NO. 2 WEATHER FORECASTS FROM SATELLITES — ABBOT 3 



to 1955. This left the solar trends and their effects unaffected by this 

 extraneous variable. This is the main reason why our present results 

 on trends of temperature at Washington are smoother than those of 

 25 years ago. The 6.6476-day period would necessarily have to be 

 taken account of, along with the solar trend effect, in any 16-day fore- 

 casts which I hope may be made from satellite observations. 



Table 1 gives in eight groups, of 180 repetitions each, the average 



Table i. — Groups of 180 — repetitions of the 6.6476-day solar period and mean 



of 1,440 of them 



Places: 12 3 4 5 6 



1924-1927 466 462 465 474 472 454 



1927-1930 450 444 438 461 456 442 



I93I-I934 461 474 456 470 459 463 



1934-1937 485 4/0 459 475 475 467 



1938-1941 467 466 457 474 481 465 



1941-1944 465 457 461 466 462 461 



1945-1948 481 486 479 482 485 478 



1948-1951 504 493 489 510 501 482 



Mean 472.4 469.0 464.2 476.5 473.9 463.9 



amplitudes of these groups, and finally the general mean of 1,440 

 repetitions of the 6.6476-day period. It will be clear from the graphical 

 expression of these data in figure 1 that, though the phases of the 

 period hardly ever differ more than a day from their mean predicted 

 positions, their amplitudes are widely differing. Moreover, as pointed 

 out in my paper (Publication 3990 of 1949) and clearly shown in 



figure 1, a period of — — — days also exists. These factors would 



have to be allowed for in any 16-day forecasts from satellite obser- 

 vations. 



Mrs. Hill and I have eliminated the effect of these two regular 

 periodic solar variations. Our results on 16-day weather control by 

 solar trends are shown graphically in figure 2 and numerically in 

 table 2. 



With figure 1 I have also given an example of the large, and I may 

 even say almost controlling, effect of the period of 6.6476 days and 

 3.3238 days on the temperatures of Washington and New York. 



