OBSERVATIONS OF THE MAGNETIC FORCE. 
305 
although the evidence on this point is obscured by the occasional irregularities of 
observation. If we treat the observed values between Dunvegan and Toronto in the 
manner already described in discussing the observations with No. 30, we obtain •3495 
as the value of m at Toronto on the 5th of February 1845, and ‘3495-1- ‘0000 12a as 
its value on any anterior day in that interval. 
The values of m with No. 17, derived from the deflections, are as follows : — 
Fort Simpson, May 2, 1844 . . "6858. 
Fort Simpson, June 12, 1844 . "6738. 
Fort Resolution, June 24, 1844. *670 7- 
Athabasca, July 2, 1844 . . . "6691. 
Fort Vermilion, July 1 1 , 1844 . "6678. 
Fort Dunvegan, July 24, 1844 . ’67 27. 
Fort Edmonton, Aug. 18, 1844. *6702. 
Norway House, Sept. 4, 1844 . *6441. 
Fort Alexander, Sept. 20, 1844 . ’6414. 
Rat Portage, Sept. 25, 1 844 . . *6423. 
Fort Francis, Sept. 29, 1844. . "6401. 
Fort William, October 10, 1844. "6405. 
Sault St. Mary, Nov. 5, 1845 . "6393. 
Toronto, March 8, 1845 . . . "6368. 
A cursory examination of these values suffices to show, that between September 4, 
1844, at Norway House, and March 8, 1845, at Toronto, a small and progressive loss 
of magnetism was sustained ; but that between the observations at Fort Edmonton 
on the 18th of August, and those at Norway House on the 4th of September, an irre- 
gular and very considerable loss occurred. A more careful examination of the times 
of vibration of this magnet, in comparison with those of No. 30 and 31, show that the 
period at which this loss took place was, after the observations at Devil’s Drum Island 
on September 1, and before those at Norway House on the 4th of the same month. 
The stations therefore at which the observations of deflection were made with this 
magnet may be divided into two series ; one antecedent to the loss thus sustained, 
and the other subsequent to it. For the latter series, viz. between Norway House 
on the 4th of September and Toronto in February 1845, the most probable values of 
m have been obtained by the method of least squares in the manner already described ; 
which gives for m at Toronto, on the 8th of March 1845, ="6363, and on anterior 
days "6363 + •000032a. But we have still to provide for the observations of vibration 
made in July and August, and for this period the arithmetical mean of the observed 
values of m between June 24 at Fort Resolution, and August 18 at Fort Edmonton, 
viz. *6700, is perhaps the least exceptionable value that can be taken ; the irregula- 
rities of the observed values during the period would scarcely justify a more precise 
deduction. 
When the values of m, resulting from the observations made with a magnet at 
several stations of a survey, give reason to infer that its magnetism has remained 
constant, it is obviously preferable to employ at each station a mean value of m de- 
rived from the whole body of the observations, as irregularities of individual deter- 
minations afford a mutual compensation. When the loss of magnetism during the 
whole or a part of the survey has been small and progressive, a probable value of ?n, 
computed for the period of the loss in the manner described, may continue to be 
