that the variation is caused by the inæquality of the projecting parts of the earth.
d emonstration of this may manifestly be made variation on imperfect terrella. by means of a
terrella in the following way: let there be a round loadstone somewhat imperfect in some part, and impaired by decay (such an one we had with a certain part corroded to resemble the atlantick or great ocean): place upon it some fine iron wire of the length of two barleycorns, as in the following figure. a b, a terrella in certain parts somewhat imperfect and of unæqual virtue on the circumference. the versoria e, f, do not vary, but look directly to the pole a; for they are placed in the middle of the firm and sound part of the terrella and somewhat distant from the imperfect part: that part of the surface which is distinguished by dots and transverse lines is the weaker. the versorium o also does not vary (because it is placed in the middle of the imperfect part), but is directed toward the pole, just as near the western azores on the earth. the versoria h and l do vary, for they incline toward the sounder parts very near them. as this is manifest in a terrella whose surface is sensibly rather imperfect, so also is it in others whole and perfect, when often one part of the stone has stronger external parts, which nevertheless do not disclose themselves manifestly to the senses. in such a terrella the demonstration of the variation and the discovery of the stronger parts is on this wise. 
let a be the pole, b the place of the variation, c the stronger regions; then the horizontal versorium at b varies from the pole a toward c: so that both the variation is shown and the stronger places of the loadstone recognized. the stronger surface is also found by a fine iron wire of the length of two barleycorns: for since at the pole of the terrella it rears up perpendicularly, but in other places inclines toward the æquator, if in one and the same parallel circle it should be more erect in one place than in another; where the wire is raised more upright, there the part and surface of the terrella is stronger. also when the iron wire placed over the pole inclines more to one part than to another. 
let the experiment be made by means of a fine iron wire of three digits length placed over the pole a, so that its middle lies over the pole. then one end is turned away from b toward c, and is not willing to lie quietly toward b; but on a terrella which is perfect220 all round and even it rests on the pole directed toward any point of the æquator you please. otherwise, let there be two 
meridians meeting in the poles a b, let iron wires be reared just at the ends d and c of the equal arcs d a and c a; then the wire at d (the stronger region) will be more raised up than that at c, the weaker. and thus the sounder and stronger part of the loadstone is recognized, which otherwise would not be perceived by the touch. in a terrella which is perfect, and even, and similar in all its parts, there is, at equal distances from the pole, no variation221. variation is shown by means of a terrella, a considerable part of which, forming a surface a little higher than the rest, does, although it be not decayed and broken, allure the versorium from the true direction (the whole terrella co-operating).
a terrella uneven in surface.
it is shown by a small spike placed over a terrella or by a small versorium; for they are turned by the terrella toward the mass that stands out and toward the large eminences. in the same way on the earth the verticity is perturbed by great continents, which are mostly elevated above the depths of the seas and make the versorium deviate sometimes from the right tracks (that is, from the true meridians). on a terrella it is thus demonstrated: the end of the versorium a is not directed straight to the pole p, if there be a large protuberance b on the terrella; so also the cusp c deviates from the pole because of the eminence f. in the middle between the two eminences the versorium g collimates to the true pole because, being at equal distances from the two eminences b and f, it turns aside to neither, but observes the true meridian, especially when the protuberances are of equal vigour. but the versorium n on the other side varies from the pole m toward the eminences h, and is not held back, stopped, or restrained by the small eminence o on the terrella (as it were, some island of land in the ocean). l, however, being unimpeded, is directed to the pole m. the variation is demonstrated in another way on a terrella, just as on the earth. let a be the pole of the earth, b the equator, c the parallel circle of latitude of 30 degrees, d a great eminence spread out toward the pole, e another eminence spread out from the pole toward the æquator. it is manifest that in the middle of d the versorium f does not vary; while g is very greatly deflected: but h very little, because it is further removed from d. similarly also the versorium i placed directly toward e does not deviate from the pole: but l and m turn themselves away from the pole a toward the eminence e.
