MAGNETIC OBSERVATIONS. 



199 



rg and r^ are neither equal to each other, nor yet to r^ and r & but are, the one 



T 



nearly three, and the other four, times as great as , . Assuming the theory 

 employed in this discussion to be correct, we should have expected to find r^, r a , 



?* 



r t , r ffi sensibly equal to each other, and all sensibly equal to ; . Such, however, 



y 16 



is not the case; and, as the results for each compass all tend in precisely the same 

 direction as ,the mean result, a doubt naturally arises whether or not the theory 

 really represents the semi-circular deviation as accurately as it does the quadrantal. 

 As this doubt is founded upon observations which may possibly have been affected 

 by some unknown cause of constant error as they were all made on a single vessel 

 during a single cruise perhaps it would not be well to insist upon it too strongly; 

 but at all events, it shows the necessity for further investigation of the subject, and 

 especially the great want of more observations. 



The probable errors of the coefficients 53, (, ^ (, for each compass, when com- 



puted from the values of A^ , , -- , , , , ), and (5, given in the table 



Ai fb f\i fd At /b 



on page 193, are as follows : 



The following table shows, for each compass, the place at which the maximum 

 value of its deviation, $, was the greatest, together with the point on which that 

 maximum value occurred, and its amount. Also, the place at which the maximum 

 value of its deviation was the least, together with the point on vvhicli that maximum 

 occurred, and its amount. These deviations are given on the compass points, and 

 in computing them the true A was used. 



