372 



hemispheres, but opposite in the two half years. In the one case 

 the effects are hemispherical, in the other semiannual. It is this pe- 

 culiarity which gives to the " April to September " branch of the 

 semiannual inequality its analogy with the diurnal variation which 

 prevails throughout the year in the middle latitudes of the northern 

 hemisphere, and to the " October to March " branch its analogy 

 with the diurnal variation which prevails throughout the year in the 

 middle latitudes of the southern hemisphere. The analogies extend 

 even to the small but apparently systematic difference which exists 

 between the turning hours of the mean variation in the two hemi- 

 spheres, and of the semiannual variation in the two half years. The 

 turning hours of the variation in the northern hemisphere, and of the 

 " April to September " semiannual branch, appear to occur system- 

 atically about an hour earlier than those of the southern hemisphere, 

 and of the "October to March" semiannual branch. This is a 

 connecting link which draws still nearer the analogies of which the 

 broader features have been frequently noticed and commented upon ; 

 and is the more remarkable on account of the diversity which in other 

 respects seems to distinguish the mode of operation by which the 

 solar influence produces in the one case hemispherical difference with 

 annual agreement, and in the other case semiannual difference with 

 hemispherical agreement. 



Thus at Pekin, regarded as a station in the middle latitudes of the 

 northern hemisphere, if we view the semiannual mean of the six 

 mouths from April to September, we see repeated the general fea- 

 tures of the annual mean, reinforced by the semiannual inequality of 

 kindred character with itself; the deflections of both having the 

 same direction at the same hours, the range becomes enlarged, but 

 its characteristics are unchanged ; the progression is still a single one, 

 as is the case in the annual mean, with but one easterly and one 

 westerly extreme, the hours of which are slightly earlier than those 

 of the annual mean, by reason of a particular feature of the semi- 

 annual inequality spoken of above. "When, on the other hand, we 

 direct our attention to the semiannual mean from October to March, 

 we see the consequence of the superposition upon the annual mean 

 of the opposite semiannual inequality belonging to these months : 

 this is most particularly shown in the effect produced upon the semi- 

 annual mean by the great semiannual loop which culminates about 



