Then comes the group 14 to 18 in stone, derived from the mouldings 1 and 2; first by truncation, 14; then by beading the truncated angle, 15, 16. The occurrence of the profile 16 in the three beautiful windows represented in the uppermost figure of Plate XVIII. Vol. I. renders that group of peculiar interest, and is strong evidence of its antiquity. Then a cavetto is added, 17; first shallow and then deeper, 18, which is the common archivolt moulding of the central Gothic door and window: but, in the windows of the early fourth order, this moulding is complicated by various additions of dog-tooth mouldings under the dentil, as in 20; or the _gabled_ dentil (see fig. 20, Plate IX. Vol. I), as fig. 21; or both, as figs 23, 24. All these varieties expire in the advanced period, and the established moulding for windows is 29. The intermediate group, 25 to 28, I found only in the high windows of the third order in the Ducal Palace, or in the Chapter-house of the Frari, or in the arcades of the Ducal Palace; the great outside lower arcade of the Ducal Palace has the profile 31, the left-hand side being the innermost. Now observe, all these archivolts, without exception, assume that the spectator looks from the outside only: none are complete on both sides; they are essentially _window_ mouldings, and have no resemblance to those of our perfect Gothic arches prepared for traceries. If they were all completely drawn in the plate, they should be as fig. 25, having a great depth of wall behind the mouldings, but it was useless to represent this in every case. The Ducal Palace begins to show mouldings on both sides, 28, 31; and 35 is a _complete_ arch moulding from the apse of the Frari. That moulding, though so perfectly developed, is earlier than the Ducal Palace, and with other features of the building, indicates the completeness of the Gothic system, which made the architect of the Ducal Palace found his work principally upon that church. The other examples in this plate show the various modes of combination employed in richer archivolts. The triple change of slope in 38 is very curious. The references are as follows: