394 



n is the number of readings in each series ; a the dip, when the marked side 

 of the needle turned to east, four readings with limb east, and four when 

 west ; b the dip, when the marked side was west, and limb both east and 

 west ; c and d the same, after changing the poles. When the weight was 

 applied, the dip is calculated by the formula — 



tans- i — CO tang a + eotang d — cotang b — cotang c . 

 cotang a . cotang d — cotang b . cotang c 



and by this method the different magnetical moment of the needle, before 

 and after the returning of the poles, is quite eliminated. The eight first ob- 

 servations, without weight, give 71° 53'. 1, the last six, with weight, 71° 54'.2, 

 and the probable error of the mean of the whole scries is 0'.851. Though 

 the probable error of the three determinations 1839, 823 (14 obs..), 1841, 7C9 

 (7 obs.), 1842, 170 (12 obs.), are only 0'.851, 0'.888, and 0'.806, their devia- 

 tions from the formula are respectively + 2'.52, + 3'.17, and — 1'.03, which, 

 I think, may principally be ascribed to Lregular variations in the dip, 



I beg your pardon for my profusion in this matter, but my long experience 

 with different instruments of this kind has made me somewhat sceptical in 

 the way of observing the dip, and I am not content before I have varied the 

 methods of observation, and multiplied the observations as much as pos- 

 sible. 



In observing the intensity by vibrations of a magnetical cylinder, you are 

 almost the only observer who has not neglected one or more of the neces- 

 sary reductions, as that for temperature, rate of the chronometer, force of 

 torsion, arc, and to observe the cylinder in the same place before and after 



