NO. 3 SOLAR VARIATION AND WEATHER — ABBOT 59 



every time holographs of the solar spectrum were made for measur- 

 ing the solar constant of radiation. This continued until 1920, when 

 we removed from regular observation on Mount Wilson. 



Our purpose was to determine if changes in the solar constant are 

 accompanied by correlated changes in the U-shaped drift curves. 

 Dr. Langley had hoped that they would be and that drift observa- 

 tions would serve as an easy means to measure solar variation. 



The entire series of comparisons between solar-constant variation 

 and solar-contrast variation is studied in chapter 7 of A.P.O. Annals, 

 volume 4, pages 217-258. The discussion of this long series of 

 careful measures gave conflicting results, hard to understand. Some- 

 times it indicated increased solar constant with increase of contrast 

 in brightness between center and limb of the sun. Indeed, for the 

 results of 1913, the correlation coefficient was -1-0.601 ±0.067, and 

 a change of 1 percent increase of the solar constant was accompanied 

 by +17 in the arbitrary solar-constant number. But at some times 

 even the sign of the correlation coefficient changed from plus to 

 minus. So the hope that Dr. Langley had held before his death in 

 1906 proved illusory. Solar-contrast observations did not yield an 

 easy way to measure the variation of solar radiation. 



In May 1952, however, P. 4088 threw new light on this diffi- 

 culty. We then knew of the family of harmonic periods in solar 

 variation. Synthetic solar constant values computed from these 

 periodic terms marched in close accord with observed values from 

 1920 to 1951 (P. 3902). So good was this agreement that I com- 

 puted the probable march of the solar constant from 1900 to 1920, 

 the years before good determinations had been possible. For, as I 

 have said, the Langley solar-constant method, though sound and 

 fundamental, must always give values too high or too low if the 

 transparency of the atmosphere changes during the several hours 

 required to measure it. The synthetic solar-constant values (see 

 fig. 50) were based on "the short method" which has no such draw- 

 back, and besides gives several values of the solar constant on each 

 day of observation, thus providing mean values. (See pp. 61, 62). 



Figure 50 shows that before 1920 there is no visible correlation 

 between the observed and the synthetic Mount Wilson solar constant 

 values. But figure 51, in its graphic comparison of synthetic solar 

 constant with observed solar-contrast values, shows a fairly high 

 degree of correlation. Increased contrast goes with increased solar 

 constants. So if the A.P.O. was still in short-method operation as 



