8 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 1 10 



as between the means of the first 9 and the last 9 of the solar-constant 

 values of the sequences. 



In order to illustrate the variation in effect of magnetic storms on 

 solar radiation depending on the magnitude of sunspots and their loca- 

 tion on the sun's disk, I ask the reader to compare plate 2 with plate i. 

 Plate 2 includes direct photographs of the sun taken at Mount Wilson 

 on November 28, 1936, and February 3, 1937. On these dates there 

 was no central sunspot group, as on March 22, 1920. On November 

 28, 1936, two large sunspot groups were at about 20° solar latitude 

 both north and south of the center of the sun's disk and another near 

 the sun's limb. On February 3, 1937, there were many small spot 

 groups upon the disk, and one very large one near the sun's limb, but 

 none near the center of the disk. Accordingly we see from the solar- 

 constant records a very large depression in March 1920 (see fig. i), 

 a conspicuous depression in November 1936, and scarcely any de- 

 pression in February 1937 (see table i), at the times of severe mag- 

 netic storms. 



MAGNETIC STORMS AND SKY CONDITIONS 



Thus the magnitude of the Rayleigh depression of solar radiation 

 resulting from 93 million miles of ionic shower proves measurable. 

 Since these ions invade the earth's atmosphere, we may look for two 

 meteorological effects. First, the captured ions are likely to act as 

 centers of condensation of water molecules and dust, and thereby 

 increase the haziness and the brightness of the sky. Second, the sur- 

 face temperatures of the earth might be affected. 



From table 24, Annals, volume 6, and unpublished later records, I 

 collected for 30 magnetic-storm dates the pyranometer measures at 

 air mass 2.5 of the brightness of the sky near the sun. The mean 

 values for these dates and the 10 days before and 10 days after, 

 together with the numbers of observations entering into each mean, 

 are given in table 2 and graphically in figure 3. 



It appears that the haziness of the sky increased suddenly on the 

 storm day,^ and sky brightness near the sun averaged 10 percent 

 higher for the 10 last days of the sequences than for the first 10 days. 

 As could be expected, the graph, figure 3, is rather irregular. It must 

 be considered that the principal causes of sky haziness lie in the lower 

 layers of the atmosphere, and are subject to great fluctuations as dust 

 and humidity float about in the changing air currents. Hence the 



3 This tends to explain the drop in atmospheric transparency shown in 

 figure I. 



