MAGNETIC SURVEY OF GREAT BRITAIN. 105 



turbing cause is merely to widen the limits of probable error. 

 The imperfect curvature of the axle, and the magnetism of the 

 limb, act however very differently. Either of these sources of 

 error must, at a given place, affect all the results in the same 

 manner; and, consequently, no repetition of observation, with 

 an instrument so circumstanced, can afford even an approxi- 

 mation to the true dip. At different places the error will be 

 different, and will vary according to no assignable law. 



The course to be pursued by the observer with reference 

 to these errors is manifest. Their existence or non-existence 

 should be ascertained at the outset by one or other of the means 

 pointed out by Major Sabine in the commencement of this me- 

 moir ; and if found to surpass certain limits, the instrument 

 should be rejected. The case is different, however, when the 

 instrument has been actually employed for some time pre- 

 viously to the detection of the error. Here we must seek, if 

 possible, to determine the probable amount of the error, and 

 apply it, with an opposite sign, as a correction to the results. 

 Where the district of observation is limited, this is practicable. 

 It will be easily understood, that the imperfect curvature of the 

 axle, or the disturbing action of the limb, must, within a moderate 

 range of dip, affect all the results in the same manner, so that 

 they will all require a correction having the same sign ; and that 

 when the range of dip is very small, the amount of the dis- 

 turbance will be nearly the same throughout, and consequently 

 the correction required will be nearly constant. In such a case 

 then we have only to determine the amount of the error at some 

 one station, by a comparison of the results with those of proved 

 needles obtained at the same place, and, if possible, at the same 

 time. 



Again, in needles whose poles are unchanged, gravity acts 

 with a certain moment with or against the directive force ; the 

 coincidence of the centre of gravity with the axle being rarely 

 attained. The observed inclination, therefore, deviates from 

 the true dip, and the amount of this deviation varies in different 

 places, according to a known law*. To obtain its actual 

 value, however, at any station, it must be known at some one ; 

 and this knowledge is to be obtained, as before, by a com- 

 parison of the results with those of other needles at that sta- 



• Fifth Report, p. 144. With needles whose poles are inverted in each 

 observation, the true dip may be inferred from the observed angles of incli- 

 nation, however considerably they may deviate from it. In such needles, 

 therefore, the non- coincidence of the centre of gravity with the axle cannot 

 properly be ranked among the sources of error. 



