Sun-spot Frequencies Contrasted. 153 



and represent a latitude much lower than that of Scan- 

 dinavia. The other data employed are from Lovering's 

 General Catalogue, pp. 195-200. This comes down to 1864, 

 and extends to earlier than the 14th century. Though 

 enumerating nearly 10,000 auroras it is probably, judging 

 by Tromholt's figures, very far from complete. The data 

 are doubtless of a very heterogeneous character, and the 

 same precautions appeared necessary as in the case of the 

 Scandinavian data. 



Recently Prof. Schuster * has gone very fully into the 

 existence of periodic variations in Wolfer's sun-spot fre- 

 quencies, and has concluded that there is evidence of the 

 existence not merely of the ordinarily recognized period of 

 11*125 years, but of others which like it are submultiples 

 of a period whose most probable value is 33*375 years. This 

 is one of the reasons why 1 have dealt with three successive 

 33-year periods, viz. 1761-1793, 1794-1826, and 1827-1859. 



§ 5. Before comparing sun-spots and auroras, I would draw 

 attention to some features of Table I., which gives mean sun- 

 spot frequencies for the several months of the year, as 

 calculated by me from Wolfer's table for a number of specified 

 combinations of years. It has been remarked by Mr. Ellis f 

 — who based his remarks on AVolfer^s data for ai-oups of 

 3 montlis (February to April, &c.) — that sun-spot frequency 

 shows no real annual period. The results in the first line of 

 Table I. cannot be said to be decisive against the existence 

 of a small annual term.thouoh inconsistent with the existence 

 of a large term having this period. The extent of the 

 difference between the mean results for the several months is 

 more easily realized in Table II., which expresses the monthly 

 values in Table I. as percentages of the arithmetic mean for 

 the 12 months. One would, I think, hardly have anticipated, 

 in means based on 153 years' data, the difference of 6 J per 

 cent, shown between the values for January and May. On 

 the other hand, if there were a true annual period, one would 

 expect the percentage figures in Table II. for the three suc- 

 cessive 33-years periods to resemble one another more closely 

 than they actually do. 



The monthly means in these three periods show very con- 

 siderable differences : in the earliest period, for instance, the 

 difference between the March and Xovember means amounts 

 to 20 per cent, of the mean value for the epoch. 



* Proc. Rov. Soc. A. vol. lxxvii. p. 145. 

 t Monthly Notices R. A. S. vol. lx. p. 142. 



