natural resources, but it was not wealthy. It was a land of small towns an prosperous rural estates, remote from the splendours and the troubles of the Mediterranean, comfortable, unimportant.
   If we turn now to Kent, we shall find, as we might expect, that the geographical and geological features of the county explain the distribution the Roman settled sites there. Almost the whole of the central area consist of high chalk downs which, to a large extent treeless and waterless, were no favoured by the more Romanised elements in the population. Whether, and how far, they (like the Sussex downs) supported native village groups we do not at present know; but it is certain that, for the mansions and farmhouse of the more prosperous inhabitants of the county, we must look rather to the flanks than to the ridges of the downlands. There, on the south, the fertile sandy clay of the gault from Folkestone to Maidstone and Sevenoaks carried a fairly extensive series of ‘villas‘ and other settlements. To the north between the downs and the sea, a fringe of lightly wooded sands attracted first, the great arterial Watling Street, and, secondly, a continuous chain of towns and ‘ villas ‘ from Richborough to Canterbury and London. These two zones of habitation are joined towards ‘the west by two transverse valleys—those of the Medway and the Dart—along both of which flowed the tide of Roman occupation. Beyond all these regions, to the south the great forest of the Weald, 120 miles in length and over 20 in breadth, formed a natural barrier to which Kent has at more than one period owed both isolation and independence. In Roman times this forest-belt constituted a hiatus in the occupation of the countryside, but, save perhaps in the fourth century, formed, so far as we know, no sort of political or cultural frontier.⁴
   Nevertheless, however undesirable for permanent habitation , the chalk downs had one sterling merit. They offered an expanse of convenient open country towards the three natural ports of south-eastern Britain. To the south, Lymne or Lympne, now long cut off from the sea, then afforded sheltered anchorage which was readily linked with Canterbury to the north-ward by the long, straight Stane Street. At Dover, the Dour cut its way through cliffs which at that time flanked a now-vanished haven. To the north, the Wansum, which then, as a navigable strait, cleft the chalk plateau of Thanet from the mainland, offered at both ends —but particularly towards the south, at Richborough convenient inlets where a fleet might ride at anchor. At all these points harbour—towns, linked with Canterbury by a rigid framework of roads, came into being during the early years of the occupation. And it was inevitable that, at all these points, in the latter days when Saxon pirates came thrusting into our south—eastern shores, fortresses should spring up to guard the gateways to the province. Since this chain of harbours was thus the dominant Imperial interest in Kent, we may justly begin our detailed survey with them.
   ⁴ In the quadruple subdivision of Roman Britain after A.D. 296 Britannia Secunda may have coincided roughly with Kent, but our only evidence is Giraldus Cambrensis, writing about A.D. 1205. (See F. Haverfield, Archaeologia Oxoniensis, 1892— 5 , p .223.)