If the stone be mounted on the goniometer and adjusted so that the edge of the prism is parallel to the axis of rotation of the instrument and if light from the collimator fall upon the table-facet and the telescope be turned to the proper position to receive the emergent beam, a spectral image of the object-slit, or in the case of a doubly refractive stone in general, two spectral images, will be seen in white light; in the light of a sodium flame the images will be sharp and distinct. Suppose that we rotate the stone in the direction of diminishing deviation and simultaneously the telescope so as to retain an image in the field of view, we find that the image moves up to and then away from a certain position, at which, therefore, the deviation is a minimum. The image moves in the same direction from this position whichever way the stone be rotated. The question then arises what are the angles of incidence and refraction under these special conditions. It is clear that a path of light is reversible; that is to say, if a beam of light traverses the prism from the facet _t_ to the facet _b_ it can take precisely the same path from the facet _b_ to the facet _t_. Hence we should be led to expect that, since experiment teaches us that there is only one position of minimum deviation corresponding to the same pair of facets, the angles at the two facets must be equal, _i.e._ _i = i´_, and _r = r´_. It is, indeed, not difficult to prove by either geometrical or analytical methods that such is the case. _A_ _A_ + _D_ Therefore at minimum deviation _r_ = ——— and _i_ = —————————, and, 2 2 since sin _i_ = _n_ sin _r_, where _n_ is the refractive index of the stone, we have the simple relation— _A_ + _D_ sin ————————— 2 _n_ = ————————————— _A_ sin ——— 2