77 



deserve particular attention in theoretical respects. At Pekin, for 

 example which may be advantageously compared with Kew, being 

 both in the same hemisphere, but Pekin some degrees nearer the 

 equator the variation is zero in the passage of the north end of the 

 magnet from east to west at 20| lunar hours, or 2| hours earlier than 

 the corresponding epoch at Kew. Again, at the Cape of Good 

 Hope, situated in the same hemisphere with Hobarton, but some 

 degrees nearer the equator, the variation is zero in the passage of the 

 north end of the magnet from west to east also at 20| lunar hours, 

 or 2| hours earlier than the corresponding epoch at Hobarton. Thus 

 there is an accord of precisely the same kind between Pekin and the 

 Cape of Good Hope that there is between Kew and Hobarton, whilst 

 there is a difference between the two pairs of stations of 2| hours in 

 the position of the moon relatively to the meridian at which she 

 ceases to exercise a deflecting influence on the magnet. Again, at 

 St. Helena, which is in the same (geographical) hemisphere as Ho- 

 barton and the Cape of Good Hope, but still nearer to the equator 

 than either, the lunar influence is zero in the passage from west to 

 east at 19| lunar hours, being one hour earlier than at the Cape, 

 and 3 1 hours earlier than at Hobarton. 



Where the whole range of the variation of which we have been 

 treating is so small (not more than a few seconds of arc in each lunar 

 day), it may be desirable to show by the accordance'of the independent 

 evidence obtained in single years, the degree of confidence which may 

 be placed in the mean results of several years. This may be seen in 

 the Table on the next page, which contains the separate results in 

 each of the five successive years of observation at Hobarton, as well 

 as their mean. 



In this Table the principal features of the variation are seen to be 

 substantially alike in each year. The individual results at the several 

 hours in single years are of course somewhat less regular than in the 

 mean of the five years : such small discrepancies are no doubt in great 

 part due to the lesser disturbances which, being below the separating 

 value of 2'* 13, have been left in the body of the observations. They 

 slightly disfigure the symmetry of the results in single years, but 

 almost entirely disappear when the mean of several years is taken. In 

 order to appreciate justly and fully the confidence to which the whole 

 investigation is entitled, it must be borne in mind that every single 



