54 



EIGHTH RBPORT 1838. 



to revolve, were successively tried in four positions, which were, as 

 nearly as could be guessed, a quarter of the circumference apart ; 

 had they been precisely so, the needle must have rested on the 

 same points of the axle, in the 1st and 3rd positions, and in the 

 2d and 4th, (as the poles are reversed in each observation), and 

 the results in those positions should have been the same ; but 

 as this can have been only approximately done, each position may 

 be considered as bringing a different set of bearings into play. 

 The observations were made as before, in Mr. Robinson's house, 

 and have therefore no reference to the true dip. 



Table II. 



Trials of the Axles of the undermentioned Dipping Needles. 

 London, June and July, 1838. 



In all these six needles a great improvement was manifested. 

 The greatest difference occurring in any tvvo positions of the 

 axle of any one of the six needles is 8', including of course ac- 

 cidental errors of all kinds. 



The imperfection of the axle is a source of error, from the 

 effects of which, if it exists, the results can scarcely be freed 

 by any mode of conducting the observation ; at least, without 

 going through the very tedious operation of observing round 

 the circumference of the axle on every occasion. When accu- 

 racy is desii'ed, therefore, only such needles should be employed, 

 as have been ascertained by preliminary trial to be nearly 

 free from this defect. Needles with revolving axles are easily 

 tried. Those of the ordinary description, in which the axle is 

 permanently fixed, may be examined by observing the angle of 

 inclination shown by the needle when the circle is turned in 

 different azimuths from that of the magnetic meridian, and by 

 computing the dip by means of appropriate formulae, from the 

 angles shown in the different azimuths. If the axle is perfect the 

 dips so computed should all accord. In the azimuths intermediate 

 between the magnetic mei-idian and its normal plane, the needle 

 rests successively on all points of the axle comprised in a por- 

 tion of the quadrant equivalent to the complement of the dip ; 



