SOLAR VARIATION AND WEATHER—ABBOT 147 
ture records at terrestrial stations, to obtain more exact periods of the 
solar changes than could be fixed by solar-constant measurements ex- 
tending only since the year 1920. 
New Haven 
aA [ot Pa Os We OT 
ence t| Et Pohl A pode Nee 9 
SHA 
Jae eee eee 
| 
TIACACLAI Te 
PEE CEA EC ae 
ML. 
Figure 16.—A periodicity of 84% months in temperatures at Copenhagen, Vienna, 
and New Haven, Conn., since the year 1700. Seasonal phase disturbances are 
excluded. 
PERIODICITIES IN WEATHER 
Since the 14 periods® simultaneously active in solar variation are 
approximately aliquot parts of 273 months, we may anticipate that 
the many weather features occurring at a station in this interval of 
nearly 23 years will tend to repeat with some measure of similarity 
in successive 23-year cycles. Experience shows that this influence 
is more effective at some stations than at others. Figure 17 shows 
what has happened at one of the most responsive stations, Peoria, IIl. 
It will be seen that especially in the last half of the cycles the tendency 
of features to repeat in Peoria precipitation is quite marked. Two 
attempts to forecast, made in 1934 and in 1938, are shown by heavy 
dotted lines, and by light full lines, respectively, in figure 17. 
I have made use of this 273-month master period to predict 
precipitation for some years in advance at a number of limited 
regions of the United States. I reduced the prediction to a purely 
routine computation, and used the percentages of normal precipi- 
tation smoothed by 5-month consecutive means. Thus for March 
5To them must be added the sunspot cycle of approximately 723 months. For though 
it does not appear in variations of the solar constant, the ionic: bombardment of the earth 
from sunspots is not negligible as a weather factor. 
