59 MAGNETIC FIELD. 



knowledge of the time or of longitude is sufficient to calculate the 

 declination by horizontal tables. Observation is especially accurate 

 when the height of the star varies very rapidly that is to say, 

 when it is to the east or the west in the vicinity of the prime 

 vertical. 



1162, In magnetic observations we ought to be certain first that 

 the wire has no initial torsion. A copper bar of the same weight 

 as the magnet is suspended to it, and the circle of the pulley is 

 turned until the bar is stationary in the plane of sight of the micro- 

 scope. The magnet is substituted for the bar, one of its ends is 

 sighted with the microscope, and, after having made the oscillations 

 very small, the direction of the microscope is adjusted by the screw 

 so that the oscillations of the median line are symmetrical in respect 

 of the system of vertical lines ; the verniers are then read, and the 

 same observation is repeated at the other end. The mean of the 

 two observations corresponds to the line of sight of the magnet. 

 The magnet is then turned in its stirrup, and two similar observa- 

 tions are repeated. The mean of four readings corresponds to the 

 magnetic axis, and from this is deduced the declination. 



As the observer is placed very near the magnet, the smallest 

 iron objects which he may carry, and even traces of iron which 

 exist in certain materials used for clothing, may produce consider- 

 able perturbations. The precautions taken in this direction are 

 rarely sufficient. 



The instrument itself should be of copper or of bronze, free 

 from iron ; for there is no mode of correction by which this 

 source of error can be eliminated. The magnet may be turned 

 end over, and the observations repeated; but the concordance of 

 the results is not an absolute guarantee, and the mean does not 

 correct the error if it exists. 



The series of four observations being of long duration, the mag- 

 netic declination may have varied in the interval of the readings. 

 The indications of a variation apparatus, or, still better, of a regis- 

 tering apparatus, would enable us to correct each of the readings 

 so as to refer them to the same period. In all cases it is better 

 to make the observations when the declination passes through a 

 maximum or a minimum. This precaution is especially necessary 

 in summer, when the amplitude of the daily variation is greatest. 



Instead of making the central line of the divisions coincide at 

 each sighting with the central wire of the network, we can deter- 

 mine the angular value of the divisions, and that of the distance 

 of the wires, which would sometimes enable us to make the 



