English translation of A. Fresnel, "Note sur la double réfraction du verre comprimé", Annales de Chimie et de Physique, Ser. 2, vol. 20, pp. 376–83 (1822), as reprinted in Oeuvres complètes d'Augustin Fresnel, vol. 1 (1866), pp. 713–18, with the corresponding extract from the "Table Analytique" in Oeuvres complètes..., vol. 3 (1870), at p. 595. Brewster has discovered that when a plate of glass between two polarizers is compressed or stretched in a single direction, it displays colors analogous to those of birefringent crystalline plates, whence he has promptly concluded that stress induces birefringence. As not all physicists were convinced that such colors were due to simple birefringence, Fresnel, in 1819, established by interference experiments that the propagation speed depends on whether the polarization is parallel or perpendicular to the compression. Although Fresnel himself was thereby convinced of the birefringence, he has thought it desirable to confirm the conclusion through actual double refraction. Four right-angled isosceles glass prisms were lined up with their long rectangular faces touching end-to-end in the same plane, and with their 90° refracting angles facing the same way. The spaces between were filled by three more prisms with the same base dimensions but a slightly shorter height; and two half-prisms, also of reduced height, were added at the ends so as to make the overall assembly rectangular. To suppress reflections, the residual air gaps were filled with turpentine with a refractive index nearly equal to that of the glass. When the first-mentioned prisms were compressed in a suitably designed vise (the others escaping the compression by being shorter), objects viewed through the length of the assembly appeared double, the two images having a separation of about 1.5 mm at one metre, with polarizations parallel and perpendicular to the direction of compression. Fresnel confidently expects that compression in two perpendicular directions to different degrees would produce biaxial birefringence, and that the inclinations of the optic axes would be easily calculable from the strains; but he acknowledges that experimental verification would be difficult because of the almost inevitable non-uniformity of compression. Returning to the experiment just reported, he concludes with the daring prediction that if the stressed glass prisms are replaced by unstressed quartz prisms with their optical axes along the length of the assembly, there will again be two images, which will appear unpolarized when viewed through an analyzer but, when viewed through a Fresnel rhomb (as we now call it), will be polarized at ±45° to the plane of reflection of the rhomb.