It would seem then that, by 47 or 48, practically the whole lowlands were in the hands of the Romans. Whether Chester had already been occupied or (as seems likelier) was first garrisoned when Ostorius attacked the Decangi, must remain uncertain; it must in any case have been occupied soon (Eph. Epigr. vii. 903; Domaszewski, Rhein. Mus. xlviii. 344). Caerleon, connected by Mommsen with Tac. Ann. xii. 32, presents more difficulty, since it has yielded hardly any datable remains earlier than about A.D. 70–80; however, no other site can be suggested on our present evidence for the hiberna of the Second Legion Augusta before 70. Wroxeter rests its claim to a fortress on two early inscriptions of Legio xiv. (Vict. Hist. Shropshire, i. 243, 244), and this may be adequate, though Domaszewski doubts it. The course of Watling Street seems to show that Wroxeter was occupied before the troops pushed on to Chester.
Mommsen’s account of the Boadicean revolt (pp. 179–181) is famous for his denunciation of Tacitus as “the most unmilitary of all authors.” It must be conceded that Tacitus is unmilitary—not so much because he is condensed or discontinuous or ignorant of geography (E. G. Hardy, Journ. Phil. xxxi. 123), as because he has a literary horror of all technical detail, and desires to give the general effect of each situation without distracting the reader by vexatious precision and difficult minutiae. But in this case his narrative (Ann. xii. 32 foll.) is better than Mommsen (or indeed Domaszewski) allows. Paullinus doubtless marched to London, as Horsley long ago observed, because it lay on the road (Watling Street) from Chester to Colchester; that he hurried on in front of his main forces is implied in the iam at the beginning of c. 34.
The conquest of Wales (p. 182) was completed, as Mommsen says, in the decade A.D. 70–80. But his statements require some re-wording. Roman remains are not “completely absent” in the interior; the continuance of native resistance to Rome is very doubtful; the existence of Celtic speech and nationality in Wales to-day is—in large part, at least—due to a Celtic revival in the late fourth or the fifth century, and to immigration of new Celtic elements at that time, and cannot therefore be cited as here. So far as present evidence goes, the district as a whole seems during the first, second, and third centuries to have closely resembled the similar mountainous districts of northern England, save only that the Welsh tribes never revolted after A.D. 80, while the Brigantes gave trouble throughout the second century. The same system of small auxiliary castella was established in Wales as in northern England. These forts are at present almost wholly unexplored. But we can detect unquestionable examples at Caerhun (Canovium, Eph. vii. 1099) and Carnarvon, in the north; at Tommen-y-mur, Llanio-i-sa, and Caio, in the west; at Caergai (Eph. vii. 863), Castle Collen near Llandrindod (ibid. 862), Caersws in the upper valley of the Severn, and the Gaer near Brecon, in the interior; at Gelligaer (Trans. Cardiff Nat. Soc. xxxv. 1903), Merthyr Tydfil, Cardiff, Abergavenny, Usk, in the south, besides others not yet satisfactorily identified as military sites. Several of these have yielded remains suggestive of the first century, and indeed of the Flavian period. The only one as yet properly excavated, Gelligaer, seems to have been occupied under the Flavians, and dismantled after no very long occupation, probably early in the second century. Such dismantlement suggests that the land was then growing less unquiet. But Wales never reached any higher degree of Roman civilisation than the north of England. Towns and country houses were always rare, and its population lived mostly, it would seem, in primitive villages (Arch. Cambrensis, 1907). Later on, in the fourth century, Celts began to come in from Ireland, much as other barbarians entered other parts of the Empire, but their dates and numbers are very little known; see my Romanisation of Roman Britain, pp. 27 foll. and refs. there given.
The invasion of Caledonia (p. 183) by Agricola has been illustrated by recent discoveries. As I have pointed out elsewhere, we have traces of Agricola’s line of forts (Tac. Agr. 23) at Camelon (Proc. Soc. Antiq. Scotland, xxxv. fig. 10) and Bar Hill (G. Macdonald, Roman Forts on the Bar Hill, Glasgow, 1906). Farther north, near the junction of the Tay and Isla, at Inchtuthill, in the policies of Delvine, a large encampment of Roman type has yielded a few objects datable to the Agricolan age (Proc. Soc. Antiq. Scot. xxxvi. pp. 237, 242), and may give a clue to the site of Mons Graupius. Farther south, the large fort lately excavated by Mr. James Curle, at Newstead, near Melrose (C. I. L. vii. 1080, 1081; Scottish Hist. Review, 1908), was certainly occupied in the Agricolan age. To this date, too, may perhaps be assigned the siege works round the native fortress on Birrenswark in Dumfriesshire, with their leaden sling-bullets (Proc. Soc. Antiq. Scot. xxxiii. 198 foll.). Evidence that the Legio ii. Adiutrix was then posted at Chester, probably forming a double-legion fortress with Legio xx., was obtained in the excavations of 1890 (Catal. of the Grosvenor Museum, Chester (1900), pp. 7 foll. and Nos. 23–35). An inscription from Camelon with the letters MILITES L·II·A·DIE may have been intended to refer to this legion, but is a forgery (Class. Review, xix. 57). No trace of Agricolan or of Flavian remains has yet been found on the line of Hadrian’s Wall, except at two points, which, strictly speaking, are near but not on the wall, Carlisle (Luguvallium), and Corbridge (Corstopitum), where the two great north roads pass on towards Caledonia. For the influence of continental frontier troubles on the British operations of Agricola see also Ritterling, Jahreshefte des österr. arch. Instituts, vii. 26.
The years between the recall of Agricola and the building of Hadrian’s Wall (roughly A.D. 85–120) are a historical blank. Even the position of the northern frontier during these years is unknown. The Romans seem to have soon withdrawn from the line of the Clyde and Forth (Macdonald, Bar Hill, pp. 14, 15). Whether they also withdrew south of Cheviot is not quite clear, in the present state of the Newstead excavations.
Hadrian’s Wall from Tyne to Solway (p. 186) has assumed a very different historical appearance since Mommsen wrote his paragraphs on it in 1885. Then, the theory of Hodgson and Bruce held the field—that the stone wall which is still visible, and the double rampart and ditch to the south of it (called by English antiquaries the “Vallum”), were both Hadrian’s work, the wall for defence against Caledonia and the “Vallum” for defence against stray foes from the south. This view was accepted by Mommsen. But later excavation and observation have shown that the “Vallum” cannot be regarded as a military work—though it is certainly Roman and connected with the wall. Excavations have also shown that the wall itself falls into two periods. At Birdoswald (Amboglanna) there was first a wall of turf (murus caespiticius); later, almost but not quite on the same line, came the wall of stone and the fort of Amboglanna in its present form. Similarly at Chesters (Cilurnum) two building periods are discernible; the character of the first is obscure, but the stone wall and the fort of Cilurnum belong unquestionably to the second (Cumberland Arch. Soc. xiv. 187, 415, xv. 180, 347, xvi. 84; Arch. Aeliana, xxiii. 9). As our ancient authorities persistently mention two wall-builders, Hadrian and Severus, and as the earlier wall of turf can be assigned to no one but Hadrian, it would seem that we may assume a first fortification of the Tyne and Solway line in turf about A.D. 120, and a rebuilding in stone, on almost exactly the same tracé, about A.D. 208 by Severus. The "Vallum" seems to have been built in relation to one or the other—more probably the earlier—of these stone walls, and may represent a civil frontier contemporaneous with it (Mommsen, Gesammelte Schriften, v. 461; Pelham, Trans. Cumberland Arch. Soc. xiv. 175). The attempt of Dr. E. Krueger (Bonner Jahrbücher, cx. 1–38) to show that the "Vallum" is an earlier independent work, built by Hadrian, while the turf and stone walls are post-Hadrianic, seems to me both unproven and contradicted by recent excavations.
Mommsen’s account of the Wall of Pius between Forth and Clyde and of the Roman occupation of Scotland also needs modification. Statistics of coins found in Scotland (printed in Antonine Wall Report, 1899, pp. 158 foll., confirmed by all later finds) show that the Romans had retired south of Cheviot by about A.D. 180, and never reoccupied the positions thus lost. The mass of inscriptions, to which Mommsen alludes, also contains nothing later than the reign of Marcus. It becomes, therefore, impossible to connect the Wall of Pius with the literary evidence relating to wall-building by Severus. That evidence must belong to the Tyne and Solway. The length which it assigns to the wall, cxxxii. miles, suits the southern line best. The numeral in any case needs emendation, but it is as easy to read lxxxii. as xxxii., and 82 Roman miles fit closer to the length of the southern line (73–1/2 English miles) than do 32 Roman miles to the 36–1/2 English miles of the northern wall. Our knowledge of the northern wall itself and of forts either north of it, like Ardoch, or south, like Lyne and Newstead, has been much widened by excavation, but the gain has been rather to the archaeologist than to the pure historian.
In the later history of north Britain the chief recent addition has been evidence of a serious rising about A.D. 158, which perhaps covered all the land of the Brigantes from Derbyshire to Dumfriesshire. Inscriptions found at Birrens, at Netherby between Birrens and Carlisle, at Newcastle-on-Tyne, and at Brough in north Derbyshire, mention a governor Iulius Verus as then specially active, and special reinforcements as then arriving from Germany (Proc. Soc. Antiq. Scot. xxxviii. 454). It is natural to connect these with the words of Pausanias (cited on p. 188, note 2), and the connection had the approval of Mommsen. For the division of the province into two by Severus see Domaszewski, Rangordnung, p. 173. The boundary between the two provinces is unknown; perhaps a line from the Humber to the Mersey is not altogether improbable. Nor is there evidence to show how long the division lasted.
Of the civil life and Romanisation of Britain (pp. 191–4) I have written somewhat fully in a paper on The Romanisation of Roman Britain. Here I may indicate some points. Mommsen’s view that the cantonal system adopted in Gaul was dropped in Britain is opposed by an inscription found at Caerwent in 1903, which records the erection of a monument by the canton of the Silures after a decree of the local senate—ex decreto ordinis respublica civitatis Silurum (Athenaeum, Sept. 26, 1903; Archaeologia, lix. 290); other inscriptions, if less decisive, suggest that the case of the Silures was not unique in the province. Indeed, a list of the cantonal capitals, and therefore of the cantons, seems to survive mutilated in the Ravennas (ed. Parthey and Pinder, pp. 425 foll.). There we meet, besides three municipalities carefully so labelled, nine or ten towns with tribal affixes—Isca Dumnoniorum, Exeter; Venta Belgarum, Winchester; Venta Silurum, Caerwent; Corinium Dobunorum, Cirencester; Calleva Atrebatum, Silchester; Durovernum Cantiacorum, Canterbury; Viroconium Cornoviorum, Wroxeter; Ratae Coritanorum, Leicester; Venta Icenorum, Caistor-by-Norwich—and perhaps Noviomagus Regentium, Chichester. Add to these Isurium Brigantum, known otherwise by this title, and Dorchester in Dorset, and there emerges a fairly complete list of just those towns which are declared by their remains to have been the chief “country towns” of Roman Britain. The reasons why so little is heard of the cantons are, I think, plain. They were smaller, poorer, and less important than those of Gaul—as, indeed, a comparison of the town-remains shows; there was, further, no British literature to mention them; and, lastly, they quickly fell before the barbarians in the fifth century.
The town-life of Roman Britain (p. 192) was somewhat more extensive than Mommsen allows. There were four coloniae—Colchester or Camulodunum, founded about A.D. 48 (Tac. Ann. xii. 32); Lincoln, Lindum, established after the transference of the Ninth Legion to York, probably in the late first century; Gloucester or Glevum, founded A.D. 96–98 (C. I. L. vi. 3346); York or Eburacum, planted at an unknown date, on the opposite bank of the Ouse to the legionary fortress; and one municipium, Verulamium, outside St. Albans, founded before A.D. 60. There were also about a dozen “country towns,” already enumerated in the last paragraph. These were for the most part not large villages, but actual towns, furnished with temples, fora, houses, and street plans of Roman fashion, and inhabited, so far as our scanty evidence goes, by populations of which both upper and lower classes spoke and wrote Latin. At Bath, Aquae Sulis, were well-built baths, and a stately temple of the goddess of the waters. At London, Londinium (later Augusta), was a prosperous and wealthy trading-centre. But London was the only town of real size or splendour. The rest, like the cantons mentioned above, were small and unimportant as compared with similar towns elsewhere, and though it is not strictly true that Gloucester and Verulam have produced no inscriptions (p. 193; Eph. Epigr. iv. p. 195), the epigraphic yield has been scanty in every town except perhaps York.