The Romans, as many of their modern followers, were for a time so inconsistent as not only to limit the arch, like the architrave, to the thickness of the upper diameter of the column, but actually interposed, without a shadow of use, a bit of entablature between the column and the arch; thus, instead of doing all they could to give steadiness to the spring of the arch, they made it as tottering in its construction as possible. This was corrected by the Romans of the Lower Empire, and the arch was placed by them, as reason would dictate, directly upon the capital, or (still more sensibly) on a strong flat impost laid on the capital; and for this most reasonable step they have in after ages been pronounced barbarous! The Romanesque architects, taught by common sense rather than by precedent, followed their example. If they imitated or re-used the Corinthian capital, they laid upon its fragile abacus a more trustworthy impost, and to give greater steadiment to the foot of the arch they made it somewhat wider than the diameter of the column--a practice which pervades Mediæval architecture, and contributes greatly both to its good construction and its beauty. The system of constructing _doorways_ is directly derived from what I have already described--as many recesses being given to the jambs as the arch has rims, and these decorated with columns if thought good. The head is often filled in with a tympanum supported by corbels in the jambs, both as a field for sculptured decoration, and to make the door itself square instead of arch-formed. If this is not done, the inner arches are made to spring from a higher level, to allow the doors to open without catching against them. The windows show the same regard to reason. The inside is nearly always widely splayed, to spread the light equally in the room. The external recess depended partly on the degree of architectural character aimed at, and partly on the depth required for the arch. Where the openings were but narrow, and the resources small, one arch-rim would suffice; and this would often be chamfered at the edges, to prevent obstruction to light. If the opening were wider, and so required a deeper arch, or if the architectural effect aimed at were greater, we find two or more such orders as the above, with, perhaps, columns supporting the outer ones; the receding orders, in either case, doing away with undue obstruction of light or view; the sill always well sloped, to throw off the water, and having usually a string-course below, to prevent it from running down and discolouring the walls. In all this, strict regard to practical reason and utility is manifest; every step is argued out on the basis of construction and requirement, and every decoration is founded on, and results from, the conclusions come to on these practical grounds.