196 OBSERVATIONS OF THE TERRESTRIAL 



Needle J, (10 obs.) . . .dip. = 6916'-1 



Needle E, (10 obs.) ... 69 18 -9 



Meyer's needle, (8 obs.) ... 59 19 -6 



Needle B (2), Admiralty, ... (10 obs.) ... 69 21 -8 



Needle P, (8 obs.) ... 69 42 '6 



Mean dip = 69 17 '3 



Thus it appears that there is a difference amounting to 41' in 

 the results of two of the needles used ; and that this difference is 

 very far beyond the limits of the errors of observation will appear 

 from the fact that the extreme difference in the partial results with 

 one of these needles B (1) does not amount to four minutes and a 

 half, while with the other (P) the extreme difference is only two 

 minutes. In fact, it so happens that these very needles which 

 differ most widely in their mean results, are those in which the 

 accordance of the partial results is most complete. Of the eight 

 results obtained with needle (P), there is one only which differs 

 from the mean of the eight by a single minute ; and yet the mean 

 of all the observations with this needle differs by more than 20' 

 from the mean of any of the others, while its excess above the 

 mean of the entire series amounts to 25'. 



These differences cannot be ascribed to any partial magnetism 

 in the apparatus ; for three of the needles (J, P, and E) were of the 

 same dimensions, and were used with the same circle, and yet their 

 results, as we see, are widely discordant. "We must seek, then, in 

 the needles themselves the cause of these perplexing discrepancies, 

 and we are forced to conclude that there may exist, even in the 

 best needles, some source of constant error which remains uncor- 

 rected by the various reversals usually made; and that accordingly 

 no repetition of observations with a needle so circumstanced can 

 furnish even an approximation to the absolute dip. If this error 

 be due to the incomplete adjustment of the needle (such as devi- 

 ation of centre of gravity from the axle, &c.), its magnitude will be 

 a function of the dip, and of the force, which may be assumed to 

 be constant where the variations of these elements are not consider- 

 able. Hence, to determine its amount for any particular needle, 

 it is necessary to make a careful series of observations with it at 

 some station for which the dip has been accurately determined 

 (from the mean of several needles) ; and the difference will be 



