposite direction to that in which the molecule escapes. Now since the evaporation is caused by the sun's heat, it must take place chiefly on that side of the body which is exposed to the sun's rays. The resultant effect of all the small recoils due to the evaporation of the different molecules will therefore be to drive the body in a direction away from the sun. If the body has a motion of rotation, the * Communicated by the Author. whole surface might in turn become exposed to the sun's rays, and evaporation would probably take place even on the side turned away from the sun. But unless the body be of a regular shape, the effect of evaporation will be to gradually stop any rotation which it might at first have; for the force of recoil from the evaporation would act upon it in the same way as the wind does on a vane, and it would at length take up a position with its longest axis in the direction of the sun. m Now let V be the average velocity relative to the body with which the molecules escape from it. Let M be the mass of the body just before the escape of a molecule of mass μ from it. Then the velocity due to the recoil as this molecule escapes will be "V. Now the molecules as they evaporate will start off in various directions, but almost always more or less towards the sun. Let us suppose, as being not far from the truth, that the inclination which their directions have to a straight line passing through the sun is on the average 45°; then the average velocity due to recoil acquired by the body on the evaporation of mass dm will be and the velocity acquired whilst the mass of the body is being reduced by evaporation from m1 to Mg will be dm V Now the average velocity of hydrogen molecules at 0° C. is 1.06, of oxygen 266, and of water vapour 35 mile per second. For the sake of illustration, let us suppose that V=35, and =1000000, these being the values we should have to assign if the body were a block of ice containing one gramme of sand or any other non-volatile substance, the block itself being equal in mass to a cubic metre of water. The velocity due to recoil by evaporation would then be 3.42 miles per second, or about 295,000 miles per day. A tail would thus be formed which would increase in length nearly a million miles in every three days. The visible portion of this tail would consist of solid or liquid matter which had resisted evaporation; but there would also be present in the tail a large portion of the gas formed during evaporation; for since the evaporating gas has a velocity relative to the body from which it is evaporating of ⚫35 mile per second, those portions of the gas which have evaporated since the body acquired by recoil a velocity greater than 35 mile per second will be also carried backwards into the tail. The estimate of the rapidity of tail-formation I have |