77 



of the meridian, the phenomena are such as have been thus cursorily 

 described. Were there no annual variation at that hour the different 

 months would all have the same mean declination, and the extended 

 figure, which in the diagram represents the annual cycle, would 

 be concentrated into one point. The annual variation differs con- 

 siderably at the different hours ; but it is a general feature amongst 

 them that the months on either side of the one solstice are either 

 congregated together towards one extremity of the annual range 

 at the hour, whilst the months on either side of the opposite sol- 

 stice are similarly congregated at the opposite extremity, or the 

 months of both solstices are contemporaneously in pretty rapid 

 transition from the one extremity to the other. It is this annual 

 variation which has been overlooked in the supposition entertained 

 by a very eminent authority, that in the vicinity of the equator the 

 magnetic direction would be found to be constant at all hours of 

 the day and night. If we group together the monthly means of 

 each period of six months separated by the equinoxes, we have 

 two semiannual mean lines, each differing comparatively very 

 slightly from any one of the months of which it is composed, but 

 the two differing very greatly from each other, and both differing 

 very considerably from the mean diurnal march in the year. If 

 the latter line, viz. the mean diurnal march in the year, be projected 

 as a straight line, as is done in the zero-line of fig. 1 in the annexed 

 woodcut, the semiannual groups take respectively the forms exhibited 

 in that figure, the continuous line being the semiannual march in the 

 half year when the sun is north of the equator, and the dotted line 

 the semiannual march when the sun is south of the equator. It is 

 in this form that the phenomena of the annual variation in different 

 parts of the globe may be most advantageously compared with each 

 other. Fig. 2 represents the analogous phenomena at Toronto in 

 43 north, and fig. 3 those at Hobarton in 43 south latitude. The 

 semiannual groups at Toronto and Hobarton have been obtained in 

 precisely the same manner as those at St. Helena ; the scale is the 

 same in the three figures, f. e. "5 of an inch to 1''0 of Declination, 

 the dotted and continuous lines refer respectively to the same periods 

 of the year, and the zero line is in each figure the mean diurnal 

 variation in the year at the station. 



" In viewing these three figures, it is scarcely possible to doubt that 



