206 ON THE PERIODICITY OF THE AURORA BOREALIS. 



whole earth, in October, though the time varies all the way from August to November 

 at different places. The winter minimium is fixed by the whole mass of observations 

 in December, but this also oscillates between November and January according to the 

 locality. 



Table XLVII. contains the principal series of observations upon the aurora which 

 have been made in the United States of America, in Canada, and in Greenland ; the 

 next contains similar series of observations made in Europe. It will be understood 

 that in both cases the monthly aggregates express not the number of auroras, but the 

 number of observations, the same aurora being frequently seen at more than a single 

 place. The monthly maxima and minima in both tables fall upon the same montliSj 

 with the exception of the autumn maximum, which for the Eastern Continent is in Sep- 

 tember, and for the Western Continent in October. 



In Tables XL1X. and L. the mean epochs of the monthly maxima and minima of the 

 year are arranged in the order of the geographical latitudes, on each continent 

 separately. These tables do not exhibit any connection between the changes in the 

 times of the maxima and minima and the position of the observer. I may add that 

 the limits within which the times of maxima and minima fluctuate, in passing from one 

 station to another on the earth's surface, would be even wider apart than I have placed 

 them, had I not excluded from the discussion special localities, where, from the extreme 

 influence of twilight, the poverty of the observations, the short period of time which 

 they cover, or the unfavorable attitude of the epoch in relation to the secular changes 

 in the phenomenon, the diurnal curve had lost all its ordinary characteristics. These 

 remarks apply to the observations made at St. Martin, Burlington, Boston, Cambridge, 

 Salem, Wilmington, Dunse, and Brussels. The yearly curve for Salem has its principal 

 maximum in July, and two other maxima in March and October; and its lowest min- 

 imum in December, with another, less conspicuous, in May and June. Some of the 

 auroras observed at Salem in July are characterized as extraordinary. I have selected, 

 in Table LI., from the auroras seen at New Haven those which belong to the period of 

 time embraced by Dr. Holyoke's observations, and I find in them a principal maximum 

 in July, with other irregularities as great as those noticed in the Salem observations, 

 though not exactly corresponding to them. Of the ninety-seven auroras seen at St. 

 Petersburg in the years 1786 to 1820 inclusive, the maxima fall, as usual, into the 

 months of March and October, and the minima into the months of January 

 and June. Only one aurora was seen in July at St. Petersburg, but thirty were 

 observed in Salem. Dalton's two catalogues of auroras for Kendal, Keswick, and 

 Great Britain refer to places not extremely different in latitude from Salem, and 



