80 Mr William Swan on a 



dered parallel by the lens, and thus enter the telescope of the 

 theodolite in a fit state to be brought to focus at the dia- 

 phragm wires, where they form a distinct image of the cross 

 fibres. A light tube represented in the figure by dotted lines, 

 slides over the whole, so as to protect the magnet from currents 

 of air ; and is furnished with an aperture at its end, covered with 

 glass, through which light is thrown by a small reflector to illu- 

 minate the cross fibres. 



The method of observation consists in first making the 

 image of the intersection of the collimator cross fibres coincide 

 with the middle diaphragm wire of the theodolite telescope, 

 which is easily effected by means of the tangent screws of that 

 instrument, and then reading off the verniers on its horizontal 

 limb. If the magnetic axis of the magnet were parallel to the 

 optical axis of the collimator, the reading on the limb for mag- 

 netic north or south would thus be at once obtained ; but as 

 such a condition can never be strictly fulfilled in practice, it 

 is necessary, where an accurate result is wanted, to repeat 

 the observation with the needle in an inverted position. For 

 that purpose the agate cap is made to screw into opposite sides 

 of the magnet, which thus admits of being suspended with 

 either side uppermost. By taking the mean of the readings 

 in the two positions of the magnet, any error caused by want 

 of parallelism in the line of collimation and the magnetic axis 

 will be either wholly or nearly eliminated. Half the differ- 

 ence of the readings in the two positions of the needle, care- 

 fully determined from a number of observations, may be regis- 

 tered and applied as an index error, when the needle has been 

 observed without having been inverted ; and such a mode of 

 observation will probably be sufficiently accurate for the ordi- 

 nary purposes of the surveyor. 



It is desirable that the instrument be adjusted so that the 

 difference of the readings in the two positions of the magnet 

 may not be great. For as the correction, for want of perfect 

 adjustment, obtained by taking the mean of those readings, 

 will generally be only approximate, it is well that any residual 

 error should be confined within as narrow limits as possible. 



In order to ascertain the variation of the compass, or to 

 apply the observations of the magnet to the ordinary purposes 



