May 19, 1 9 10] 



NATURE 



355 



When we pass to high latitudes the preeminence of the 

 uinox as a season for magnetic storms seems to dis- 

 appear entirely. This is shown by Table V., which com- 

 pares declination results at Kew and at the Discovery's 

 winter quarters. 



T.\BLE V. 



' rcentage of Days having Range above 20' at Kew, and 

 ibove 120' at the "Discovery's" Winter Quarters 



w7° 51' S.). 



At Kew, out of every 100 days at midsummer (May to 

 July), only nine had an absolute range above 20', the 

 corresponding figure for the four equinoctial months being 

 sixteen, or nearly double ; but in the Antarctic eighty- 

 one out of every 100 days at midsummer had a range 

 exceeding 120', while the corresponding figure for the 

 equinoctial months was only thirty-one. 



The phenomena of magnetic storms appear, at least at 

 some stations, to be largely influenced by the hour of the 

 day. Table VI. gives some figures for Greenwich derived 

 from the hours of beginning and ending in Mr. Maunder 's 

 lists for the years 1848 to 1903, as well as some figures 

 which Dr. Van Bemmelen has given for Batavia. 



T.ABLE VI. 

 Diurnal Variation in Magnetic Storms. 



At Greenwich no less than 60 per cent, of the storms 

 commenced during the eight hours 1 to 8 p.m., while only 

 9 per cent, then ended. 



There is yet another influence on magnetic changes 

 which requires to be considered, viz. sun-spots. 



Table VII. 



Connection betu/een Sun-spot Frequency and Declination 



Ranges. 



While Prof. Wolfer's figures are given in Table VII. 

 as a measure of sun-spot activity, it may be added that 

 closely parallel results would be derived from the 

 Astronomer Royal's figures for sun-spot areas. There was 

 a well-marked maximum in 1893. The remarkable 

 parallelism between the changes in sun-spot frequency and 

 in the diurnal inequality ranges appeals to the eye. 



Passing to the absolute daily range, we have a quantity 

 which is considerably influenced by magnetic storms. 

 Here, again, the ranges in the years of many sun-spots 

 are conspicuously the larger, but the parallelism with sun- 

 spot frequencies is less close. 1893, the year of sun- 



NO. 2 II 6, VOL. 83] 



spot maximum, shows at both Kew and Pavlovsk a dis- 

 tinctly smaller absolute range than either of the adjacent 

 years, especially 1892. Of the last two lines in Table VII., 

 the first gives the arithmetic mean of the differences 

 observed at Pavlovsk between the extreme positions of the 

 compass needle during each month of the year, while the 

 second gives its total range during the year. In both cases 



1892 occupies the premier, and 1894 the second, position. 



1893 lags far behind ; in the case of the annual range it 

 even follows 1900, which had the smallest sun-spot fre- 

 quency of the whole eleven years. The close parallelism 

 visible between sun-spot frequency and the regular diurnal 

 inequality becomes more and more obliterated as we pass 

 from the regular to the less regular, and from these to 

 the highly irregular daily changes of terrestrial magnetism. 



A general parallelism between sun-spot frequency and 

 the range of the regular diurnal inequality is far from 

 proving any intimate connection between the two pheno- 

 mena on the same day. Table VIII. gives the results of 

 an attempt to find out whether the parallelism extends to 

 individual days' results. 



T.ABLE VIII. 



Relation of Sun-spot Area (Greenwich) to .Absolute 

 Declination Range {Kew) on same Day and on Three 

 Subsequent Days. 



The days of each month were divided into three groups. 

 The first group included the ten days in which the Green- 

 wich sun-spot areas were the largest, the third group the 

 ten days in which they were least. If any close parallelism 

 existed between the solar and magnetic phenomena on 

 the same day, we should expect the mean of the absolute 

 declination ranges from the first group of days to be much 

 larger than the mean for the whole month, and that from 

 the third group to be much less. Taking all the months 

 of the years 1890-1900, there is a difference in the direc- 

 tion indicated, but it is exceedingly small. 



To provide for the possibility that the solar influence 

 takes one or more days to travel to the earth, mean 

 declination ranges were formed, not merely for the ten 

 days of largest or smallest sun-spot area, but also for the 

 ten days immediately following these, for the ten days 

 separated by two days, and yet again for the ten days 

 separated by three days, from the days constituting the 

 sun-spot groups. The results appear in Table VIII., and 

 are somewhat more favourable for an association between 

 the magnetic phenomena and the solar phenomena two or 

 three days previously than for an association between the 

 phenomena on the same day. Individual years, however, 

 e.g. 1894 and 1895, give conflicting results. 



In the preceding discussion declination has been chiefly 

 referred to, because it is the most familiar element. In 

 some respects, however, declination records during mag- 

 netic storms are inferior in interest to those of horizontal 

 force. Fig. i shows two successive days' records — 

 November 12-14, '^4 — 0^ this element at Kew. The 

 first day's trace, which was quiet, helps to bring out 

 two important features. .\ little after 2 p.m. on 

 November 13 there is a very small decrease of force (down- 

 ward movement), followed by a much larger increase. 

 These sudden commencements to storms are not unusual, 

 and seem to occur simultaneously all over the earth. The 

 type at most stations is very similar. The initial slight 

 fall in force is only sometimes seen ; the rise is generally 

 substantial. In the .Antarctic the oscillatory character is 

 unusually prominent. 



By 8 or lo a.m. of November 14 the disturbance is 



