NO. 17 WASHINGTON AND NEW YORK WEATHER — ABBOT 3 



precipitation is scarcely above 50-50. Recent press accounts of sur- 

 prising accuracy in these predictions for individual days of past years 

 refer merely to lucky happenings. 



The basic statistical tabulation from 1924 to 1941, to which I re- 

 ferred above, and on which table 2 is based, began January i, 1924. 

 The length deduced for the cycle is 27.0074 days. In 352 cycles of this 

 length there are 9506.6048 days. In the years 1924 to 1949, inclu- 

 sive, there were 9497 days. Hence the Roman cycle date I falls on 

 January ii, 1950, as given in table 2, being 10 days later in January 

 than the original Roman cycle date I, which fell on January i, 1924. 



B. TEMPERATURE AT WASHINGTON 



In previous papers I have drawn attention to a regular periodic 

 variation of 6.6485 days' length in the output of radiation from the 

 sun.^ Though quite regular intervals occur in the solar variation, 

 terrestrial responses thereto are sometimes i, 2, or rarely 3 days from 

 their expected dates. This is due to the complexity of the atmos- 

 pheric constituents and reactions. All terrestrial responses to solar 

 impulses are subject to lag. For instance, the warmest part of the 

 day occurs from i to 6 hours after noon at various stations of the 

 earth. The lag is not constant from day to day at any station. Hence, 

 from analogy, the irregularity of terrestrial responses to the 6.6485- 

 day solar variation is not surprising. Nevertheless they are notable 

 in magnitude. As shown in earlier papers they range from 2° to 

 20° F. in the temperature of Washington. This statement will be 

 found confirmed in figures 2 and 3. 



Notwithstanding the differences in lag just referred to, which cause 

 displacements of the terrestrial responses, it seemed to me worth 

 while, in January 1948, to predict for the ensuing year the 55 dates 

 when minima of temperature with respect to surrounding days might 

 be anticipated in Washington. In doing so I recognized that actual 

 minima would sometimes fall i, 2, or even rarely 3 days from the dates 

 predicted. In January 1949 the prediction was compared with the 

 event. Figure i shows the numbers of days when the observed minima 

 coincided, or fell i, 2, or 3 days from the predicted dates in 1948. 



The published predictions for 1948 and 1949 (above cited) were 

 made with the original value of the length of the period. I now give 

 in table 3 new dates to replace those published for 1949 in a previous 



3 Smithsonian Misc. Coll., vol. 107, No. 4, 1947; vol. in, No. 6, 1949; vol. in, 

 No. 13, 949. The period was originally determined as 6.6456 days. But in the 

 last of the three papers cited a correction of 0.0029 days was found, thus making 

 the preferred period 6.6485 days. 



