Sun-spat Frequencies Contrasted. 159 



auroral frequency is connected with that oi: sun-spots by a 



formula oi: type (1), then the constant a does not vanish : i. e.. 

 a total absence of sun-spots would not be accompanied by a 

 total absence of auroras. A difficulty, however, arises here, 

 which the fio-ures in Table IV. for the several districts of 

 Scandinavia will serve to explain. 



The ratio borne to the frequency of auroras in the average 

 year of the period 1794-1827 by the corresponding frequency 

 for the combined periods 1761-94 and 1827-60 is roughly 



2 : 1 for the whole of Scandinavia ; but is 9 : 1 for district V., 



3 : 1 for district IV., 3 : 2 for district III., and less than 

 1 : 1 for district II. Any comparison for district I. would 

 be misleading. 



These figures suggest that with decrease in sun-spot 

 frequency the diminution in auroral frequency is enormously 

 greater in the south than in the north of Scandinavia. 

 Several authorities have called attention to analogous pheno- 

 mena, and the theory has even been advanced that the 

 difference in auroral frequency in years of many and few 

 sun-spots really arises from the alternate expansion and 

 contraction of Fritz's isochasms * (curves of equal auroral 

 frequency). The theory does not seem to be strongly 

 supported, but there seems little if any doubt that a sub- 

 stantial difference really exists between the long period changes 

 of auroral frequency in different regions. Results from several 

 stations in Greenland — whose substantial accuracy seems 

 accepted by Prof. A. Paulsen, one of the leading authorities 

 on the subject — appear to indicate that auroral frequency is 

 there very considerably less when sun-spots are many than 

 when they are few. 



§ 9. We now pass to the consideration of the figures from 

 Lovering's general catalogue in the last five lines of Table III. 

 As before, the monthly figures are expressed as percentages 

 of the yearly total. The figures corresponding to the whole 

 period covered by the table are given by Lovering f himself. 

 Though derived from very heterogeneous data, they represent 

 very fairly the type of annual variation which is character- 

 istic of lower temperate latitudes. The midsummer minimum 

 and the equinoctial maxima are less pronounced than thev 

 are even in the South of England. 



Comparing the totals from the two periods 1761-93 and 

 1827-59, we see that the excess from the later period is even 

 greater than it was in the case of Scandinavia. The annual 



* Cf. Angot, ' Les Aurores Polaires,' p. 138. 



t L. c. pp. '200 & 216. The total for December oa p. 200 should 

 apparently be 907, and not 1007 as printed 



