to the Antarctic Regions, 227 



the magnet is in the direct and inverted positions. Half the sum of 

 these readings is the point of the scale corresponding to the mag- 

 netic axis of the magnet bar ; and half their difference (converted 

 into angular measure) is the deviation of the line of collimation of 

 the telescope from the magnetic meridian. The telescope should 

 be tnoved through this angle in the opposite direction. 



5. In order to take out the remaining torsion of the thread, the 

 magnet is again to be removed, and the unmagnetic bar (with its 

 small magnet attached) substituted. The deviation of this bar from 

 the magnetic meridian should then be read off on its divided scale, and 

 the moveable arm of the torsion circle turned through a given angle 

 in the opposite direction. The deviation being again read, a simple 

 l^roportion will give the remaining angle of torsion; and the moveable 

 iirm being turned through this angle in the opposite direction, another 

 observation will serve to verify the adjustment. The plane of de- 

 torsion then coincides with the magnetic meridian ; and the magnet 

 being replaced, the instrument is ready for use. 



Observations. — The observations to be made with this instrument 

 are, 1. of the absolute declination; 2. of the variations of the decli- 

 nation ; and 3. of the absolute intensity. 



For measurements of the absolute declination each observatory is 

 furnished with a small transit instrument having an azimuth circle. 

 This instrument being placed in the magnetic meridian of the decli- 

 nation instrument, the point of the scale coinciding with the central 

 wire of the transit telescope is to be observed ; the interval between 

 this point and the point * corresponding to the magnetic axis of the 

 bar, converted into angular measure, is the deviation (^) of the line 

 of collimation of the transit telescope from the magnetic meridian. 

 The verniers of the horizontal circle being then read, the telescope 

 is turned, and its central wire made to bisect a distant mark, whose 

 azimuth (a) has been accurately determined. If a denote the angle 

 read oft' on the horizontal circle, it is manifest that the angle between 

 the magnetic and the astronomical meridians is 



G -|- a -F ^, 



a and Z being aft'ected with their proper signs. The angle a is sup- 

 posed to have been previously determined by the help of the transit 

 instrument. 



But instead of referring the transit telescope directly to the mag- 

 netic meridian by means of the moving collimator, the same result 

 will be obtained, and probably in a better manner, by referring it 

 to the line of collimation of the fixed telescope, with which the changes 



* In determining this point by the mean of two readings of the scale with the 

 bar erect and inverted, care must be taken to ehminate the dechnation changes 

 which may occur in the interval of the two parts of the observation. The hori- 

 zontal force magnetometer may be applied to the purpose of this elimination. But 

 perhaps the simjjlest course is to take a series of readings as rapidly as possible, 

 alternately in the two positions of the bar, choosing for the time of observation a 

 period when the declination changes are slow and regular. By comparing each 

 result with the mean of the })recediug and subsequent, and then taking the mean 

 of all these partial means, a very acciu"ate dctermiuation mav be obtained. 



Q2 



