306 
MAGNETIC SURVEY IN NORTH AMERICA. 
preferable at the stations where the individual values were observed. For other 
portions of a survey where the observed discrepancies are considerable, and where 
they present an alternation of increasing - and decreasing values, an arithmetical mean 
may furnish the most probable value : but when the observed values at successive 
stations have varied to an amount which considerably exceeds what may reasonably 
be ascribed to observation error, — and whilst they manifest a progressive loss, afford no 
very decided indication of its regularity in correspondence with intervals of time, — a 
satisfactory combination may not be possible, and it maybe safer to employ the indi- 
vidual values furnished at each station of observation. Thus the arithmetical mean 
of the four values obtained with No. 31, from June 12 at Fort Simpson to July 11 at 
Fort Vermilion, appears preferable to the values themselves, or to any other deduction 
that might be made for that period ; and in the case of No. 17, the values observed 
on May 2 and June 12 are probably preferable to any others which could be assigned 
for the respective epochs, but their differences are too great to permit a value to be 
derived from them for any intermediate or an earlier period. 
The following Table contains the times of vibration of the three magnets, Nos. 30, 
31 and 17, at the whole of the stations in 1844, with the values of m either observed 
or deduced in the manner which has been described, and the absolute horizontal 
7T 2 K 
force computed by X = ^?p. 
* Values uncertain. 
