Society of Antiquaries some bald and unintelligent notes. For the rest, the evidence that might have told the tale of Roman Canterbury was found, ignored, destroyed, or buried.. Today we can only piece together fragments of what might have been a real picture. We can discern a town which looks as if it were a member, though perhaps an unimportant member, of the class of Romano-British country-towns already mentioned (p. 7). We find no difficulty in believing that this country town was also a tribal capital. But if anyone asks for details, the size of the town, the line of its ramparts, the plan of its streets, the fashion of its houses, the periods of its first beginning or its greatest prosperity, we have no proper answers to give. We can merely attempt a critical survey of the extent and value of our present evidence. There is the more reason for this inasmuch as nothing of the kind has been essayed before.
   Our survey begins with the walls. It is a priori likely that Durovernum was walled. It is equally probable that the line of its Roman walls, like those of Rochester and similar towns, was to some extent followed by its medieval ramparts. Three or four pieces of definite evidence can be cited in support of these probabilities. They indicate, first, that two gateways and a part of the medieval wall, on the south side of the town, stand on the site of two Roman gateways and a Roman town-wall. The two medieval gateways are Worth Gate and Riding Gate. These preserved, as late as the 18th century, some ancient arches, which have often been imagined to be Roman work. The gates themselves have since been altered out of all recognition, but sketches survive which support the Roman origin of the vanished masonry. Near Worth Gate, moreover, some masonry was found in 1867—8, which might reasonably be thought part of a Roman town-wall. This evidence suggests that the two gates represent Roman gateways and that the ‘whole southern angle of the medieval enceinte between and near them—some 600 yards in length—follows the course of the Roman wall. This conclusion, however, involves a difficulty. The piece of wall in question lies some way outside the area actually occupied by Roman buildings, and the interval contains both Roman burials and also some watercourses which were seemingly open in Roman times (pp. 69, 71). Neither of these things suits the interior of a Roman town, and their occurrence here is hard to explain away. But the evidence, as at present available, does not lend itself to profitable discussion. The problem, like so many others, must be left to the spade. Nearly the whole stretch of wall now in question is accessible to excavation. If it stands on the old Roman line, Roman masonry can easily be laid bare. When the citizens of Canterbury spend a few pounds on digging here, the discussion of the problem ,may be resumed.
   From the Riding Gate, a useful piece of evidence, only lately identified, carries the Roman wall nearly 600 yards to the north-eastward, presumably on the line of its medieval successor. To the east of the Cathedral, on the site of the former Quenin Gate, long covered by sheds, is a blackened fragment of the jamb and arch of a third Roman gateway (fig. 13). The remains consist of large jamb-stones and the first few courses of a brick arch embedded in the existing town wall, and the whole structure was apparently similar in type to the supposed Roman Worth and Riding gates destroyed in the 18th century (P1. XI).