148 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1944 
Jan.+Feb.+Mar.+Apr.+ May 
5 
,etc. Then for the expected smoothed 
use the summation 
Feb. + Mar.+Apr.+May+June 
5 
percentage precipitation value for January of a future year, take the 
smoothed value for April 46 years previously, plus the smoothed value 
for February 23 years previously, and divide by 2. This simple rule 
works very well for some stations. Thus for Eastport, Maine, from 
1935 to 1942, inclusive: 
, and for April, 
For 96 months, average observed minus predicted___.___________ 16 percent. 
Number observed and predicted on same side of 100 percent______-____-__-___ 78 
Number observed and predicted on opposite sides of 100 percent___-_-___~ 18 
Number, though observed and predicted on opposite sides of 100 percent, 
their difference observed minus predicted, less than +16 percent___--____ 6 
YEARS s 
4 160 +++ Js SRD ON ND AT TR WaT, (aes ALP 
= 0 (a Ae a Gd SY A A Wl TE WB 
eee DAREN EAE PSY VERS [ND | LPS Pe dW PS Pe |] 
pe | ee ae EE a ay PH EL, a ro TAT ATI 
(hie int a OG 97) EY ek 
C72 VRE CY lS Al WR DS ODT 
iB Ie) 1 ey, | ee ee ee ee ee are, 
A TO BB, WP DP a 
LN Nha! el TAT WAN Wn fT ; 
27 A PL ON A 
tf NY 
: Le! | Lt 
SG | Pn | Ae ee ES 
A OT PN 
LWA YP AVN tg ORR od | 
RE AT TA ic P| a a | SY: 
SEA Ra (Na fae yA, a a a a FP Se 
MEERA A 0 a WE Sd A 
FO ME RF Ps | a HW |, 
7: Na i i a No AL fA \( 
WIN ATS AL TT Ve hg A eT A A | 
PTV AW WO TW ot TT he yCaRuIm IY Weyl 7 | OV 
RL aR YT a ih sie Bee) 
BE Se a Pe RS 
In Dac Gael WG a ot 
Ficurre 17.—Precipitation at Peoria, Ill., smoothed by 5-month running means, 
arranged in 23-year cycles. Letters represent similar features in successive 
cycles. Forecasts (dotted line, from 1934; thin line, from 1938) made by con- 
sideration of preceding cycles. 
1898 
1921 100 
From these figures one may fairly claim that of 96 months predicted 
84 were useful predictions, or a measure of success of 87 percent. 
For 12 New England stations similarly evaluated, 807 were useful, 
845 unsatisfactory months’ predictions. 
Several years ago, at the resquest of a Colonel of Engineers I 
made in this way from records of 10 stations a prediction for 3 
months of the expected precipitation in the Tennessee Valley region. 
My prediction of precipitation was 84 to 87 percent normal. The 
event was 87 percent normal. 
