NO. l8 SUN SPOTS AND WEATHER ABBOT 5 



the north and south solar hemispheres and that the order continues 

 unchanged through each ii-year sun-spot period, but reverses at the 

 beginning of the next period. Thus it requires two ii-year periods 

 to bring the sun through a full cycle of magnetic changes. 



The writer first noticed about July 1933 that the periodicities found 

 in solar radiation and in the weather were closely related to Hale's 

 magnetic cycle. But numerous studies led the writer to assign to it 

 the length of exactly 23 years, or 276 months. Dividing this period 

 into submultiples, we find as follows : 



Name a b c d e f g h i j k 



Divisor 3 46 8 11 13 15 i8 25 34 39 



Period, 



months 92 6g 4634-1/225-1/11 21-3/13 18-2/5 15-1/3 11-1/25 8-2/17 7-:/i3 



Of these periodicities, b, c, e, f, i, j, and k are (within the error of 

 determination) the same that I found in the variation of solar radia- 

 tion, and the others have been found in terrestrial temperature 

 departures. 



Inasmuch, therefore, as Hale's magnetic cycle is the least common 

 multiple of so many periodicities in solar and weather variation, it 

 seemed probable that the weather features would be found to repeat 

 themselves at intervals of 2;^ years. As an illustration, figure 3 shows 

 the smoothed percentages of normal monthly precipitation found at 

 Nagpur in South Central India from 1856 to 1930. The values are 

 arranged in 23-year cycles, so chosen that the year 1875 begins a 

 cycle so as to fit with most of the lists in " World Weather Records." 

 Lines have been drawn to guide the reader's eye to what seem to me 

 to be homologous features in the four cycles illustrated. I would like 

 to call special attention to the regions 1865-1870, 1888-1893, 1912- 

 191 7. In 1865, 1868, and 1870 we find three pillarhke features of 

 high percentage precipitation bounding two features of subnormal 

 precipitation. Thus there stand out two intervals of three and two 

 years, respectively, as if guarded by these sentinel features, but em- 

 bracing nearly a score of subordinate features. The reader's attention 

 is now invited to similar features, 1888- 1893, and 191 2- 191 7, in which 

 nearly all the details seem to be recognizable. 



The separation between the first and second of these occurrences 

 is almost exactly 23 years, but there is a delay of nearly a year in the 

 appearance of the third. A similar delay marks, however, all of the 

 features from 1899 to 191 8. after which the cycle returns approxi- 

 mately to its earlier phase-status. Compare, for illustration, the year 

 1929 with i860. 



