178 ANNUAL REPOKT SMITHSONIAN INSTITUTION, 1931 



Hence, in the year 1902, we began a solar-constant research destined 

 to be a very long one. At that time solar-constant values ranging 

 from 1.76 to 4.0 calories were given in standard textbooks. 



Our earliest Washington work raised a question which has en- 

 grossed us for over 25 years. Our results of 1902, 1903, and 1904 

 seemed to indicate that the sun's output of radiation decreased rather 

 suddenly in March, 1903, by about 10 per cent, and continued low 

 thereafter. We should certainly have attributed this to obscure 

 error if the temperature of the entire north temperate land area of 

 the world had not shown a decrease at the same time. We think now 

 that we were misled by an atmospheric turbidity caused by a vol- 

 canic eruption in southern Mexico. At all events, we began then to 

 suspect that the sun is a variable star and that its fluctuations pro- 

 duce important weather changes. I hope now, after more than 25 

 years of investigation, to present evidence to convince you that such 

 is really the case. 



The determination of the solar constant of radiation involves two 

 principal parts. First, the exact measurement of the intensity of 

 solar radiation as received at the observing station ; second, the exact 

 estimation of the loss which the rays suffer in traversing the atmos- 

 phere. The first requirement involves an accurate pyrheliometer. 

 The second requirement involves exact measurements of the atmos- 

 pheric transmission coefficients for all important solar spectrum rays. 

 Besides this, there must be estimates of the relative transmissibility 

 of the spectral rays in the optical apparatus, and of the atmospheric 

 transmission of those feeble parts of the solar spectrum lying in the 

 ultra-violet and the infra-red beyond the limits of the spectrum 

 region usually observed. 



THE SILVER-DISK PYRHELIOMETER 



When we began solar-constant work in 1902, the beautiful electrical 

 compensation pyrheliometer of Knut Angstrom was already avail- 

 able, though not fully perfected. Following, however, in Langley's 

 path, we developed for our use the older form of pyrheliometer of 

 Pouillet and Tyndall. With us it became in 1910 the well-known 

 silver-disk pyrheliometer, of which the Smithsonian Institution has 

 since furnished more than 60 standardized copies at cost to solar- 

 radiation observers in all parts of the world. 



The silver-disk pyrheliometer, though simple, effective, and ac- 

 curate, and capable of maintaining a constant scale of readings for 

 many years, is not an independent standard. Means are required 

 to reduce its scale to true calories per square centimeter per minute. 

 For this purpose we developed the standard water-flow and water- 



