Declination being investigated on the earth itself by means of a declination instrument, we may use either a short or a very long versorium, if only the magnetick virtue of the stone that touches it is able to permeate through the whole of its middle and through all its length. For the greatest length of a versorium has no moment or perceptible proportion to the earth's semi-diameter. On a terrella, however, or in a plane near a meridian of a terrella, a short versorium is desirable, of the length, say, of a barleycorn; for longer ones (because they reach further) dip and turn toward the body of the terrella suddenly and irregularly in the first degrees of declination. [Illustration] For example, as soon as the long versorium is moved forward from the aequator A to C, it catches on the stone with its cusp (as if with a long extended wing), when the cusp reaches to the parts about B, which produce a greater rotation than at C. And the extremities of longer wires also and rods turn irregularly, just as iron wires and balls of iron and other orbicular loadstones are likewise turned about irregularly by a long non-orbicular loadstone. Just so magneticks or iron bodies on the surface of a terrella ought not to have too long an axis, but a very short one; so that they may make a declination on the terrella truly and naturally proportionate to that on the earth. A long versorium also close to a terrella with difficulty stands steady in a horizontal direction on a right sphere, and, beginning to waver, it dips immediately to one side, especially the end that was touched, or (if both were touched) the one which felt the stone last. * * * * * {195} CHAP. V. That declination does not arise from the attraction of the loadstone, but from a disposing and _rotating influence_.