100 



Dip and Variation of the Magnetic Needle. 



As the computed dip falls between the two observations, it can- 

 not be pronounced inadmissible. 



At Albany, observations made within about a year of each other 

 differ 11'. Here then is evidence of local attraction. The least 

 observed dip reduced to 1840 differs 4^5 from the computed dip. 

 At Princeton, the dip is derived from a single observation. At 

 Philadelphia, observations made within a year of each other differ 

 23.2. Here is undoubted local attraction. My own observation 

 differs 7'.5 from the computed dip. 



In order to test the hypotheses upon which the dip is compu- 

 ted, let us classify the differences. The mean difference at the 

 ten most northerly stations is +0'.3 ; at the ten most southerly 

 — 0'.3; at the five most eastern O'.O; at the five most western 

 4-2^2; at the ten intermediate stations —I'.l. It may be infer- 

 red then that within the limits of the preceding table, the hypoth- 

 esis of straight, parallel, and equidistant isoclinal lines, is well 

 nigh rigorously exact. There is then reason to believe that these 

 differences between the observed and computed dip, arise chiefly 

 from local attraction and errors of observation. As the dip ob- 

 served at Charlottesville, Ya. in 1834, 71^ 8'.7, is about half a de- 

 gree greater than observations made in other parts of the United 

 States would lead us to expect, I have thought best to omit it in 

 this comparison. 



Table II. — Observations of the Magnetic Dip from 78° to 83° west longitude. 



