Churches, it will be recollected, were commonly built with a wide nave and narrower aisles, and it was in the Norman period customary to vault the aisles and cover the nave with a ceiling. There was no difficulty in so spacing the distances apart of the piers of the main arcade that the compartments (usually termed bays) of the aisle should be square on plan; and it was quite possible, without doing more than the Romans had done, to vault each bay of the aisles with a semicircular intersecting vault (_i.e._ one which has the appearance of a semicircular or waggon-head vault, intersected by another vault of the same outline and height). This produced a simple series of what are called groined or cross vaults, which allowed height to be given to the window heads of the aisle and to the arcades between the aisles and nave. After a time it was desired to vault the nave also, and to adopt for it an intersecting vault, so that the heads of the windows of the clerestory might be raised above the springing line of the vault, but so long as the arches remained semicircular, this was very difficult to accomplish. The Romans would probably have contented themselves with employing a barrel vault and piercing it to the extent required by short lateral vaults, but the result would have been an irregular, weak, curved line at each intersection with the main vault; and the aisle vaults having made the pleasing effect of a perfectly regular intersection familiar, this expedient does not seem to have found favour, at any rate in England. Other expedients were however tried, and with curious results. It was for example attempted to vault the nave with a cross vault, embracing two bays of the arcade to one of the vault, but the wall space so gained was particularly ill suited to the clerestory windows, as may be seen by examining the nave of St. Stephen's at Caen. In short, if the vaulting compartment were as wide as the nave one way, but only as wide as the aisle the other way, and semicircular arches alone were employed, a satisfactory result seemed to be unattainable. In the search for some means of so vaulting a bay of oblong plan that the arches should spring all at one level, and the groins or lines of intersection should cross one another in the centre of the ceiling, the idea either arose or was suggested that the curve of the smaller span should be a pointed instead of a semicircular arch. The moment this was tried all difficulty vanished, and groined (_i.e._ intersecting) vaults, covering compartments of any proportions became easy to design and simple to construct, for if the vault which spanned the narrow way of the compartment were acutely pointed, and that which spanned it the wide way were either semicircular or flatly-pointed, it became easy to arrange that the startings of both vaults should be at the same level, and that they should rise to the same height, which is the condition essential to the production of a satisfactory intersection.