1. The absorption spectrum observed through a crystal varies with the direction of the rectilinear luminous vibration which propagates itself in this crystal. 2. The bands or rays observed through the same crystal have, in the spectrum, fixed positions, their intensity alone varying. 3. For a given band or ray there exist in the crystal three rectangular directions of symmetry, according to one of which the band generally disappears, so that for a suitable direction of the luminous vibrations the crystal no longer absorbs the radiations corresponding to the region of the spectrum where the band question appeared. These three directions may be called the principal directions of absorption, relative to this band. 4. In the orthorhombic crystals, by a necessary consequence of crystalline symmetry, the principal directions of absorption of all the bands coincide with the three axes of symmetry. We may thus observe three principal absorption spectra. In uniaxial crystals the number of absorption spectra is reduced to two. 5. In clinorhombic crystals one of the principal directions of absorption of each crystal coincides with the only axis of symmetry; the two other principal rectangular directions of each band may be found variously disposed in the plane normal to this axis.

Most commonly these principal directions are very near to the principal corresponding directions of optical elasticity. 6. In various crystals the characters of the absorption phenomena differ strikingly from those which we might expect to find after an examination of the optical properties of the crystal. We have just seen that in clinorhombic crystals the principal absorption directions of certain bands were completely different from the axis of optical elasticity of the crystal for the corresponding radiations. If we examine this anomaly, we perceive that the crystals manifesting these effects are complex bodies, formed of various matters, one, or sometimes several, of which absorb light and give each different absorption bands. Now, M. De Senarmont has shown that the geometric isomorphism of certain substances does not necessarily involve identity of optical properties, and in particular in the directions of the axes of optical elasticity in relation to the geometric directions of the crystal. In a crystal containing a mixture of isomorphous substances, each substance brings its own influence, which may be made to predominate in turn according to the proportions of the mixture.

We may, therefore, admit that the molecules of each substance enter into the crystal retaining all the optical properties which they would have if each crystallized separately. The principal directions of optical elasticity are given by the resultant of the actions which each of the component substances exerts on the propagation of light, while the absorption of a given region of the spectrum is due to a single one of these substances, and may have for its directions of symmetry the directions which it would have in the absorbing molecule supposing it isolated. It may happen that these directions do not coincide with the axes of optical elasticity of the compound crystal. If such is the cause of the anomaly of certain principal directions of absorption, the bands which present these anomalies must belong to substances different from those which yield bands having other principal directions of absorption. If so, we are in possession of a novel method of spectral analysis, which permits us to distinguish in certain crystals bands belonging to different matters, isomorphous, but not having the same optical properties.

Two bands appearing in a crystal with common characters, but presenting in another crystal characters essentially different, must also be ascribed to two different bodies.

[Continued from SUPPLEMENT, No. 585, page 9345.]