It has been suggested that the Aurora, inasmuch as it presents a structural character, may afford traces of polarization. Having reference to the fact that the striæ of the electric discharge in vacuum-tubes present no such feature, the probability of the suggestion may be doubted. But it will still be worth while to put the question to an experimental test. If traces of polarization be detected, it must not at once be concluded that the light of the Aurora is polarized; for the Aurora may be seen on the background of a sky illuminated by the moon, or by the sun, if not too far below the horizon, and the light from either of these sources is, in general, more or less polarized; therefore, if the light of the Aurora is suspected to be polarized, the polariscope should be directed to an adjacent portion of clear sky, free from Aurora, but illuminated by the moon or sun as nearly as possible similar, and similarly situated to the former portion; and the observer must then judge whether the polarization first observed be merely due to the illumination of the sky. The presence of polarization is to be determined:— (1) With a Nicol’s prism, by observing the light through it by turning the prism round on its axis, and by examining whether the light appears brightest in some positions and least bright in others. If such be the case, the positions will be found to be at right angles to one another. The direction of “the plane of polarization” will be determined by that of the Nicol at either of these critical positions. The plane of polarization of the light transmitted by a Nicol, is parallel to the longer diagonal of the face; and, accordingly, the plane of polarization, or partial polarization, of the observed light is parallel to the longer diameter of the Nicol when the transmitted light is at its greatest intensity, or to the shorter when it is at its least. (2) The observation with a double-image prism is similar to that with a Nicol. This instrument, as its name implies, gives the images which would be seen through the Nicol in two rectangular positions, both at once, so that they can be directly compared; and when in observing polarized light the instrument is turned so that one image is at a maximum, the other is simultaneously at a minimum. Both these methods of observation, (1) and (2), are especially suitable for faint light; because in such a case the eye is better able to appreciate differences of intensity than differences of colour.