ANALYSIS OF SOLAR RECORDS 69 



1676 July and August, 3 spots seen. 



1677 April, a group visible. 



1678 2 or 3 spots seen. 

 1677-83 almost spotless. 



1680 May, June, and August, spots observed. 



1681 May and June, spots seen. 

 1684 Several groups visible. 

 1686 2-3 .groups seen. 



1688 several spots seen. 



1689 a few spots seen. 



1695 after "6 years of no spots," in May we find a "great spot; 



only great one in 11 years." 



1700 November, spots seen. 



17041 first occurrences of 2 groups. 



1705/ at once for 60 years. 



1707 2 groups seen twice 



1710 January-October spots visible. 



October 1710 to May 1713, no spots seen. 



1716-21 spot maximum in 1718. 



After this, maxima occurred near 1728, 1739, and 1750. 



These scattered observations have formed the basis of estimates of maxima 

 on Newcomb's 11.13-year cycle. It is probably safe to accept estimated 

 dates of maxima between 1610 and 1640 and from 1718 on, although they are 

 weighted by Maunder about 2 to 5 in comparison with a weight of 10 for 

 well-observed maxima. Any assignment of maxima during the dearth 

 seems unsafe, since great bursts of spots, even those visible to the naked eye, 

 have appeared in all parts of the cycle, though more likely to occur at maxi- 

 mum. We are convinced that no minima at all can possibly be assigned 

 during that dearth interval, as has been attempted. 



When the actual data are placed in a cyclogram as in Plate 15B, we find 

 a good solution for the entire series 1610 to 1930 at 11.2 and another at 11.6 

 years, giving us a choice of 28 or 27 cycles in that total interval. Each of 

 these solutions appeared in the automatic periodogram of the sunspot num- 

 bers made in 1913, Plate 12A. 1 Well-established maxima for the last 

 hundred years give a strong value at 11.4; thus it does not decide between 

 11.2 and 11.6. If, however, we take the obvious facts of the cyclogram, we 

 find for the series between 1615 and 1788 a solution at about 10.2 comparing 

 very favorably with 11.2 judging by the decreased average residual obtained 

 on applying Maunder's weights to the maxima. These two solutions can 

 be applied to those data and it will later appear that during the dearth a 

 10-year cycle appears in tree-ring records while the 11-year cycle seems to 

 have failed. 



Other Cycles in the Annual Numbers — In the sunspot cyclogram other 

 cycles appear prominently as the tests are carried through a range from 5 

 to 25 years. After the 10 and 11 year lengths, two others promptly attract 



1 Published in the Astrophysical Journal, October 1914, and in Climatic Cycles and 

 Tree Growth, vol. I, 1919, Plate 9A. 



