This illusion may be dissipated by a process of reasoning arising from the introduction of a new circumstance in the experiment. Thus, let RL, Fig. 17, be the inverted seal, which consequently appears raised, and let an opaque and unpolished pin, A, be placed on one side of the seal. Its shadow will be of course opposite the candle as at B. In this case the seal, which had become a cameo by its inversion, will now sink down into a cavity by the introduction of the pin and its shadow; for as the pin and its shadow are inverted, as shown in Fig. 18, while the candle retains its place, the shadow of the pin falling in the direction AB is a stronger proof to the eye that the light is coming from the right hand, than the actual knowledge of the candle being on the left hand, and therefore the cameo necessarily sinks into a cavity, or the shadow is now on the same side as the light. This experiment will explain to us why on some occasions an acute observer will elude the deception, while every other person is subject to it. Let us suppose that a particle of dust, or a little bit of wax, capable of giving a shadow, is adhering to the surface of the seal, an ordinary observer will take no notice of this, or if he does, he will probably not make it a subject of consideration, and will therefore see the head on the seal raised into a cameo; but the attentive observer, noticing the little protuberance, and observing that its shadow lies to the left of it, will instantly infer that the light comes in that direction, and will still see the seal hollow. [Illustration: _Fig. 18._] I have already mentioned that in some cases even the sense of touch does not correct the erroneous perception. We of course feel that the part of the hollow on which the finger is placed is actually hollow; but if we look at the other part of the hollow it will still appear raised. By using two candles yielding different degrees of light, and thus giving an uncertainty to the direction of the light, we may weaken the illusion in any degree we choose, so as to overpower it by touch, or by a process of reasoning. I have had occasion to observe a series of analogous phenomena arising from the same cause, but produced without any instrument for inverting the object. If AB, for example, is a plate of mother-of-pearl, and LR a circular or any other cavity (Fig. 19) ground or turned in it, then if this cavity is illuminated by a candle or a window at S, in place of there being a shadow of the margin L of the hollow next the light, as there would have been had the body been opaque, a quantity of bright refracted light will appear where there would have been a shadow, and the rest of the cavity will be comparatively obscure, as if it were in shade. The necessary consequence of this is, that the cavity will appear as an elevation when seen only by the naked eye, as it is only an elevated surface that could have its most luminous side at L. [Illustration: _Fig. 19._]