The ordinary solar spectrum extends, as is well known, to about λ2913, the more ultra-violet parts being cut off by ozone absorption in the upper atmosphere. We have thus no direct knowledge of the distribution of intensity in the solar spectrum beyond λ2913, as it will appear to an observer situated outside the atmosphere of the earth. But it is now recognized that a number of physical phenomena is directly caused by the photochemical action of this part of sunlight on the constituents of the upper atmosphere. Such phenomena are (1) the luminous spectrum of the night sky and of the sunlit aurora, (2) the ionization in the E, F and other layers which is now being intensely studied by radio-researchers all over the world, (3) the formation and equilibrium of ozone (see Ladenburg 1935), (4) magnetic storms and generally the electrical state of the atmosphere. Formerly it was a debatable point whether some of these phenomena were not to be ascribed to the action of streams of charged particles emanating from the sun. There seems to be no doubt that the polar aurora and certain classes of magnetic storms are to be ascribed to the bombardment of molecules of N
2
and O
2
by such charged particles, for these phenomena show a period which is identical with the eleven year period of the sun, and are found in greater abundance, the nearer we approach the magnetic poles. But there now exists no doubt that the ionization observed by means of radio-methods in the E and F
1
regions, their variation throughout day and night, and at different seasons is due to the action of ultra-violet sunlight. This was decisively proved by observations during several total solar eclipses since 1932 (Appleton and Chapman 1935). The luminous night-sky spectrum, though it has certain points of similarity to the polar aurora, is on the whole widely different, and is found on nights free from electrical disturbances. The prevailing opinion is that it is mainly due to the ultra-violet solar rays, i. e. in the course of the day sunlight is stored up by absorption by the molecules in the upper atmosphere, and again given up during the night, in one or several steps, as a fluorescence spectrum. According to S. Chapman (1930) the formation of the ozone layer and its equilibrium under different seasonal conditions is also to be mainly ascribed to the action of ultra-violet sunlight. In the following paper an attempt will be made to discuss some of these questions in as rigorous a way as is possible with our present knowledge. It is evident that an adequate discussion is possible only if we have a good knowledge of (1) the distribution of intensity in the solar spectrum beyond λ2900, (2) the photochemical action of light of shorter wave-length than λ2900 on the constituent molecules of the upper atmosphere, which are mainly oxygen and nitrogen. We shall first consider (1).
VI. The ultra-violet band of ammonia, and its occurrence in the solar spectrum