There is, however, one division of classical glass—we can hardly call it Roman, although most of the finer specimens may be traced back to Rome or to the tombs of Central and Southern Italy—which forms in some degree a transition from our primitive family to the true blown glass of imperial times. This is the so-called Millefiori Glass. We have, doubtless, in this a development of the ‘fused mosaics’ of the Egyptians, worked out on a larger scale, and employed for other objects than flat slabs and fragments for inlay. In the millefiori bowls of Greco-Roman times we can distinguish two predominant types: the madrepore design in the first place, which closely imitates the pattern on a polished slab of coralline limestone, with the addition that the ground is of a deep translucent green or of a purple of subdued tone. In this class may be placed such exceptional pieces as the bowl from Crete, in the British Museum; here we have rosettes of yellow, green, and red upon an opaque ground of a rich blue. The second type is equally characteristic, but more difficult to describe. Short, loosely rolled scrolls of an opaque white float in a more or less transparent base, interspersed with a few quadrangular masses of gilt glass. It would be difficult to say what natural substance is imitated in this case—perhaps some kind of fossiliferous _lumachella_ marble, which may have been in vogue at one time at Alexandria. We may be quite sure that the Roman glass-workers would not have failed to imitate the famous Murrhine vases, which seem to have been originally carved from a natural stone, and it is among the millefiori glasses that such imitations may probably be looked for. These millefiori bowls are evidently built up with more or less spirally arranged fragments of glass mosaic,[23] the individual pieces having been probably cut from a cane of glass, itself formed by a combination of minute rods, as in the case of the Egyptian ‘fused mosaics.’ These pieces were arranged in the mould in a coil, starting from the centre, but how far, if at all, during the subsequent partial fusion, they were subjected to any blowing operation, is a moot point. In any case, the final effect is the result of an elaborate process of cutting on the wheel and subsequent polishing. [Illustration: _PLATE V_ ROMAN MILLEFIORI GLASS BOWLS, IMITATING NATIVE STONES]