SOLAR RADIATION ABBOT 111 



Fifty years ago Langley invented the bolometer, an exquisitely 

 delicate electrical thermometer sensitive to a millionth of a degree. 

 He took it to Mount Whitney, Calif., in 1881, to measure the rays 

 of the solar spectrum under the purest of skies. He also perfected 

 and ap23lied a method for determining the losses suffered by solar 

 rays in traversing the turbid and absorbing ocean of atmosphere 

 which always overlies even the choicest of observing stations. With 

 some improvements, we still use the Langley bolometer and the 

 Langley method. 



For the past 12 years the Smithsonian Institution has maintained 

 stations on high mountains in desert lands where daily measurements 

 of the solar radiation are made. Our best station, at Mount Monte- 

 zuma, Chile, 9,000 feet in altitude, lies in a desert where rain seldom 

 falls and where neither animal nor vegetable life can exist. The ob- 

 servers must bring even water itself from the town 12 miles distant. 

 The observations are carried on in such a way that the losses caused 

 by the atmosphere are determined accurately. Thus we are able to 

 measure the intensity of solar radiation as it would be found outside 

 our atmosphere altogether, as if one were on the moon, for instance. 

 We allow for the ellipticity of the earth's orbit, and thus reduce the 

 results to a constant solar distance. Following long custom, we call 

 the resulting value " the solar constant of radiation." It is on the 

 average 1.94 calories per square centimeter per minute. 



Yet the solar-radiation intensity is not perfectly constant. It 

 varies through a range of several per cent. Figure 1 shows how 

 two of our stations, one at Montezuma, Chile, the other at Table 

 Mountain, Calif., agree within 0.2 per cent in tracing the variation 

 of it by their monthly mean results over the past five years. The 

 total range of variation shown by these monthly means is 1.5 per cent. 

 Figure 2 shows in curve A the monthly Monteziuna values since 1918. 

 The total range in this period is 2.5 per cent. Curves C, D, E, F, G 

 are regular periodic curves of 68, 45, 25, 11, and 8 months whose sum, 

 given in curve B, almost exactly reproduces the variation shown in 

 the original observations given by curve A. Other shorter periods 

 may be found in solar variation, as indicated for the year 1924 in 

 curve H. I have ventured to forecast in curve I the probable march 

 of solar variation in the years 1931 and 1932.^ 



Short-interval changes are also found as shown in Figure 3. I 

 have indicated b}'^ curved lines, full and dotted, respectively, over 100 

 cases each of rising and of falling sequences extending over several 

 daj^s each. The smallest ranges considered in these short-period 

 changes are 0.45 per cent, and the largest found is 2.5 per cent. I 

 have compared with these sequences the weather of Washington, 



' Footnote added January 1933 : The observations of 1931 and 1932 closely verified 

 this prediction. 



