158 



Collecting these results into one point of view, we find in the 

 first place that Needles 1 and 2 from June 3 to Aug. S, 1837, 

 give mean dips not differing, on the average of 12 observations, 

 more than 0' 05, and again 10 observations from Aug. 26 to 

 July 3 give mean dips not differing more than 0' 1, quantities 

 much within the probable errors of observation. We may 

 therefore combine into one result the dips indicated by Needles 

 1 and 2. Hence — 



Obs. 

 1837, June 3 to 5 mean dip = 70°,, 48 -6 (20) 



14—5 = 70 46 4 (6) 



Aug. 1 to 3 = 70 51 1 (8) 



The mean epoch of this whole series * is June 20, and the mean dip 



of the whole series* (Obs. 34) ... 70 48 80 



For 1837, mean epoch June 20, Obs. 



mean dip 70 48 8 (34) 



In like manner the results for 1838, 

 give mean epoch Sept. 17, 



mean dip 70 50 34 (5) 



For 1839, mean epoch June 16, 



mean dip 70 43 91 (10)' 



For 1846, mean epoch Aug. 26, 



mean dip 70 19 03 (2) 



* "Weight being allowed in proportion to number of observations. 



