In examining the process of the experiment just given, we find that in the one case we have, to appearance, extended the white edge upon the dark surface; in the other we have extended the dark edge upon the white surface, supplanting one by the other, pushing one over the other. We will now endeavour, step by step, to analyse these and similar cases. 204. If we cause the white disk to move, in appearance, entirely from its place, which can be done effectually by prisms, it will be coloured according to the direction in which it apparently moves, in conformity with the above laws. If we look at the disk _a_[3] through a prism, so that it appear moved to _b_, the outer edge will appear blue and blue-red, according to the law of the figure B (fig. 1), the other edge being yellow, and yellow-red, according to the law of the figure C (fig. 1). For in the first case the white figure is, as it were, extended over the dark boundary, and in the other case the dark boundary is passed over the white figure. The same happens if the disk is, to appearance, moved from _a_ to _c_, from _a_ to _d_, and so throughout the circle. 205. As it is with the simple effect, so it is with more complicated appearances. If we look through a horizontal prism (_a b_[4]) at a white disk placed at some distance behind it at _e_, the disk will be raised to _f_, and coloured according to the above law. If we remove this prism, and look through a vertical one (_c d_) at the same disk, it will appear at _h_, and coloured according to the same law. If we place the two prisms one upon the other, the disk will appear displaced diagonally, in conformity with a general law of nature, and will be coloured as before; that is, according to its movement in the direction, _e.g._:[5] 206. If we attentively examine these opposite coloured edges, we find that they only appear in the direction of the apparent change of place. A round figure leaves us in some degree uncertain as to this: a quadrangular figure removes all doubt. 207. The quadrangular figure _a_,[6] moved in the direction _a b_ or _a d_ exhibits no colour on the sides which are parallel with the direction in which it moves: on the other hand, if moved in the direction _a c_, parallel with its diagonal, all the edges of the figure appear coloured.[7] 208. Thus, a former position (203) is here confirmed; viz. to produce colour, an object must be so displaced that the light edges be apparently carried over a dark surface, the dark edges over a light surface, the figure over its boundary, the boundary over the figure. But if the rectilinear boundaries of a figure could be indefinitely extended by refraction, so that figure and background might only pursue their course next, but not over each other, no colour would appear, not even if they were prolonged to infinity. [1] Plate 2, fig. 1.