Sometimes I find when adjusting the quadrantal headings, which I performed the last after having adjusting the cardinal magnetic headings , that moving the spheres in or out I do not succeed to correct all the deviations which appear in these quadrantal headings NE,SE,SW,NW ,and this is due that the Spheres which are of a standard diameter are or too big or on the contrary too small.
In both cases to solve the problem, if they are too big the appropriate distance should be to have the brackets with more longitude or spheres with less diameter, which is not available, so you have to leave the spheres in its maximum distance from the compass and note the residual deviations.
In the other case if they are too small, the spheres must be changed by other bigger, which is, also, not available, so the only way is to install the spheres close to the compass and again note the deviations which will be written in the certificate.
Sometimes if you take out the spheres from the binnacle there are less deviations ,but this is not convenient because the spheres give accuracy and stability to the compass.
With Japanese boxes which have plates inside, the same thing could happen, either introducing into the boxes plates the deviations increase, in this case you have to take out the plates or introducing all the possible plates you do not succeed to correct all the deviations which appear, in the first case take them out and in the second case leave all the possible plates and write down the residual deviations.
CAPT. A.ROS COMPASS ADJUSTER Nº 86 BY DGMM (SPANISH ADMINISTRATION)