80 



approximately at the equinoxes. The approximation, particularly 

 at the September equinox, is very distinctly and definitely marked. 

 The day of the equinox is the 21st of September; if a mean be 

 taken of the diurnal march in the three weeks from the 1st to the 

 21st of September, the line which represents it scarcely differs 

 sensibly at any hour of the twenty-four from the mean line of the pre- 

 ceding half-year, taken from the 22nd of March to the 20th of Sep- 

 tember ; thus showing that the phenomena of that semiannual group 

 are unchanged up to the time of the equinox. If in the same way a 

 mean be taken of the diurnal march in the three weeks following the 

 21st of September, the line which represents them shows that the 

 passage from the phenomena of one semiannual group to those of the 

 other has not only commenced, but that in half the period of three 

 weeks, i. e. within eleven days of the equinox, the change has already 

 advanced very far towards its completion ; and by the middle of Oc- 

 tober it is found to be quite complete, the mean in October retaining 

 no trace of those semiannual characters which had undergone no 

 modification ten days before the equinox." The facts thus stated 

 were illustrated by diagrams. 



"At the March equinox the commencement of the change is 

 equally definite : no trace of change can be discovered in the mean 

 from the 1st to the 20th of March, when compared with the mean 

 of the six months from the 22nd of September to the 20th of March ; 

 the change then commences, but from some cause not yet apparent, 

 the conversion from the phenomena of the one half-year to those of 

 the other is effected less rapidly at this than at the September 

 equinox. The mean of the month of April retains the distinct traces 

 of the group which it has quitted, and is in fact a month of transition 

 between the two groups, but in May the conversion is quite com- 

 plete ; the phenomena of that month have no characteristic distin- 

 guishable from those of June, July and August. 



" From what has been stated in the preceding paragraphs, it will 

 be evident that the epochs of the sun's passage of the equator have a 

 very marked influence on the phenomena under consideration, and 

 that the influence is the same and produces similar effects whether 

 the station itself be north or south of the equator, and however di- 

 verse may be its climatic or magnetic conditions. The semiannual 

 characteristics continue unchanged up to the days of the respective 



