NO. 4 SUN S VARIATION AND TEMPERATURES — ABBOT 23 



Employing table i of my paper "Weather Predetermined by Solar 

 Variation," above cited, I selected 14 cases of large, and 13 cases of 

 small apparent solar-constant rise, which agreed in phase with the 

 6.6456-day period, as now settled upon. Unfortunately I could not 

 find any more extreme cases occurring when the records were suf- 

 ficiently complete to tabulate satisfactorily. The numbers tabulated 

 are far too small, therefore, to yield good curves. Corresponding to 

 all these 27 dates, I had already tabulated the solar-constant values 

 from 8 da}s before to 11 days after zeroth day. As explained above, 

 I had interpolated such values as were necessary to fill up gaps. I 

 computed separately the mean march of the solar constant for high 

 observed changes, and for low observed changes. The results are 

 as follows : 



Days — 8 —7 —6 —s —4 —3 —2 —I o i 2 3 4 5 6 7 8 9 10 ii 



Mean high values. 443 453 427 432 399 396 423 474 464 454 Remainder of curve is too irregular 



to be useful 

 Mean low values.. 466 468 446 438 444 444 452 478 455 456 447 444 441 462 468 441 44s 443 44s 44S 



The amplitude of the curve for high observed solar changes is nearly 

 0.5 percent, while the amplitude of the curve for low observed solar 

 changes is only about 0.2 percent. Both curves decidedly exceed the 

 amplitude, 0.13 percent, of the mean curve from 310 computed dates, 

 which I give in figure 9. This result, indeed, was to be expected. For 

 the 310 cases no doubt include many cases when the solar-constant 

 change was too small to recognize individually in the incomplete 

 solar-constant record, affected as it is by accidental errors. Had I been 

 able to use all the very highest apparently observed solar-constant 

 changes, many of which I omitted because of too incomplete recording, 

 no doubt the mean result, instead of being 0.5 percent, would have 

 been nearly as large as my estimate of 0.7 percent, made in 1936. 

 But, as stated above, no doubt some of the large apparent changes 

 used in 1936 were made larger by favorable trends of the accidental 

 errors of observations. 



It appears from this investigation that the amplitudes of solar vari- 

 ations of 6.6456 days period are indeed variable. Possibly this 

 variation in solar amplitudes may be reflected in the terrestrial tem- 

 perature responses. Unfortunately the solar-constant record is too 

 imperfect to decide this. Many other observing stations would have to 

 be permanently occupied to determine the solar constant with highest 

 possible accuracy to decide it. 



