Even to the last age of French Gothic it is rare to see any mouldings along the ridge of the vault, and the groining, with the exception of the oblique groining which I attempted once to describe to you, is generally simple, and varies very little from first to last. Before I left England this subject had excited my attention, but I did not arrive at any satisfactory conclusion. In some of our Saxon architecture, for example, we have groins, where the ridge of the cross vaulting is considerably arched; but I am not sure that the principal vault forms, on the line of the ridge, a series of arches, as it does in the continental architecture, and so remarkably in the church of St. Germain des Prés. In our Gothic of the thirteenth century I believe the cross vaulting is always arched on the ridge, and the ridge of the principal vault forms also a series of arches nearly flat; but whether these arches are the same in the cross vaulting and principal vaulting, whether they are pointed, or segments of circles, is what I have not been able to determine. In our last style of Gothic architecture, two very different modes of vaulting prevailed; one was nearly that formed by a portion of a sphere cut off by four vertical planes, or like a handkerchief exposed to the wind, and held by the four corners. The arch was, however, seldom quite circular, but had something of a point; the other was formed by courses spreading out beyond each other half round a common centre, so that the vault was formed by a number of funnels touching each other; or rather, while they touched, to form the principal vault, they cut each other to form the cross vaulting. These funnels were universally concave on the section, not straight lines forming regular cones; the intermediate spaces were either left flat, or filled up with pendentives smaller than the funnels, but of the same form. This sort has been called the palm-tree vaulting, because the ribs, gracefully spreading from the tops of the little shafts, present something of the form of a palm-tree, and as this is a much prettier name than funnel-shaped vaults, we will, if you please, adopt it. If you imagine these palm-trees to close against each other in all directions, you will have an arrangement differing less in its appearance from the first than you would readily imagine. If you suppose a square room covered with a true groin, and cut off in the height of the vaulting, the plan at the section would take the form shewn at _a_,