We must thus imagine that the radiation of the sun will be stronger in times of strong eruptive activity than during the days of low sun-spot frequencies. Direct observations of the intensity of the solar radiation which have been made by Saveljeff in Kieff confirm this assumption. It must be pointed out, however, that another phenomenon investigated by Köppen seems to contradict this conclusion. Köppen ascertained that in our tropics the temperature was by 0.32° Cent. (nearly 0.6° F.) lower during sun-spot maxima than the average, and that five years later, a year before the sun-spot minimum, it reached its maximum value of 0.41° Cent. (0.7° F.) above the average. A similar peculiarity can be traced to other zones, but on account of the less steady climates it is much less marked there than in the tropics. A French physicist, Nordmann, has fully confirmed the observations of Köppen. On the other hand, Very, an American astronomer, has found that the temperature in very dry (desert) districts of the tropics (near Port Darwin, 12° 28´ S., and near Alice Springs, 23° 38´ S., both in Australia) is higher at sun-spot maxima than at minima; but Very was in this research guided merely by the records of maximum and minimum thermometers. From Very’s investigation it would appear that the solar radiation is really more intense with larger sun-spot numbers.[11] This, it must be remarked, is only noticeable in exceedingly dry districts in which there is no cloud formation worth mentioning. In other districts the cloud formation which accompanies sun-spot maxima interferes with the simplicity of the phenomena. The cooling effect of the clouds seems in these cases by far to surpass the direct heating effect of the solar rays, and in this manner Köppen’s conclusion would become explicable. If we could observe the temperatures of the atmospheric strata above the clouds, their variation would no doubt be in the same degree as that in the desert. [Footnote 11: According to Memery (_Bull. Soc. Astr._, March 7, 1906, p. 168) an instantaneous rise of temperature is observed immediately when a sun-spot is first seen, and the temperature sinks again when the sun-spot disappears.] Finally, we have to note another period in the phenomena of the polar lights—the so-called tropical month, whose length is 27.3 days. The nature of this period is little understood. It is possibly connected with the electric charge of the moon. The peculiarity of this period is that it acts in an opposite way in the northern and southern hemispheres. When the moon is above the horizon, it seems to prevent the formation of polar lights; but for this case the difficulties of observation caused by the moonlight must, of course, be taken into consideration.