The author has observed a fact which furnishes a remarkable demonstration of the law of the production of the dark bands which he has detected in the spectrum of oxygen. The phenomena of elective absorption in oxygen gas are manifested in two mutually distinct spectral systems. A first system, formed of fine rays, follows the law of the product of the gaseous system traversed by its density. The second system is formed of bands much less easily resolved, is governed by the law of the product of the thickness by the square of the density. This second law being quite novel in spectral analysis, the author has instituted experiments necessary to prove that this system of obscure bands really belongs to oxygen. These experiments range from pressures of 100 atmospheres down to those of a few units, and with lengths of tubes from 0.42 meter to 60 meters. At the same time prolonged observations have been made upon the atmosphere, brought into connection with the experiments in the tubes. These observations, and especially those made during autumn last on the Pic du Midi, prove that all the bands of the spectrum of oxygen are found in the spectrum of the solar light if it is allowed to traverse a sufficient thickness of the atmospheric medium. Further, on comparing, by the aid of photography, the intensities of the bands of the atmospheric spectrum with those given in the tubes, the author has found that the intensities of these atmospheric bands fulfill the law of the square. It appears from _Wiedemann's Annalen_ that M. Olszewski, when liquefying oxygen, examined its spectrum and ascertained the existence of the bands in question with a stratum of 7 mm. of liquid oxygen.--_J. Jansen._ ON A THEORY CONCERNING THE SUDDEN LOSS OF MAGNETIC PROPERTIES OF IRON AND NICKEL. By Mr. H. TOMLINSON, B.A. Experiments by himself and other observers have shown that the temperatures at which iron and nickel lose their magnetic properties depend on the specimens used and the magnetizing forces employed; but the temperatures at which they _begin to lose_ these properties are definite--for nickel about 300° C., and iron about 680° C. The author's own experiments on "Recalescence of Iron" show two critical temperatures; and Pinchon has shown by calorimetric measurement that between 660° and 720° C., and between 1,000° and 1,050° C., heat becomes latent. All these facts seem to indicate a molecular rearrangement about these temperatures.