In the case of this later engraved glass, the lapidary’s wheel was applied at times to produce a rough design by a series of burr-like marks, or again the pattern was built up of a number of shallow, mostly oval depressions; in other examples the glass was deeply undercut, so that the designs appear to float round the vessel, to which indeed they are only attached by small rods not easily visible. Of the last kind is the work that may conveniently be called _diatretum_, although it is by no means certain that the _diatretarii_, mentioned by Ulpian and others, were necessarily workers in glass, seeing that carvings of this description, whether in metal, in hard stones, or in our material, were equally in favour at this time. We have, unfortunately, no complete example of this undercut work easily accessible in our public collections. A fragment, however, in the British Museum throws much light upon the process of manufacture. On this piece there remains a portion of the outer frame in the form of a few letters that have formed part of an inscription; most of these letters, however, have been broken away, and we are thus enabled to see the base of the rods that supported them. The sharp angles of these little rods, and the marks on the surface of the glass, point unmistakably to the use of a cutting-tool, nor is there, I think, any trace of soldering at the base of the rods. We must turn again to the marvellous collection of late classical and mediæval objects that has been so long preserved in the treasury of St. Mark’s at Venice for the most complete specimen of this undercut glass. Here will be found a _situla_, or bucket-shaped vessel, of slightly greenish glass, about eleven inches in height (Plate XIV.). On the upper zone is a hunting scene with two horsemen, treated with a certain energy that calls to mind some of the Byzantine and even Sassanian work of the fourth and fifth centuries. Below we have a raised network, or rather grating—for the motive seems to be taken from a grille of iron or bronze—formed of four rows, each built up of fifteen tangential circles bound together at the points of contact. About half of these circles are more or less broken, and neither on the ground nor on the supporting rods thus disclosed was I able on close examination to discover any of those marks of a cutting-tool so prominent on the British Museum fragment. Indeed it is very possible that this late example may be built up of separately cast pieces soldered on to the base.