The semicircular error is due to a great extent to permanent magnetism, but also in part to induced magnetism. It is so called because when the ship's head is turned through 360°, the error attains a maximum on two courses 180° apart. It may amount to over 20° in an ordinary iron vessel, and to 30° or 40° in an armour-clad. It is corrected by two sets of steel magnets placed with their centres under the needle in the binnacle. One set have their lengths fore and aft, the others in the thwart-ship direction. These magnets annul the error on the north and south and on the east and west courses, due to the two horizontal components of magnetic force produced mainly by the permanent magnetism of the ship. A regular routine of swinging the ship when marks on the shore (the true bearings of which from the ship are known) are available, is followed for the adjustment. The quadrantal error is so called because its maxima are found on four compass courses successively a quadrant, or 90°, from one another. It amounts in general to from 5° to 10° at most. It is due to induced magnetism, and is corrected by a pair of soft-iron spheres, placed on the two sides of the compass with their centres in a line transverse to the ship, through the centre of the compass needle. There are, however, exceptional cases in which they are placed in the fore and aft line one afore, the other abaft, the needle. When the quadrantal error has once been annulled it is always zero, for as the induced magnetism changes, so does that of the spheres, and the adjustment remains good. In a new ship the permanent magnetism slowly alters, and so the semicircular correction has to be improved from time to time by changing the magnets. These adjustments are not quite all that have to be made; but enough has been stated to show how the process of compensation can be carried out with the Thomson compass. The immensely-too-large magnets used formerly as compass needles, through a mistaken notion, apparently, that more directive force would be got by their means, rendered the quadrantal adjustment an impossibility. The card swinging round brought the large needles into different positions relatively to the iron balls, when these were used, and exerted an inductive action on them which reacted on the needles, producing more error, perhaps, than was corrected. Thomson invented also an instrument called a "deflector," by which it is possible to adjust a compass when sights of sun or stars, or bearings of terrestrial objects, cannot be obtained. By means of it the directive forces on the needles on different courses can be compared. Then the adjustment is made by placing the correctors so that the directive force is as nearly as may be the same on all courses. The compass is then quite correct.