§ 16. Now, it will be observed by any one familiar with ordinary principles of arrangement of form (on which principles I shall insist at length in another place), that while the dream introduces these changes bearing on the expression of the scene, it is also introducing other changes, which appear to be made more or less in compliance with received rules of composition,[10] rendering the masses broader, the lines more continuous, and the curves more graceful. But the curious part of the business is, that these changes seem not so much to be wrought by imagining an entirely new condition of any feature, as by _remembering_ something which will fit better in that place. For instance, Turner felt the bank on the right ought to be made more solid and rocky, in order to suggest firmer resistance to the stream, and he turns it, as will be seen by comparing the etchings, into a kind of rock buttress, to the wall, instead of a mere bank. Now, the buttress into which he turns it is very nearly a facsimile of one which he had drawn on that very St. Gothard road, far above, at the Devil's Bridge, at least thirty years before, and which he had himself etched and engraved, for the Liber Studiorum, although the plate was never published. Fig. 1 is a copy of the bit of the etching in question. Note how the wall winds over it, and observe especially the peculiar depression in the middle of its surface, and compare it in those parts generally with the features introduced in the later composition. Of course, this might be set down as a mere chance coincidence, but for the frequency of the cases in which Turner can be shown to have done the same thing, and to have introduced, after a lapse of many years, memories of something which, however apparently small or unimportant, had struck him in his earlier studies. These instances, when I can detect them, I shall point out as I go on engraving his works; and I think they are numerous enough to induce a doubt whether Turner's composition was not universally an arrangement of remembrances, summoned just as they were wanted, and set each in its fittest place. It is this very character which appears to me to mark it as so distinctly an act of dream-vision; for in a dream there is just this kind of confused remembrance of the forms of things which we have seen long ago, associated by new and strange laws. That common dreams are grotesque and disorderly, and Turner's dream natural and orderly, does not, to my thinking, involve any necessary difference in the real species of act of mind. I think I shall be able to show, in the course of the following pages, or elsewhere, that whenever Turner really tried to _compose_, and made modifications of his subjects on principle, he did wrong, and spoiled them; and that he only did right in a kind of passive obedience to his first vision, that vision being composed primarily of the strong memory of the place itself which he had to draw; and secondarily, of memories of other places (whether recognized as such by himself or not I cannot tell), associated, in a harmonious and helpful way, with the new central thought.