SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I46 



Fig. 8a. — Solar constant measures repeat after 273 months. 



ence for 616 days is ±0.0076 calorie. In 1953, Mrs. L. S. Hill made 

 several similar comparisons. Mount Montezuma, Chile, is compared 

 with Table Mountain, Calif., and with Tyrone, N. Mex. Table Moun- 

 tain is compared with Tyrone. Her data cover, respectively, 891, 283, 

 and 202 identical days. 



From this abundant evidence, with no indication of appreciable 

 seasonal influence, I confidently take the average probable error of 

 one day's observation at one station by the "short method" to be 

 1/6 of 1 percent of the solar constant. 



In volumes 6 and 7, Annals, A. P.O., are published about 6,000 daily 

 values of the solar constant obtained by the short method, often by 

 two or three stations observing upon the same day. Fully half of 

 them are "very good" and at least as accurate as those referred to in 

 Table 1. The mean value for one month would rest on at least 10 of 

 them, and frequently 20 to 30. The month will be the unit I shall 

 use in the discussion to follow. Its probable error can justly be 

 regarded as 1/VlOx 1/6=1/20 of 1 percent of the solar constant. 



