Geographically, Britain consists of two parts: (1) the comparatively flat lowlands of the south, east and midlands, suitable to agriculture and open to easy intercourse with the continent, i.e. with the rest of the Roman empire; (2) the district consisting of the hills of Devon and Cornwall, of Wales and of northern England, regions lying more, and often very much more, than 600 ft. above the sea, scarred with gorges and deep valleys, mountainous in character, difficult for armies to traverse, ill fitted to the peaceful pursuits in agriculture. These two parts of the province differ also in their history. The lowlands, as we have seen, were conquered easily and quickly. The uplands were hardly subdued completely till the end of the 2nd century. They differ, thirdly, in the character of their Roman occupation. The lowlands were the scene of civil life. Towns, villages and country houses were their prominent features; troops were hardly seen in them save in some fortresses on the edge of the hills and in a chain of forts built in the 4th century to defend the south-east coast, the so-called Saxon Shore. The uplands of Wales and the north presented another spectacle. Here civil life was almost wholly absent.

No country town or country house has been found more than 20 m. north of York or west of Monmouthshire. The hills were one extensive military frontier, covered with forts and strategic roads connecting them, and devoid of town life, country houses, farms or peaceful civilized industry. This geographical division was not reproduced by Rome in any administrative partitions of the province. At first the whole was governed by one legatus Augusti of consular standing. Septimius Severus made it two provinces, Superior and Inferior, with a boundary which probably ran from Humber to Mersey, but we do not know how long this arrangement lasted. In the 5th century there were five provinces, Britannia Prima and Secunda, Flavia and Maxima Caesariensis and (for a while) Valentia, ruled by praesides and consulares under a vicartus, but the only thing known of them is that Britannia Prima included Cirencester.

Fig 1.  Plan of Housesteads. Fig. 1. - Plan of Housesteads (Borcovicium) on Hadrian's Wall.

The army which guarded or coerced the province consisted, from the time of Hadrian onwards, of (1) three legions, the Second at Isca Silurum (Caerleon-on-Usk, q.v.), the Ninth at Eburācum (q.v.; now York), the Twentieth at Deva (q.v.; now Chester), a total of some 15,000 heavy infantry; and (2) a large but uncertain number of auxiliaries, troops of the second grade, organized in infantry cohorts or cavalry alae, each 500 or 1000 strong, and posted in castella nearer the frontiers than the legions. The legionary fortresses were large rectangular enclosures of 50 or 60 acres, surrounded by strong walls of which traces can still be seen in the lower courses of the north and east town-walls of Chester, in the abbey gardens at York, and on the south side of Caerleon. The auxiliary castella were hardly a tenth of the size, varying generally from three to six acres according to the size of the regiment and the need for stabling. Of these upwards of 70 are known in England and some 20 more in Scotland. Of the English examples a few have been carefully excavated, notably Gellygaer between Cardiff and Brecon, one of the most perfect specimens to be found anywhere in the Roman empire of a Roman fort dating from the end of the 1st century A.D.; Hardknott, on a Cumberland moor overhanging Upper Eskdale; and Housesteads on Hadrian's wall.

In Scotland excavation has been more active, in particular at the forts of Birrens, Newstead near Melrose, Lyne near Peebles, Ardoch between Stirling and Perth, and Castle Cary, Rough Castle and Bar Hill on the wall of Pius. The internal arrangements of all these forts follow one general plan. But in some of them the internal buildings are all of stone, while in others, principally (it seems) forts built before 150, wood is used freely and only the few principal buildings seem to have been constructed throughout of stone.

We may illustrate their character from Housesteads, which, in the form in which we know it, perhaps dates from Septimius Severus. This fort measures about 360 by 600 ft. and covers a trifle less than 5 acres. Its ramparts are of stone, and its north rampart coincides with the great wall of Hadrian. Its interior is filled with stone buildings. Chief among these (see fig. 1), and in the centre of the whole fort, is the Headquarters, in Lat. Principia or, as it is often (though perhaps less correctly) styled by moderns, Praetorium. This is a rectangular structure with only one entrance which gives access, first, to a small cloistered court (x. 4), then to a second open court (x. 7), and finally to a row of five rooms (x. 8-12) containing the shrine for official worship, the treasury and other offices. Close by were officers' quarters, generally built round a tiny cloistered court (ix., xi., xii.), and substantially built storehouses with buttresses and dry basements (viii.). These filled the middle third of the fort. At the two ends were barracks for the soldiers (i.-vi., xiii.-xviii.). No space was allotted to private religion or domestic life.

The shrines which voluntary worshippers might visit, the public bath-house, and the cottages of the soldiers' wives, camp followers, etc., lay outside the walls. Such were nearly all the Roman forts in Britain. They differ somewhat from Roman forts in Germany or other provinces, though most of the differences arise from the different usage of wood and of stone in various places.

Forts of this kind were dotted all along the military roads of the Welsh and northern hill-districts. In Wales a road ran from Chester past a fort at Caer-hyn (near Conway) to a fort at Carnarvon (Segontium). A similar road ran along the south coast from Caerleon-on-Usk past a fort at Cardiff and perhaps others, to Carmarthen. A third, roughly parallel to the shore of Cardigan Bay, with forts at Llanio and Tommen-y-mur (near Festiniog), connected the northern and southern roads, while the interior was held by a system of roads and forts not yet well understood but discernible at such points as Caer-gai on Bala Lake, Castle Collen near Llandrindod Wells, the Gaer near Brecon, Merthyr and Gellygaer. In the north of Britain we find three principal roads. One led due north from York past forts at Catterick Bridge, Piers Bridge, Binchester, Lanchester, Ebchester to the wall and to Scotland, while branches through Chester-le-Street reached the Tyne Bridge (Pons Aelius) at Newcastle and the Tyne mouth at South Shields. A second road, turning north-west from Catterick Bridge, mounted the Pennine Chain by way of forts at Rokeby, Bowes and Brough-under-Stainmoor, descended into the Eden valley, reached Hadrian's wall near Carlisle (Luguvallium), and passed on to Birrens. The third route, starting from Chester and passing up the western coast, is more complex, and exists in duplicate, the result perhaps of two different schemes of road-making. Forts in plenty can be detected along it, notably Manchester (Mancunium or Mamucium), Ribchester (Bremetennacum), Brougham Castle (Brocavum), Old Penrith (Voreda), and on a western branch, Watercrook near Kendal, Waterhead near the hotel of that name on Ambleside, Hardknott above Eskdale, Maryport (Uxellodūnum), and Old Carlisle (possibly Petriana). In addition, two or three cross roads, not yet sufficiently explored, maintained communication between the troops in Yorkshire and those in Cheshire and Lancashire. This road system bears plain marks of having been made at different times, and with different objectives, but we have no evidence that any one part was abandoned when any other was built.

There are signs, however, that various forts were dismantled as the country grew quieter. Thus, Gellygaer in South Wales and Hardknott in Cumberland have yielded nothing later than the opening of the 2nd century.

Fig 2.  Hadrian's Wall. Fig. 2. - Hadrian's Wall.

From Social England, by permission of Cassell & Co., Ltd.

Besides these detached forts and their connecting roads, the north of Britain was defended by Hadrian's wall (figs. 2 and 3). The history of this wall has been given above. The actual works are threefold. First, there is that which to-day forms the most striking feature in the whole, the wall of stone 6-8 ft. thick, and originally perhaps 14 ft. high, with a deep ditch in front, and forts and "mile castles" and turrets and a connecting road behind it. On the high moors between Chollerford and Gilsland its traces are still plain, as it climbs from hill to hill and winds along perilous precipices. Secondly, there is the so-called "Vallum," in reality no vallum at all, but a broad flat-bottomed ditch out of which the earth has been cast up on either side into regular and continuous mounds that resemble ramparts. Thirdly, nowhere very clear on the surface and as yet detected only at a few points, there are the remains of the "turf wall," constructed of sods laid in regular courses, with a ditch in front. This turf wall is certainly older than the stone wall, and, as our ancient writers mention two wall-builders, Hadrian and Septimius Severus, the natural inference is that Hadrian built his wall of turf and Severus reconstructed it in stone.

The reconstruction probably followed in general the line of Hadrian's wall in order to utilize the existing ditch, and this explains why the turf wall itself survives only at special points. In general it was destroyed to make way for the new wall in stone. Occasionally (as at Birdoswald) there was a deviation, and the older work survived. This conversion of earthwork into stone in the age of Severus can be paralleled from other parts of the Roman empire.

Fig 3.  Section of Hadrian's Wall. Fig. 3. - Section of Hadrian's Wall.

The meaning of the vallum is much more doubtful. John Hodgson and Bruce, the local authorities of the 19th century, supposed that it was erected to defend the wall from southern insurgents. Others have ascribed it to Agricola, or have thought it to be the wall of Hadrian, or even assigned it to pre-Roman natives. The two facts that are clear about it are, that it is a Roman work, no older than Hadrian (if so old), and that it was not intended, like the wall, for military defence. Probably it is contemporaneous with either the turf wall or the stone wall, and marked some limit of the civil province of Britain. Beyond this we cannot at present go.