Repeated measurements of 100 lines of the solar spectrum have enabled me, upon the basis of Ångström’s Atlas (‘Spectre normal de Soleil’), to express the indications of the screw directly in wave-lengths. In place of the cross wires originally introduced into the focus of the observing telescope, I have inserted a tiny polished steel cone, the very fine point of which reaches to the centre of the field of vision. The axis of this cone stands perpendicular to the length of the spectrum, therefore parallel with the spectral lines, and the setting of the point of the cone on the latter is accomplished with great sharpness. If the spectrum is very faint, or consists only of bright lines, the cone is lighted by a small lamp. For this purpose, opposite to the point of the cone, there is an opening in the telescope, through which, regulated by a blind, light can be thrown on the point. As the latter is polished, a fine line of light thus appears, which extends to the centre of the field of vision, and the brilliancy of which can be altered by withdrawing the lamp to a greater distance or lowering the blind, so that even the faintest lines of a spectrum can be brought with facility and certainty into coincidence with this line of light. The head of the micrometer-screw is divided into 100 parts, and each part, in the neighbourhood of the Fraunhofer line F, answers to about ·00016 wave-length. The probable error of position on one of the well-marked lines in the sun’s spectrum amounts to about 0·008 of a turn of the screw with the lowest eyepiece of the telescope. I have subjected the screw itself to a thorough examination with reference to such range, as well as to periodical inequalities in the single worms of the screw, but could discover no error exceeding 0·01 of a turn of the screw. I have to mention, further, that after each observation in the position in which the instrument was used, readings followed on the sodium-lines, or on some of the hydrogen-lines, in order to eliminate errors which might arise in the unavoidable disturbance of any particular part of the spectral apparatus. 1. _Observations of the Aurora._ 1870, Oct. 25th.—A very bright Aurora. In the brightest parts, besides a very bright line between D and E, several other fainter lines were to be discerned, situated further towards the blue end of the spectrum. They appeared on a dimly-lighted ground, and stretched out over the Fraunhofer lines E and _b_ to about midway between _b_ and F. Towards the red end the spectrum was terminated by the bright line first mentioned. No measurements could be taken, as the apparatus had not yet undergone the above-mentioned alterations, and even the brightest line of the spectrum did not diffuse sufficient light to be able to perceive the fine cross wires. The red rays of the Aurora were not examined.