§ III. For, in the preceding inquiry, I have always supposed either that a single shaft would reach to the top of the building, or that the farther height required might be added in plain wall above the heads of the arches; whereas it may often be rather expedient to complete the entire lower series of arches, or finish the lower wall, with a bold string course or cornice, and build another series of shafts, or another wall, on the top of it.
§ IV. This superimposition is seen in its simplest form in the interior shafts of a Greek temple; and it has been largely used in nearly all countries where buildings have been meant for real service. Outcry has often been raised against it, but the thing is so sternly necessary that it has always forced itself into acceptance; and it would, therefore, be merely losing time to refute the arguments of those who have attempted its disparagement. Thus far, however, they have reason on their side, that if a building can be kept in one grand mass, without sacrificing either its visible or real adaptation to its objects, it is not well to divide it into stories until it has reached proportions too large to be justly measured by the eye. It ought then to be divided in order to mark its bulk; and decorative divisions are often possible, which rather increase than destroy the expression of general unity.
§ V. Superimposition, wisely practised, is of two kinds, directly contrary to each other, of weight on lightness, and of lightness on weight; while the superimposition of weight on weight, or lightness on lightness, is nearly always wrong.
1. Weight on lightness: I do not say weight on _weakness_. The superimposition of the human body on its limbs I call weight on lightness: the superimposition of the branches on a tree trunk I call lightness on weight: in both cases the support is fully adequate to the work, the form of support being regulated by the differences of requirement. Nothing in architecture is half so painful as the apparent want of sufficient support when the weight above is visibly passive: for all buildings are not passive; some seem to rise by their own strength, or float by their own buoyancy; a dome requires no visibility of support, one fancies it supported by the air. But passive architecture without help for its passiveness is unendurable. In a lately built house, No. 86, in Oxford Street, three huge stone pillars in the second story are carried apparently by the edges of three sheets of plate glass in the first. I hardly know anything to match the painfulness of this and some other of our shop structures, in which the iron-work is concealed; nor, even when it is apparent, can the eye ever feel satisfied of their security, when built, as at present, with fifty or sixty feet of wall above a rod of iron not the width of this page.