According to my earlier investigations (Pogg. Ann. vol. cxxviii. p. 562, vol. cxxix. p. 211, 1866), a prime angle of incidence and a prime azimuth can still be spoken of when the light is reflected in air from a thin, transparent lamina of metal lying on a glass plate. The difference of phase and ratio of amplitudes of the components polarized parallel and perpendicular to the plane of reflection can thence be calculated for any angle of incidence, just as with non-transparent metallic mirrors.

I have therefore determined the prime angle of incidence and prime azimuth on transparent laminæ of gold, platinum, and silver, as well as on non-transparent metals. (See Table II.) They lay on plates of mirror-glass, the back of which was blackened with oil-colour in order to destroy their reflection.

The gold was obtained by Wernicke's process (Pogg. Ann. vol. cxxxiii. p. 183, 1868). The silver was obtained thus: from the same freshly prepared Martin's silvering-fluid, on strips of the same plate glass, dressed simultaneously in the same manner, I let silver be precipitated during 1, 2, 3, 6, and 15 minutes respectively. The thicker layers were converted into silver iodide; and from the colour of the iodide layer the thickness of the silver was calculated (cf. Pogg. Ann. vol. cxxix. p. 208, 1866). The platinum was a mirror, such as have been sold in France for some years past. The thickness given of the gold and platinum layers is only from a very rough estimate.

The observations on the thinnest layer of silver, as it was not everywhere of equal thickness, deserve but little confidence. The rest of the mirrors of silver, however, were most perfect and homogeneous, and reflected the light strongly, although they had merely been rinsed with water and dried but not polished with leather.

The polarization-angles of the unsilvered glass plates were calculated from the refraction-exponents of the surface, determined by a peculiar process. The glass for the gold layers had exactly, and that for the platinum very nearly, the same exponents of refraction as for silver.

These measurements agree substantially with the results of the previous investigations, which yet were made with only onecoloured red light. They show that the prime angle of incidence and the prime azimuth increase as the thickness of the metal increases, but in different degrees for different colours. The values of the prime azimuths for the different Fraunhofer lines approximate to one another with increasing thickness of silver, recede from one another with greater thickness of gold.

With silver the prime azimuth exhibits a maximum value for a certain Fraunhofer line, which maximum moves towards the red end of the spectrum as the thickness is augmented.

Gold, in transparent as in non-transparent layers, has a minimum angle of incidence for Fraunhofer's line F.

The values of H and B were not altered by magnetizing or electrizing the metallic mirrors.

By polishing and pressure of the layer of metal the value of the prime angle of incidence is augmented, that of the prime azimuth diminished. I have observed this also with other metals than those cited in Tables I. and II., for example cobalt and nickel. The treatment may vary much for the different colours. Only with reflection from silver into glass was there shown a slight increase of the prime azimuth through pressure, which, however, was not exerted in this case on the reflecting front surface of the metal, but on its hinder surface bounded by the air. The effect of pressure, or of the distance of the particles of silver, upon the constants of reflection was most strikingly exhibited with silver-collodion films, for which I am indebted to M. F. A. Nobert, of Barth. They were obtained by the process used in photography-a glass plate coated with collodion (containing iodide of potassium ?) being immersed in a silver solution, browned by exposure to daylight, and made black by a strong solution of pyrogallic acid. After drying, the plate was strongly heated, in order to make the collodion film adhere more fimly to the glass.

Such a collodion film contains uniformly distributed finely divided particles of silver, is opaque even when very thin, but reflects too little light for the double-reflection method. With Babinet's compensator and homogeneous red or blue glass, trial was made with reflection in air :

After gentle rubbing with soft buckskin, when the plate took a

polish :

:

An ordinarily transparent collodion film, rendered opaque by finely divided particles of silver, shows therefore a slight alteration of the prime angle of incidence, but a great alteration of the prime azimuth, which approximates to that of pure silver.

If by a slight pressure the silver particles are brought nearer

to one another, the prime angle of incidence and prime azimuth change very considerably in the same direction, only more than with a layer of pure silver particles on glass. Moreover, with a suitable pressure and reflection in air the prime azimuth may be nearly the same for blue as for red.

Whether the silver-collodion film be unpolished or polished, the particles of silver have always a different distance from one another, or a different distribution, in the vicinity of the glasssurface and in the vicinity of the free surface bounded by air; so that the reflections in air and in glass are not directly comparable.

According to the distribution of the silver particles in the interior of the collodion, which varies of course with the nature of the exposure in its preparation, the prime angle of incidence and prime azimuth vary for different colours, and the silver collodion-film (photographic plate) exhibits various colours under the incidence of natural white light, which may be different for reflection in air and in glass.

Precisely similar is the behaviour of substances with so-called surface-colours, with which H and B increase and diminish with increasing wave-length.

Carthamin, dried in a red-coloured transparent layer on the faces of the above-mentioned parallelepipeds, exhibited, with reflection in the same substance, different reflection-constants in different parts, according as the distance between the particles was less or greater. The parallel-mirror method showed, for light of the given wave-lengths A (in millionth parts of a millimetre), the following mean values of the prime angle of incidence H and prime azimuth B: