72 IIACXETIC SURVEY OF P E N N S Y L V A X I A . 



The probable error of any single observation is + 0°.12 = + 7'. 2; the pro- 

 bable error of arvj observation with the regular dip needles, and the Lloyd needles 

 combined, is ± ()°.13: with the latter needles alone, ± 0°.ll. This shows that the 

 irregularities in the observed dip are due to local attractions rather than to imper- 

 fections in the needles employed. It is proper, therefore, to assign equal weights 

 to results by the direct and indirect method of observing. 



If we apply Peirce's criterion for the rejection of observations differing too much 

 from the regular value indicated by all other observations, we find the limit of 

 rejection to be ± 0°.4(5, or ± 28' ; the maximum difference in the preceding table 

 is 25'; hence no observation is excluded. 



General Sabine's resulting isoclinal lines, in his seventh contribution to terrestrial 

 magnetism {Phil. Trans. Roy. Soc., Part III, 1846, p. 237), refer to an average 

 period between 1840 and 1842, and correspond in their position very closebj to 

 those now presented ; they are deduced from independent data. 



