202 ON DETERMINATIONS OF THE ABSOLUTE VALUE, SECULAR CHANGE, 
striictions for their use, were found on the experience of the first few years to be not 
fully adequate to the satisfactory attainment of all the purposes for which they had 
been designed. This was especially the case in regard to the Bifilar Magnetometer, 
the magnet of which, 15 inches long, was directed to be used, in conjunction with the 
magnet of the Declinometer (which was of similar length), for the determination of the 
horizontal component of the magnetic force. It was soon found that at a suitable 
distance from each other, necessarily regulated by their length, the magnets were 
too far apart in the experiments of deflection, to produce angles of deflection bearing 
a sufficient proportion to the unavoidable errors of observation ; whence the results 
so obtained were charged with a probable error far too great for tlie purposes con- 
templated. It was also found, that, whenever the magnets of the Bifilar and Decli- 
nometer were thus employed, a break was unavoidably made in the series with the 
Bifilar, on the continuity of which its value as a differential instrument, and therefore 
as an instrument for determining secular changes, must necessarily in great measure 
depend. To obviate these inconveniences, and particularly the latter, “ Revised In- 
structions” were drawn up and circulated by the Royal Society, recommending the 
employment of an auxiliary apparatus, by which the experiments of vibration and de- 
flection required for the determination of the horizontal component of the Force 
might be made in a separate apartment, and therefore without disturbing the magneto- 
meters of the observatory, which from thenceforward would be employed solely as 
differential instruments. The magnets of the auxiliary apparatus were reduced from 
15 inches to 12 and 9 inches in length. It was found however on trial, that when these 
magnets were placed at the minimum admissible distance from each other correspond- 
ing to their length, their distance apart was still too great to produce angles of deflec- 
tion of sufficient magnitude, to lessen the probable error of the results to an amount, 
which would enable the secular change from month to month, or even from year to year, 
to be ascertained by their means within any reasonable period of observation. A few 
months’ additional experience also served to show that when the Bifilar was thus set 
free to be appropriated wholly as a differential instrument, and when the magnet was 
left perfectly undisturbed, the conclusions which might be derived from it were still 
subject to two instrumental irregularities, which did not conveniently admit of eli- 
mination by any known process of experiment, and which prevented the observations 
made with the Bifilar from being strictly comparable with each other for more than 
very short intervals of time. One of these sources of irregularity had been antici- 
pated, and it was hoped might have been surmounted ; but the other had been 
apparently wholly unforeseen. The first consisted in the liability of the magnet bar 
to lose a portion of its magnetism. It was hoped if the magnet were entirely undis- 
turbed by removals, that after a few months it might gradually attain a state in 
which its magnetism might undergo no further change (at least while it remained 
undisturbed), and that even whilst the period of change continued, the loss might be 
of a sufficiently uniform character to admit of its being allowed for. This has however 
