t ake a terrella of the best strong loadstone, and homogeneous throughout, not weakened by decay or by a flaw in any parts; let it be of a fair size, so that its diameter is six or seven digits; and let it be made exactly spherical. having found its poles according to the method already shown, mark them with an iron tool; then mark also the æquinoctial circle. afterwards in a thick squared block of wood, one foot in size, make a hemispherical hollow, which shall hold half of the terrella, and such that exactly one half of the stone shall project above the face of the block. divide the limb close to this cavity (a circle having been drawn round it for a meridian) into 4 quadrants, and each of these into 90 degrees. let the terminus of the quadrants on the limb be near the centre of a quadrant described on the block, also divided into 90 degrees. at that centre let a short, slender versorium (its other end being rather sharp and elongated like a pointer) be placed in æquilibrio on a suitable pin. it is manifest that when the poles of the stone are at the starting points of the quadrants, then the versorium lies straight, as if in æquilibrio, over the terrella. but if you move the terrella, so that the pole on the left hand rises, then the versorium rises on the meridian in proportion to the latitude, and turns itself as a magnetick body; and on the quadrant described on the flat surface of the wood, the degree of its turning or of the declination is shown by the versorium. the rim of the cavity represents a meridional circle, to which corresponds some meridian circle of the terrella, since the poles on both sides are within the circumference of the rim itself. these things clearly always happen on the same plan on the earth itself when there is no variation; but when there is variation, either in the direction or in the declination (a disturbance, as it were, in the true turning, on account of causes to be explained later), then there is some difference. let the quadrant be near the limb, or have its centre on the limb itself, and let the versorium be very short, so as not to touch the terrella, because with a versorium that is longer or more remote, there is some error; for it has a motion truly proportionate to the terrella only on the surface of the terrella. but if the quadrant, being far distant from the terrella, were moved within the orbe of virtue of the terrella toward the pole on some circle concentrick with the terrella, then the versorium would indicate the degrees of declination on the quadrant, in proportion to and symmetrically with that circle, not with the terrella.
