2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. Ill 



ture, though the phases do not shift in solar variation. However, it 

 seemed to me probable that the period could be used for long-range 

 predictions of temperature, though such predicitions may sometimes 

 be out of phase by from i to 3 days. I propose to make a predic- 

 tion of Washington temperature for the year 1948 now [January 22, 

 1948J, seal and deposit it with the Smithsonian Treasurer, to be 

 opened and checked by the Director of the Smithsonian Astrophysi- 

 cal Observatory in January 1949. 



The enclosed graph ^ contains plots of the departures from normal 

 temperature at Washington for the years 1945-46-47. From a 

 marked repetition of the 6.6456-day period of March 1946 I observed 

 the phases of minimum temperature at that ^ time. Assuming no 

 changes of phase, I indicated corresponding dates of minimum tem- 

 perature for all repetitions of the period from January 1945 to 

 December 1947. As these indicators in most cases bounded curves 

 of higher temperature between each pair of indicators, I was en- 

 couraged to continue with my project. 



Using the method described on page 4 of the paper above cited, 

 and allowing for the fact that now I am concerned, not with "zero" 

 dates but with dates of minima, I compute the following dates in 

 the year 1948 as likely to be at or near dates of minima in the march 

 of temperature at Washington : 



1948 Jan. 



24 31 



30 



July .... 2 



Aug 4 



Sept 7 



Oct 3 



Nov 5 



Dec 2 



30 



28 



To verify the prediction I am about to make, the departures from 

 normal temperatures at Washington, as published monthly by the 

 United States Weather Bureau, should be taken for each of the dates 

 tabulated above, and also the departure occurring at the highest 

 temperature which occurred between each pair of dates. Then the 

 mean value for the year 1948 of the departures for the dates above 

 tabulated should be found to indicate a lower mean of departure 

 temperatures than the mean departure of the said dates of "highest 

 temperature." 



I will also predict what the difference will be in degrees Fahren- 



1 Not here reproduced. 



2 See figure herewith. 



