56 
MR. AIRY ON THE OBSERVED DEVIATIONS OF THE COMPASS 
fig'. 3, let SH represent the magnitude and direction of the polar-magnet-force of 
induced magnetism, directed from the ship’s stern to her head (this diagram having 
no relation to the direction of terrestrial magnetism), and let HP 
represent the magnitude and direction of the subpermanent mag- 
netism, which, inasmuch as its direction is invariable with respect 
to the ship, is inclined at a constant angle to SH : then the force 
resulting from the composition of these two will be repre-sented in 
magnitude and direction by SP, which is invariable in magnitude, 
and inclined at a constant angle to SH. And this force will appear, 
in the phenomena of compass-disturbance at any one locality, as a 
whole, and cannot immediately be separated into the two parts SH 
and HP. 
All that can be done is this. At a given locality we can find the direction of SP 
with regard to the ship’s keel, if, by methods to be explained below, we can find the 
“ neutral position of the ship in reference to the polar-magnet-force,” or (which is 
the same thing) the azimuth of the ship’s head, or of the line SH, when the polar- 
magnet-deviation vanishes, or when the line SP coincides with the magnetic meridian. 
And we can find the magnitude of SP by methods to be explained below. Therefore 
we can find SQ and QP. And if we assume FIQ to be constant, and SH proportional 
to the earth’s vertical magnetic force at the given locality, we shall be able, by com- 
parison of the results at different localities where the vertical force has different 
magnitudes, to discover the value of SIT at each place and the value of HQ. The 
value of QP however requires no combination of results found at different places, and 
is not liable to any uncertainty. 
But without at present insisting on this separation of the subpermanent magnetism 
from the polar-magnet-force of induced magnetism, we can lay down the following 
rule : — 
(3). The whole disturbance of the compass, whether the ship be wood-built or iron- 
built, will be represented by the sum of the effects of two forces, which separately 
would produce these two disturbances: one, a polar-magnet-deviation whose neutral 
point may be in any direction ; the other, a quadrantal deviation, which may be 
expected to be a positive quadrantal deviation, following the law of signs -\ 1 — 
as depending on the quadrants of azimuth of the ship’s head. And the whole 
disturbance will be very nearly (but not exactly) the algebraic sum of these two 
disturbances : the slight departure from that law will be the subject of examination 
below. 
The practical problem, then, of analysing a given series of compass-deviations, is 
reduced to the dividing of them into two parts, of which one follows the law of polar- 
tnagnet-deviation, and the other follows the law of quadrantal deviation, subject to 
the trifling correction to which I have just alluded. 
In the experiments with the ‘Rainbow’ and ‘Ironsides’ (which are treated in my 
