Take as an illustration of the contrary practice such a masterpiece in its way as Gerome's "Eminence Grise." In this picture, skilfully and satisfactorily composed, the relative values of all the colors are admirably, even beautifully, observed. The correspondence of the gamut of values to that of the light and dark scale of such an actual scene is perfect. Before Manet, one could have said that this is all that is required or can be secured, arguing that exact _imitation_ of local tints and general tone is impossible, owing to the difference between nature's highest light and lowest dark, and the potentialities of the palette. In other words, one might have said, that inasmuch as you can squeeze absolute white and absolute black out of no tubes, the thing to do is first to determine the scale of your picture and then make every note in it bear the same relation to every other that the corresponding note in nature bears to its fellows in its own corresponding but different scale. This is what Gerome has done in the "Eminence Grise"--a scene, it will be remembered, on a staircase in a palace interior. Manet inquires what would happen to this house of cards shored up into verisimilitude by mere _correspondence_, if Gerome had been asked to cut a window in his staircase and admit the light of out-of-doors into his correspondent but artificial scene. The whole thing would have to be done over again. The scale of the picture running from the highest palette white to the lowest palette dark, and yet the key of an actual interior scene being much nearer middle-tint than the tint of an actual out-of-doors scene, it would be impossible to paint with any verisimilitude the illumination of a window from the outside, the resources of the palette having already been exhausted, every object having been given a local value solely with relation, so far as truth of representation is concerned, to the values of every other object, and no effort being made to get the precise value of the object as it would appear under analogous circumstances in nature.