IN WOOD-BUILT AND IRON-BUILT SHIPS. 
65 
Starboard angle = — OP'06 ; which subtracted from approximate starboard angle gives 
P . 
True Starboard Angle =31P'33. And ^ X cosine of twice approximate starboard 
angle =0-020; adding this to unity and multiplying the Approximate Modulus, we 
have the True Modulus =0-352. 
11. Expressing the magnitude of the terrestrial magnetic force in the manner 
introduced by Gauss for Absolute Measure, and adopting the English foot and English 
grain as the units of length and weight, the measure of terrestrial horizontal force at 
Greenhithe is 3-79, and that of terrestrial vertical force is 9-66. Forming the quan- 
tities “True Modulus X cosine True Starboard Angle x Terrestrial Horizontal 
Force,” and “ True Modulus X sine True Starboard Angle X Terrestrial Horizontal 
Force,” we have for the ‘Trident’ at Greenhithe in 1852, 
-f 1 -375 = H -fN X 9-663 
— 0-182 = S; 
and these results, with that just obtained for the Coefficient of Quadrantal Deviation, 
are the most advanced that can be obtained from the deviations of the compass in 
the ‘Trident’ observed at Greenhithe only. 
I shall now exhibit, in a tabular form, the results of the twenty-nine series of devi- 
ations which have reached me. Nos. 1 to 13 are extracted from the work of the late 
Captain Johnson, R.N., “ On the Deviations of the Compass.” The signs of the 
compass-deviations of the ‘ Erebus’ at St. Helena are changed, on the authority of 
Colonel Sabine, as conveyed to me by Archibald Smith, Esq. Nos. 14 to 29 have 
been communicated to me by Captain Washington, R.N., Hydrographer to the 
Admiralty. 
It must be remarked that the first column, in every case, is the registered deviation 
as given by the observer (a few numbers in brackets being supplied by interpolation), 
and not the deviation as cleared of Jiiean error or index error. In some cases this 
mean error is large (thus with the ‘ Simoom’ at Simon’s Town it amounts to 1° 47')3 
and here it greatly modifies the true deviation, and even causes the original deviation 
to appear less on some points than the residual error. The residual error is formed 
by computing the polar-raagnet-deviation from the approximate elements at the top 
of the Table, and subtracting it from the deviation corrected for constant mean error 
only; it therefore contains the quadrantal deviation, the small terms produced by 
combination of polar-rnagnet-deviation with quadrantal deviation, and the accidental 
errors of observation. The coefficients of quadrantal deviation below are formed by 
omitting the residual errors for QP, 8P, 16 p, 24P, and taking one-fifth part of the sums 
of the residual errors in the groups between them ; and the mean coefficient is formed 
by changing the signs of the second and fourth coefficients of quadrantal deviation, 
and taking one-fourth of the algebraical sum. The true elements are formed by 
correcting the approximate elements in the manner just explained. 
MDCCCLVI. 
K 
