78 REPORT OF THE SECEETAEY. 



Tlio rating of tlio cliarnctor of the observation is made rhlefly by inspection 

 of the logarithmic plots used in determining the atmospheric transmission, 

 similar to those published facing page 80 of your report for the year ending 

 June 30, 190o. On an excellent day the observations thus plotted, like those 

 just referred to. He close to representative straight lines, but such days are 

 unfortunately very rare in Washington, and especially within the last two 

 years, owing in part to building and other operations producing smoky and non- 

 uniform atmospheric conditions. The bearing of the results given in Tal)le 2 

 on the question of the variability of the sun has already been given. 



Before leaving the subject of solar constant observations it may well be 

 remarked that Mr. Fowle has published within the year a valuable paper giv- 

 ing a comparison between the solar constant values deduced by the method of 

 homogeneous rays and those deduced for the same days by the old method of 

 high and low sun observations of the total radiation of the sun by the 

 actinometer alone.« This old method, the method of Pouillet, gives necessarily 

 too low results, as you long ago demonstrated. & 



What is valuable in Mr. Fowle's paper is that he shows that the defect is 

 almost constant at Washington, no matter what the time of the year, the trans- 

 parency of the air, or the humidity, provided the extrapolation is made with 

 moderate solar zenith distances. Thus if observations be made in Washington 

 with the actinometer alone and extrapolated by the aid of a logarithmic plot 

 to the limits of the atmosphere, and a correction of 14 per cent is then added, 

 the result will be practically the same as if the spectrobolometric method 

 had been used. If additional measurements should confirm this result (that 

 V constant difference between the two methods holds here and at other locali- 

 ties), then the process of detecting variations in solar radiation outside our 

 atmosphere would be much easier, for it could be made to depend on actinometer 

 measures alone, and, indeed, old series of observations made years ago could 

 be utilized. 



I give in illustration the following ta!)le, taken in part from Mr. Fowle's 

 paper above cited, showing how closely solar constant values deduced from 

 pyrheliometer measures alone, by the application of the 14 per cent correction, 

 agree with values deduced when possible from spectrobolometric work of the 

 same days. 



Table 2a. — Solar constant values from Tpyrheliomeiry. 



Date. 



Range of 



air 

 masses. 



Log. a 



Radia- 

 tion com- 

 puted 

 from 

 pyrheli- 

 ometry. 



Grade of 

 pyrheliom- 

 eter obser- 

 vations. 



Radia- 

 tion com-I 



puted by Grade of spec- 



spectro- Itro-bolometric 



bolo- I observations. 



metric 

 method. 



1902. 



October 9 . 



1903. 

 February 19. . . 



March 25 



March 26 



April 17 



April 29 



July 7 .-- 



October 14 



December 7... 

 December 23-, 



1.5-3.0 



1.8-2.9 

 1.4-2.6 

 1.3-2.3 

 1.2-1.9 

 1.1-1.6 

 1.6-1.5 

 1.6^3.0 

 2.2-3.7 

 2. 2-3. 1 



0.113 



.114 

 .148 

 .133 

 .126 

 .170 

 .177 

 .107 

 .0&5 

 .077 



0.771 



.748 

 .676 

 .665 



.780 

 .822 

 .838 



2.18 



2.25 

 2.23 

 2.07 

 2.17 

 1.99 

 2.36 

 2.13 

 1.93 

 1.99 



Very good _ 



do_ 



Excellent -- 



do 



Very good . 



Bad 



Short, good 

 Very good . 

 Excellent .. 

 Doubtful... 



2.19 



Pair. 



0- F. E. Fowle, Smithsonian Miscellaneous Collections, vol. 



6S. P. Langley, American Journal of Science (3), XXVIII, p. 163, 1884, 



