ON THE PERIODICITY OF THE AURORA BOREALIS. 207 



to nearly the same period of years, but they do not exhibit the anomalies which are 

 so prominent in the Salem curve. The Brussels catalogue shows a large number 

 for July, and is otherwise irregular. A sufficient number of observations was not ob- 

 tained at Montmorenci and Paris, during the interval of time under consideration, to 

 warrant an argument from them. 



The yearly curve for Cambridge is also singular. According to the mean values 

 of all the observations, the ordinary maxima occur in March and September, and the 

 minima in June and December or January. But the number of auroras observed in 

 July is unusually large ; and if we take Winthrop's observations, amounting to two 

 hundred and fifty-four auroras, by themselves, the principal maximum is in July, with 

 inferior maxima in March and September, and minima in May and January. Tbe 

 yearly curve for this period, between tbe years 1742 and 1779 inclusive, resembles the 

 Salem curve, though the latter refers to different years. The number of auroras ob- 

 served at New Haven between the years 1763 and 1864 is as great for July as for the 

 spring maximum. The observations made at St. Petersburg, between the years 1742 

 and 1779 inclusive, come to their maxima in April and October, and to their minima in 

 July and January. The whole number of auroras is four hundred and fifty, of which 

 only two were in July ; whereas, out of two hundred and fifty-four auroras observed by 

 Winthrop, thirty-one were in July. If we acid Wigglesworth's observations to those of 

 Winthrop, so as to extend the period under consideration to the year 1793 inclusive, we 

 shall reach the same general conclusion. A comparison of the Cambridge curve with 

 those of Berlin, Jena, Paris, and Montmorenci, as well as with those of St. Petersburg 

 and Upsal, proves clearly that the differences above indicated cannot be wholly explained 

 by inequalities in the length of daylight and twilight. The yearly curves for Boston, 

 Wilmington, and Brussels are also remarkable, but they are based on scanty observations, 

 and the peculiarities may be only accidental. The Brussels curve, drawn in accordance 

 with the ancient observations, differs considerably from that which would represent Que- 

 telet's observations at the same place. At Hammerfest, Jakobshavn, Gothaab, Fort Re- 

 liance, and in Scandinavia the winter minimum and the spring and autumn maxima are 

 almost obliterated by the great changes in the length of clay and night in different parts 

 of the year. 



In different places the ratio of the maxima and minima numbers varies greatly. 

 In the general table the spring maximum is five per cent, larger than the autumn 

 maximum, and the December minimum is nearly three times larger than the June mini- 

 mum, and twenty per cent, smaller than the spring maximum. In particular places, 

 where the June minimum is zero, the ratio between maximum, and minimum becomes 



