4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL, IO7 



and yet the very next month might show i to 3 days shifting of phase 

 from them. ■*' 



I made a prehminary study of the changes in the solar constant 

 since 1924 as observed at Montezuma, Chile. The loss of days of 

 observation and accidental errors hindered the fixing of the phases. 

 But I settled on January 3, 1924, as a date when the solar constant 

 probably began to rise. This date depended on use of the period of 

 6.6456 days, and of all the conspicuous occasions of observed rise of 

 solar radiation occurring in all months from 1924 to 1944. As I shall 

 show, January 3, 1924, did not prove finally to be a date when the 

 solar constant began to rise, but a date near maximum of periodic 

 variation. However, it is as good a date as any to base my tabulations 

 upon, and all the work about to be displayed is based on January 

 3.0000, 1924. 



3. TEMPERATURE TABULATIONS 



To fix ideas, I now employ tables and charts. For the convenience 

 of any who may wish to use the periodicity I have discovered, table i 



Table i. — Corresponding dates in 1924 for successive cycles of 6.6456 days 



contains all the periodic zero-dates in the year 1924, as I have used 

 them. 



Since 55 cycles of 6.6456 days comprise 365.5080 days, we may 

 obtain corresponding dates in any year next succeeding a year in 

 which we know the periodic dates, by the following simple rules. For 

 the 10 months, March to December, add 0.5080 day to give the new 

 dates, except for leap years. In case the next succeeding year is to 

 be a leap year, subtract 0.4920 day to give corresponding dates in 

 these 10 months in such leap years. For the months January and 

 February, add 0.5080 day, unless the year when dates are known is a 

 leap year. In such cases of leap years, subtract 0.4920 day to obtain 

 corresponding dates in January and February, in years next suc- 

 ceeding leap years. 



The decimal parts of days serve to indicate the nearest whole 

 number date. If the fraction is less than 0.5, use the integer of the 

 date as found. If over 0.5, use the next larger integer in tabulations. 



I now give, in table 2, a sample tabulation for the month of March 



