With an achromatic object-glass composed of three glasses, this experiment may be made step by step, if we do not mind taking out the glasses from their setting. The two convex glasses of crown-glass in contracting the sun's image towards the focus, the concave glass of flint-glass in dilating the image beyond it, exhibit at the edges the usual colours. A convex glass united with a concave one exhibits the colours according to the law of the latter. If all three glasses are placed together, whether we contract the sun's image towards the focus, or suffer it to dilate beyond the focus, coloured edges never appear, and the achromatic effect intended by the optician is, in this case, again attained. 349. But as the crown-glass has always a greenish tint, and as a tendency to this hue may be more decided in large and strong object-glasses, and under certain circumstances produce the compensatory red, (which, however, in repeated experiments with several instruments of this kind did not occur to us,) philosophers have resorted to the most extraordinary modes of explaining such a result; and having been compelled, in support of their system, theoretically to prove the impossibility of achromatic telescopes, have felt a kind of satisfaction in having some apparent ground for denying so great an improvement. Of this, however, we can only treat circumstantially in our historical account of these discoveries. XXIX. COMBINATION OF SUBJECTIVE AND OBJECTIVE EXPERIMENTS. 350. Having shown above (318) that refraction, considered objectively and subjectively, must act in opposite directions, it will follow that if we combine the experiments, the effects will reciprocally destroy each other. 351. Let the sun's image be thrown upwards on a vertical plane, through a horizontally-placed prism. If the prism is long enough to admit of the spectator also looking through it, he will see the image elevated by the objective refraction again depressed, and in the same place in which it appeared without refraction. 352. Here a remarkable case presents itself, but at the same time a natural result of a general law. For since, as often before stated, the objective sun's image thrown on the vertical plane is not an ultimate or unchangeable state of the phenomenon, so in the above operation the image is not only depressed when seen through the prism, but its edges and borders are entirely robbed of their hues, and the spectrum is reduced to a colourless circular form. 353. By employing two perfectly similar prisms placed next each other, for this experiment, we can transmit the sun's image through one, and look through the other. 354.