162 Royal Society. 



Observed Deviations over these Tabular Deviations ought to consist 

 simply of Quadrantal Deviation and Errors of Observations, 



From the neutral point and the modulus, with the absolute mea- 

 sure of the terrestrial horizontal force, it is easy to form the abso- 

 lute measures of the apparent permanent forces of the ship, in the 

 directions of " headward " and " starboard " respectively. The star- 

 board force (on the assumption of general symmetry) can arise from 

 nothing but subpermanent magnetism ; the headward force will con- 

 sist of subpermanent magnetism added algebraically to a multiple of 

 the terrestrial vertical force, the multiplier being an unknown con- 

 stant different for each different ship. 



The process is then applied by the author to four wood-built sail- 

 ing ships, two wood-built steamers, and five iron-built steamers, 

 whose compass-deviations have been observed at twenty-nine sta- 

 tions in all. The results are as follows : — 



1 . In aU cases, the principal part of the deviation follows the law 

 of polar-magnet-deviation. 



2. When the polar-magnet-deviation is computed accurately from 

 the Table, and subtracted from the observed deviation, the residual 

 quantity in all cases follows very closely the law of quadrantal devia- 

 tion, leaving very little to be accounted for by errors of observation. 



3. For each ship, the coefficient of quadrantal deviation is sen- 

 sibly the same in all localities. The small deviations from exact 

 equality cannot be referred to geographical position, and evidently 

 depend on accidental changes in the distribution of the iron, espe- 

 cially of that which is near to the compass. 



4. As the test of theory must reside in the comparison of residual 

 quantities, and as it appears that these residual quantities obey with 

 great exactness the law which theory assigns, it follows that the 

 theory, "that the deviation may in all cases be represented by a 

 combination of two deviations, of which one is a polar-magnet- 

 deviation, and the other is a quadrantal deviation whose angular 

 coefficient, for the same ship, is constant under all circumstances," 

 is practically accurate. 



5. Consequently in every case the deviation at any locality may 

 be perfectly corrected ; by the application of a steel magnet to neu- 

 tralize the polar-magnet-force ; and of a mass of soft iron on one 

 side of the compass, and at the same level, to correct the quadrantal 

 deviation (as the author had previously explained) . 



6. The mass of soft iron will not require to be changed, under 

 any circumstances. It will depend on the variability or constancy 

 of the subpermanent force to determine whether the steel magnet 

 must be or must not be changed in different localities or after the 

 lapse of time. 



7. On forming the expressions for the absolute values of the head- 

 ward and starboard polar-magnet-forces, it appears that in some 

 ships the subpermanent magnetism is really (to sense) a permanent 

 magnetism, but that in others there is a sensible change. In one 

 instance the change was such that, supposing the deviations to be 

 accurately corrected by magnets and soft iron in England, there 

 would have been at the Cape of Good Hope an error whose maxi- 



