140 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1944 
that a fair estimate of the dates of changes in solar activity could be 
made. With allowance made for the difference of 2 days in phase, the 
basic curves used were those derived for Washington temperatures 
' from solar-constant work. The general result was as summarized 
below: 
Days 
Totals :predicheds 2s a te Sik 8 a Se es or eae a en ee 201 
Observed and predicted: ‘sameysign 22 oS se ee ee ee 139 
Observed and: predicted, opposite Signa ee ee 62 
Observed departures: Plus 65, minus 136. 
Predicted departures: Plus 64, minus 137. 
Num- Percent- 
bers ages 
Differences VOLO 2 eee eee le Ce ee eee eee en 66 82.7 
STAM KO Bae: aed ae Rell Paes Rec eal US eat 43 21.3'6°, 69.8% 
Be COT Oe a Ue ot 2 J Be 32 15.8 
General oa 0d 0 Yt SG) WD ae PE PR EE ee 33 16.3 
mean= aL AIA oC 014s gpk ep SE Uk SP AN Ps OE SL es a 20 10.0'6°, 30.3% 
5°.35. Oven toe Sie Pane Ua ee Ae eae 8 4.0 
Correlation coefficient=56.9+3.2 percent. 
This preliminary test, which is a forecast based on solar data alone, 
gives some ground for hope that with more accurate and continuous ob- 
servation of solar-constant values, when these are obtained every 
single day, such solar forecasts, supplemented and corrected by the ex- 
tensive knowledge of terrestrial influences now available to meteorolo- 
gists, may in that combination greatly promote longer-range weather 
forecasting. Since solar changes are a major weather factor it is 
difficult to see how long-range weather forecasts can be made if they 
are neglected as always heretofore. 
THE 27-DAY PERIOD IN WASHINGTON PRECIPITATION 
I now present a curious result of investigation of the sunspot rota- 
tion period of 27 days in connection with the precipitation at Wash- 
ington. In the year 1942 I collected values of the daily precipitation 
at Washington from 1924 to 1941. These values I arranged in cycles 
of 27 days. Since 27 such cycles fill 2 complete years, lacking 1 or 
2 days depending on leap year, it was convenient to tabulate the values 
in nine 2-year tables, and take the mean values for each of them. 
I was immediately struck by the circumstance that for the mean of 
every 2-year tabulation, the 11th day of the cycle in the earlier years 
and the 12th day of the cycle in the later years was from 2 to 3 times 
as rich in precipitation as the 6th and 7th day. The cycle, whose true 
period seems to be 27.0074 days, was always taken in the phase as of 
January 1-27, 1924. On taking the general mean of 243 cycles, the 
characteristic of high values about the 12th day was very marked, but 
other parts of the cycle also were conspicuous as high or as low in 
