226 ON THE PERIODICITY OF THE AURORA BOREALIS. 



as soon as auroras became more frequent and more splendid. How much or how little 

 our confidence in the mean curve can reasonably be impaired by interruptions in the " 

 data may be found by comparing the mean curve with the particular curves, for special 

 places, and also by examining to what extent many observers simply duplicate each 

 other's observations, without adding to the number of recorded auroras. If an exami- 

 nation is made of the records of auroras seen since 1700, it reveals the fact that the dis- 

 plays reached a maximum of frequency about the year 1730. The St. Petersburg 

 series agrees with the general table (after excluding cases which fall within Mairan's ex- 

 ceptions) in assigning the maximum to the year 1730. The miscellaneous observations 

 which Mairan groups under the title of " Philosophical Transactions of London," place 

 this maximum as early as 1726, and the Wittemberg series as late as 1742, though 

 there was another inferior maximum in 1731. In Berlin it was the year 1732, in Eng- 

 land, according to Huxham, it was the year 1736, and in Italy it was the year 1737 

 which claimed the maximum. The series of auroras observed in Sweden discloses one 

 maximum in 1730, and another slightly greater in 1741. 



This maximum was followed, after an interval of about twenty-five years, by a mini- 

 mum, which, on the average of all the reliable series, belongs to the year 1758. The 

 only four series of observations which extend over this period of time correspond 

 strikingly with each other, as also with the mean. The Cambridge series gives a mini- 

 mum to the year 1755, the Swedish series to the same year, the St. Petersburg set to 

 the years 1753-4, and the observations at Abo to the year 1757. Another sudden and 

 low depression in the year 1766 is worthy of mention. Not a single aurora was ob- 

 served that year either in Cambridge or St. Petersburg. One was seen in England and 

 six at Abo, this secondary minimum appearing the next year at Abo, when only four 

 auroras were observed ; whereas in 1757, the year of the principal minimum, there was 

 one solitary observation at that place. 



About the year 1787 another maximum is reached. The observations made at Abo, 

 Ratisbon, Prague, Mannheim, and Paris agree separately on the same year as the mean 

 curve gives. The series of Cambridge, Berlin, and Brussels transfer this maximum to 

 the year 1786 ; those of Jena, Kendal and Keswick, and Sagan, to the year 1788 ; those 

 of St. Petersburg to 1786, though a more decided maximum occurred at that place in 

 the year 1774 ; the observations at New Haven bring this maximum as early as 1781, 

 but another maximum almost as great is found in the year 1786, whereas those at 

 Stockholm postpone it to 1790. There is a maximum in the series at Montmorenci in 

 the year 1787, but a greater one in 1779 ; a maximum appears to have come at least as 

 early as 1779 at Carlsruhe. At Copenhagen one maximum occurred in 1787, and a 



