Theory of Colours — trans. Charles Lock Eastlake (1840)

A just impression with regard to this important phenomenon will, however, much depend on some nice and even troublesome conditions, which are necessary to produce the illusion in question. Paper should be tinged with vermilion or the best minium for the red square, and with deep indigo for the blue square. The blue and red prismatic edges will then unite imperceptibly with the real surfaces where they are respectively homogeneous; where they are not, they vitiate the colours of the squares without producing a very distinct middle tint. The real red should not incline too much to yellow, otherwise the apparent deep red edge above will be too distinct; at the same time it should be somewhat yellow, otherwise the transition to the yellow border will be too observable. The blue must not be light, otherwise the red edge will be visible, and the yellow border will produce a too decided green, while the violet border underneath would not give us the impression of being part of an elongated light blue square.

269.

All this will be treated more circumstantially hereafter, when we speak of the apparatus intended to facilitate the experiments connected with this part of our subject.[2] Every inquirer should prepare the figures himself, in order fairly to exhibit this specimen of ocular deception, and at the same time to convince himself that the coloured edges, even in this case, cannot escape accurate examination.

270.

Meanwhile various other combinations, as exhibited in the plate, are fully calculated to remove all doubt on this point in the mind of every attentive observer.

271.

If, for instance, we look at a white square, next the blue one, on a black ground, the prismatic hues of the opposite edges of the white, which here occupies the place of the red in the former experiment, will exhibit themselves in their utmost force. The red edge extends itself above the level of the blue almost in a greater degree than was the case with the red square itself in the former experiment. The lower blue edge, again, is visible in its full force next the white, while, on the other hand, it cannot be distinguished next the blue square. The violet border underneath is also much more apparent on the white than on the blue.

272.

If the observer now compares these double squares, carefully prepared and arranged one above the other, the red with the white, the two blue squares together, the blue with the red, the blue with the white, he will clearly perceive the relations of these surfaces to their coloured edges and borders.

273.

The edges and their relations to the coloured surfaces appear still more striking if we look at the coloured squares and a black square on a white ground; for in this case the illusion before mentioned ceases altogether, and the effect of the edges is as visible as in any case that has come under our observation. Let the blue and red squares be first examined through the prism. In both the blue edge now appears above; this edge, homogeneous with the blue surface, unites with it, and appears to extend it upwards, only the blue edge, owing to its lightness, is somewhat too distinct in its upper portion; the violet border underneath it is also sufficiently evident on the blue. The apparent blue edge is, on the other hand, heterogeneous with the red square; it is neutralised by contrast, and is scarcely visible; meanwhile the violet border, uniting with the real red, produces a hue resembling that of the peach-blossom.

274.

If thus, owing to the above causes, the upper outlines of these squares do not appear level with each other, the correspondence of the under outlines is the more observable; for since both colours, the red and the blue, are darks compared with the white (as in the former case they were light compared with the black), the red edge with its yellow border appears very distinctly under both. It exhibits itself under the warm red surface in its full force, and under the dark blue nearly as it appears under the black: as may be seen if we compare the edges and borders of the figures placed one above the other on the white ground.

275.