152 Dr. 0. Chree : Auroral and 



comparable with Wolfer's mean annual sun-spot frequencies, 

 I have calculated from the Scandinavian monthly totals data 

 for years commencing in January. In addition to yearly 

 totals — from July to June — for the whole of Scandinavia, 

 Tromholt's Table E gives yearly totals for five subdivisions 

 of the country, numbered I. to V. according to latitude. 

 I. includes all districts north of 68°*5, II. extends from 68°'5 

 to 65°, III. from 65° to 61°*5, IV. from 61 3 «5 to 58°*5, while 

 V. includes the extreme south of Scandinavia from 58°"5 

 to 55°. 



For a considerable time subsequent to 1761, observations 

 from district I. were very few. a fact due probably more to 

 lack of observers than anything else. This possesses some 

 importance for the following reason. The annual variation 

 in auroral frequency is largely dependent on the fact that 

 aurora is seldom vivid enough to be visible until the sun is 

 several degrees below the horizon. In high latitudes there 

 is daylight throughout the whole 21 hours near midsummer, 

 and no daylight near midwinter, and the auroral frequency 

 in these regions, as was pointed out many years ago by 

 Lovering, has a single maximum near midwinter, and a 

 single minimum answering to a total absence of aurora near 

 midsummer *. This state of matters is at least approached 

 in district I., and to a lesser extent in district II. (cf. loc. cit. 

 Table G, p. 420). Further south in Scandinavia the annual 

 variation is similar to that in England, showing two maxima 

 near the equinoxes, a principal minimum at midsummer, and 

 a secondary minimum at midwinter. The mean annual 

 variation deduced for the whole of Scandinavia will clearly 

 depend to some extent on how far the several districts con- 

 tribute to the general result. Assuming an increasing 

 relative contribution from district I., if annual variations be 

 calculated from two different periods, one may not unreason- 

 ably expect the later period to show the equinoctial maxima 

 less prominently and the midsummer minimum more promi- 

 nently than the earlier period. This is one of my reasons 

 for contrasting one period of a special type with two of an 

 opposite type, the one earlier the other later. 



Lovering gives annual variations for a number of separate 

 stations. Most of these, however, are based on too few years' 

 observations to suit the present enquiry. 



Of the data for separate stations or districts, I propose to 

 use only those for New York State (I.e. p. 181). These 

 extend over 26 years, including 1205 separate observations, 



* Cf. A. Angot's ' Les Aurores Polaires," Paris, 1895, p. 12-5. 



