Measurements of Ultra-Violet Solar Radiation

T1ieultra-z~ioletsolarradiometer.-This instrument (P1. XIa) consists of a pair of quartz lenses, one silvered and one gilded, mounted on a rotating disk (Fig. i) electrically con- trolled in such a manner that images of the sun in X 0.32 and 0.5 /h are made to fall alternately upon the receivers of a compensated thermocouple for consecutive peri- ods of one minute. A complete determination of the ratio of ultra-violet solar radiation at Xo.32/~ to green radiation at Xo.5 ~ is obtained each four minutes of time, the gal- vanometer deflections being registered photographically (Fig. 3) together with an hour- angie scale recorded by the control clock (P1. XIb). Secular varialicm of the measurements.-The radiometer has been in operation on most of the clear days since June, 1924. Each galvanometer record has been measured, and the logarithms of the ratio ultra-violet to green, plotted against sec z (Fig. 3), furnish a determination of this ratio for the zenith and for no atmosphere for that day. A plot of the monthly means (Fig. ~) shows a considerable correlation with the sun-spot curve, but the amplitude (0.95-1.56) is too great to be explained by variations of solar temperature alone. Since no check observations are available, it is possible that at least a part of this amplitude is due to atmospheric or instrumental effects not under-. stood at present. Atmospheric and instrumental sources of error.-The effects of the variation of at- mospheric ozone on the radiometer measurements were tested experimentally with an ozone tube placed over the radiometer and were found to be negligible. The small varia- tion in atmospheric transparency (Fig. 4) does not seem to show any correspondence to the radiometer-curve. A certain white haze affects the measurements in the afternoon, and, on this account, only morning observations have been used. The transmissions of the metallic films (Fig. 2) seem not to have changed during the seven years they have been in use. Ultra-violet light from the sky.-By means of a silvered photoelectric cell it was found that the intensity on a horizontal plane of ultra-violet sky light at Xo.32/h is about equal to that of direct sunlight when the sun is near the zenith at Pasadena (eleva- tion 845 ft.). The same result was obtained at X 0.31 ~. Change in ekvation in- creases the direct sunlight about as much as the intensity of sky radiation diminishes at this wave-length. Seasonal and weather effects (Fig. 5) studied with an automatic device show that we receive on a horizontal plane about 25 per cent as much radiation of X 0.32 j~ on a cloudy as from sun plus sky on a clear day. The least amount observed was io per cent, recorded during a rainstorm. We receive about as much on a clear day at the winter solstice as during a cloudy day in midsummer. After the sun reaches an al- titude of 6°, the intensity of sky radiation at Xo.32 ~ on a horizontal plane increases pro- portionally with the solar akitude. The distribution of radiation at Xo.32 ~ over the sky (Fig. 7) is rather uniform, being somewhat greater toward the southern horizon under the sun and less toward the northern horizon. The sudden increase near the sun takes place at greater angular distances on hazy days. Ultra-violet limit of the spectra of skylight and direct sunlight.-A quartz spectro- graph and a concave grating, crossed with a quartz monochromator, were used. In midsummer the ultra-violet limit of skylight is about Xo.29ójh and of direct sunlight a little less than Xo.290/.L. In midwinter the limits are Xo.298 and o.296/~, respec- tively. It is believed that it would be difficult to detect the spectrum of the sky muc