In this millefiori glass the sections of the canes are arranged with a studied irregularity (so as, in a measure, to mask the spiral arrangement), and a further variety is given by setting up many of them obliquely to the surface. On the other hand we can seldom, perhaps never, find any trace of the distortion, which would inevitably be caused by the subsequent use of the blowing-tube. In other cases, the individual fragments may be built up of irregular longitudinal bands, so as to give the general effect of an agate breccia, as in a fine bowl at South Kensington. When the contorted bands are continuous we have another important type, founded apparently upon the endless varieties of banded agate and other native stones that have been formed by slow deposition in the hollows of rocks. One variety imitates amethystine quartz, but here, as elsewhere, rich combinations of colour, which can have no prototype among natural stones, are often introduced. We have an exceptionally beautiful example of this in certain cigar-shaped alabastra, said to have come from Sidon. Meandering bands of emerald green, powdered with gold, are divided by lines of white and deep blue. Good examples of this ‘peacock’ decoration may be found in the British Museum, at South Kensington, and in the Gréau collection.[24] Allied to these, and still more rare, are the little globular bottles with bands of green and gold, of which there are exquisite specimens in both our great Museums. In the Etruscan Museum of Gregory XVI. in the Vatican, the millefiori glass is well represented by a series of bowls from Greek and Etruscan tombs. There is a choice collection of fragments of millefiori and banded glass in the British Museum,[25] and a still larger one in the Industrial Museum at Vienna. A broken fragment of glass will indeed often tell us more than a complete vase. We can, for example, see from it whether the pattern passes continuously through the whole thickness of the glass, or whether it has been inlaid, or perhaps pressed into the surface when hot. In one case we have a process that reminds us of mosaic work; in the other there is some approach to a _champlevé_ enamel, only with a base of glass instead of metal. In some rare examples we find the glass inlay surrounded by a fine ribbon of gold, suggesting the _cloisonné_ enamels of the Byzantine jeweller. There is a minute example of this delicate work in the Slade collection (_Catalogue_, Pl. III. No. 4). COLOURS OF ROMAN GLASS