The Romanization of Roman Britain

CHAPTER VI

ROMANIZATION IN THE LOCAL GOVERNMENT AND LAND-SYSTEM

I have dealt with the language and the material civilization of the province of Britain. I pass to a third and harder question, the administrative and legal framework of local Romano-British life. Here we have to discuss the extent to which the Roman town-system of the colonia and municipium, and the Roman land-system of the villa penetrated Britain. And, first, as to the towns. Britain, we know, contained five municipalities of the privileged Italian type. The colonia of Camulodunum (Colchester) and the municipium of Verulamium (St. Albans), both in the south-east of the island, were established soon after the Claudian conquest. The colonia of Lindum (Lincoln) was probably founded in the early Flavian period (A.D. 70-80), when the Ninth Legion, hitherto at Lincoln, was probably pushed forward to York. The colonia at Glevum (Gloucester) arose in A.D. 96-98, as an inscription seems definitely to attest. Lastly, the colonia at Eburacum (York) must have grown up during the second or the early third century, under the ramparts of the legionary fortress, though separated from it by the intervening river Ouse.[1] Each of these five towns had, doubtless, its dependent ager attributus, which may have been as large as an average English county, and each provided the local government for its territory.[2] That implies a definitely Roman form of local government for a considerable area—a larger area, certainly, than received such organization in northern Gaul. Yet it accounts, on the most liberal estimate, for barely one-eighth of the civilized part of the province.

[Footnote 1: The fortress was situated on the left or east bank of the Ouse close to the present cathedral, which stands wholly within its area. Parts of the Roman walls can still be traced, especially at the so-called Multangular Tower. The municipality lay on the other (west) bank of the Ouse, near the railway station, where various mosaics indicate dwelling-houses. Its outline and plan are, however, not known. Even its situation has not been generally recognized.]

[Footnote 2: If the evidence of milestones may be pressed, the 'territory' of Eburacum extended southwards at least twenty miles to Castleford, and that of Lincoln at least fourteen miles to Littleborough (Ephemeris Epigraphica, vii. 1105=ix. 1253, where the last two lines are AVGG EB|MP XX (or XXII), and vii. 1097). The general size of these municipal 'territoria' is amply proved by Continental inscriptions.]

Of the rest, some part may have been included in the Imperial Domains, which covered wide tracts in every province and were administered for local purposes by special procurators of the Emperor. The lead-mining districts—Mendip in Somerset, the neighbourhood of Matlock in Derbyshire, the Shelve Hills west of Wroxeter, the Halkyn region in Flintshire, the moors of south-west Yorkshire—must have belonged to these Domains, and for the most part are actually attested by inscriptions on lead-pigs as Imperial property. Of other domain lands we meet one early instance at Silchester in the reign of Nero[1]—perhaps the confiscated estates of some British prince or noble—and though we have no further direct evidence, the analogy of other provinces suggests that the area increased as the years went by. Yet it is likely that in Britain, as indeed in Gaul,[2] the domain lands were comparatively small in amount. Like the municipalities, they account only for a part of the province.

[Footnote 1: Tile inscribed NERCLCAEAVGGER, Nero Claudius Caesar Augustus Germanicus (Eph. ix. 1267). It differs markedly from the ordinary tiles found at Silchester, and plainly belongs to a different period in the history of the site. Possibly the estate, or whatever it was, did not remain Imperial after Nero's downfall; compare Plutarch, Galba, 5. The Combe Down Principia (C. vii. 62), which are certainly not military, may supply another example, of about A.D. 210 (Vict. Hist. Somerset, i. 311; Eph. ix. p. 516).]

[Footnote 2: Hirschfeld in Lehmann's Beiträge zur alten Geschichte, ii. 307, 308. Much of the Gaulish domain land appears to date from confiscations in A.D. 197.]

Throughout all the rest of the British province, or at least of its civilized area, the local government was probably organized on the same cantonal system as obtained in northern Gaul. According to this system the local unit was the former territory of the tribe or canton, and the local magistrates were the chiefs or nobles of the tribe. That may appear at first sight to be a native system, wholly out of harmony with the Roman method of government by municipalities. Yet such was not its actual effect. The cantonal or tribal magistrates were classified and arranged just like the magistrates of a municipality. They even used the same titles. The cantonal civitas had its duoviri and quaestors and so forth, and its ordo or senate, precisely like any municipal colonia or municipium. So far from wearing a native aspect, this cantonal system merely became one of the influences which aided the Romanization of the country. It did not, indeed, involve, like the municipal system, the substitution of an Italian for a native institution. Instead, it permitted the complete remodelling of the native institution by the interpenetration of Italian influences.

We can discern the cantonal system at several points in Britain. But the British cantons were smaller and less wealthy than those of Gaul, and therefore they have not left their mark, either in monuments or in nomenclature, so clearly as we might desire. Many inscriptions record the working of the system in Gaul. Many modern towns—Paris, Reims, Chartres, and thirty or forty others—derive their present names from those of the ancient cantons, and not from those of the ancient towns. In Britain we find only one such inscription (Fig. 15),[1] only one town called in antiquity by a tribal name—and that a doubtful instance[2]—and no single case of a modern town-name which is derived from the name of a tribe.[3] We have, however, some curious evidence from another source. There is a late and obscure Geography of the Roman Empire which was probably written at Ravenna somewhere about A.D. 700, and which, as its author's name is lost, is generally quoted as the work of 'Ravennas'. It consists for the most part of mere lists of names, about which it adds very few details. But in the case of Britain it notes the municipal rank of the various coloniae, and it further appends tribal names to nine or ten town-names, which are thus distinguished from all other British place-names. For example, we have Venta Belgarum (Winchester), not Venta simply; Corinium Dobunorum (Cirencester), not Corinium simply. The towns thus specially marked out are just those towns which are also declared by their actual remains to have been the chief country towns of Roman Britain. This coincidence can hardly be an accident. We may infer that the towns to which the Ravennas appends tribal names were the cantonal capitals of the districts of Roman Britain, and that a list of them, perhaps mutilated and imperfect, has been preserved by some chance in this late writer. In other words, the larger part of Roman Britain was divided up into districts corresponding to the territories of the Celtic tribes; each had its capital, and presumably its magistrates and senate, as the above-mentioned inscription shows that the Silures had at Venta Silurum. We may suppose, indeed, that the district magistrates—the county council, as it would now be called—were also the magistrates of the country town. The same cantonal system, then, existed here as in northern Gaul. Only, it was weaker in Britain. It could not impose tribal names on the towns, and it went down easily when the Empire fell. In Gaul, Lutetia Parisiorum became Parisiis and is now Paris, and Nemetacum Atrebatum became Atrebatis and is now Arras. In Britain, Calleva Atrebatum (Silchester) remained Calleva, so far as we know, till it perished altogether in the fifth century.[4]

[Footnote 1: Found at Venta Silurum (Caerwent) in 1903: … leg. legi[i] Aug. proconsul(i) provinc. Narbonensis, leg. Aug. pr. pr. provi. Lugudunen(sis): ex decreto ordinis respubl(ica) civit(atis) Silurum—a monument erected by the cantonal senate of the Silures to some general of the Second legion at Isca Silurum, twelve miles from Caerwent—perhaps to Claudius Paulinus, early in third century (Athenaeum, Sept. 26, 1903; Archaeologia, lix. 120; Eph. ix. 1012). Other inscriptions mention a civis Cantius, a civitas Catuvellaunorum and the like, but their evidence is less distinct.]

[Illustration: FIG. 20. INSCRIPTION FOUND AT CAERWENT (VENTA SILURUM)
MENTIONING A DECREE OF THE SENATE OF THE CANTON OF SILURES.]

[Footnote 2: Icinos in Itin. Ant. 474. 6 may well be Venta Icenorum (Victoria Hist. of Norfolk, i. 286, 300).]

[Footnote 3: Canterbury may seem an exception. But its name comes ultimately from the Early English form of Cantium, not from the Cantii. In the south-west and in Wales, tribal names like Dumnonii (Devonshire), Demetae, Ordovices, have lingered on in one form or another, and, according to Professor Rhys, Bernicia is derivable from Brigantes. But these cases differ widely from the Gaulish instances.]

[Footnote 4: Ravennas (ed. Parthey and Pinder), pp. 425 foll. I have given a list of the towns in my Appendix to Mommsen's Provinces of the Empire (English trans., 1909), ii. 352.]

Of the smaller local organizations, little can be said. Towns existed, but many of them were the tribal capitals mentioned in the last paragraph, and these, as I have said, were doubtless ruled by the magistrates of the tribes. It is idle to guess who administered the towns that were not such capitals or who controlled the various villages scattered through the country. Nor can we pretend to know much more about the size and character of the estates which corresponded to the country-houses and farms of which remains survive. The 'villa' system of demesne farms and serfs or coloni[1] which obtained elsewhere was doubtless familiar in Britain; indeed, the Theodosian Code definitely refers to British coloni.[2] But whether it was the only rural system in Britain is beyond proof, and previous attempts to work out the problem have done little more than demonstrate the fact.[3] It is quite possible that here, or indeed in any province, other forms of estates and of land tenure may have existed beside the predominant villa.[4] The one thing needed is evidence. And in any case the net result appears fairly certain. The bulk of British local government must have been carried on through Roman municipalities, through imperial estates, and still more through tribal civitates using a Romanized constitution. The bulk of the landed estates must have conformed in their legal aspects to the 'villas' of other provinces. Whatever room there may be for survival of native customs or institutions, we have no evidence that they survived, within the Romanized area, either in great amount or in any form which contrasted with the general Roman character of the country.

[Footnote 1: The term 'villa' is generally used to denote Romano-British country-houses and farms, irrespective of their legal classification. The use is so firmly established, both in England and abroad, that it would be idle to attempt to alter it. But for clearness I have thought it better in this paper to employ the term 'villa' only where I refer to the definite 'villa' system.]

[Footnote 2: Cod. Theod. xi. 7.2.]

[Footnote 3: For instance, Mr. Seebohm (English Village Community, pp. 254 foll.) connects the suffix 'ham' with the Roman 'villa' and apparently argues that the occurrence of the suffix indicates in general the former existence of a 'villa'. But his map showing the percentage of local names ending in 'ham' in various counties disproves his view completely. For the distribution of the suffix 'ham' and the frequency of Roman country-houses and farms do not coincide. In Norfolk, for instance, 'ham' is common, but there is hardly a trace of a Roman country-house or farm in the whole county (Victoria Hist. of Norfolk, i. 294-8). Somerset, on the other hand, is crowded with Roman country-houses, and has hardly any 'hams'.]

[Footnote 4: Professor Vinogradoff, Growth of the Manor (chap. ii), argues strongly for the existence of Celtic land-tenures besides the Roman 'villa' system. 'There was room (he suggests) for all sorts of conditions, from almost exact copies of Roman municipal corporations and Italian country-houses to tribal arrangements scarcely coloured by a thin sprinkling of imperial administration' (p. 83). As will be seen, this is not improbable. But I can find no definite proof of it. If northern Gaul were better known to us, it might provide a decisive analogy. But the Gaulish evidence itself seems at present disputable.]

CHAPTER VII

CHRONOLOGY OF THE ROMANIZATION

From this consideration of the evidence available to illustrate the Romanization of Britain, I pass to the inquiry how far history helps us to trace out the chronology of the process. A few facts and probabilities emerge as guides. Intercourse between south-eastern Britain and the Roman world had already begun before the Roman conquest in A.D. 43. Latin words, as I have said above (p. 24), had begun to appear on the native British coinage, and Arretine pottery had found its way to such places as Foxton in Cambridgeshire, Alchester in Oxfordshire, and Southwark in Surrey.[1] The establishment of a municipium at Verulamium (St. Albans) sometime before A.D. 60, and probably even before A.D. 50,[2] points the same way. The peculiar status of municipium was granted in the early Empire especially to native provincial towns which had become Romanized without official Roman action or settlement of Roman soldiers or citizens, and which had, as it were, merited municipal privileges. It is quite likely that such Romanization had begun at Verulam before the Roman conquest, and formed the justification for the early grant of such privileges. Certainly the whole lowland area, as far west as Exeter and Shrewsbury, and as far north as the Humber, was conquered before Claudius died, and Romanization may have commenced in it at once.

[Footnote 1: Babington, Anc. Cambridgeshire, p. 64; E. Krüger, Westd. Korr.-Blatt, 1904, p. 181; my note, Proc. Soc. Antiq. Lond., xxi. 461 Journal of Roman Studies, i. 146. Mr. H.B. Walters has dealt with the Southwark piece in the Proceedings of the Cambridge Antiq. Society, xii. 107, but with some errors. The Alchester piece may be later than A.D. 43.]

[Footnote 2: The grant is very much more likely to have been made by Claudius than by Nero, and more likely to belong to the earlier than to the later years of Claudius.]

Thirty years later Agricola, who was obviously a better administrator than a general, openly encouraged the process. According to Tacitus, his efforts met with great success; Latin began to be spoken, the toga to be worn, temples, town halls, and private houses to be built in Roman fashion.[1] Agricola appears to have been merely carrying out the policy of his age. Certainly it is just at this period (about 75-85 A.D.) that towns like Silchester, Bath, Caerwent, seem to take definite shape,[2] and civil judges (legati iuridici) were appointed, presumably to administer the justice more frequently required by the advancing civilization.[3] In A.D. 85 it was thought safe to reduce the garrison by a legion and some auxiliaries.[4] Progress, however, was not maintained. About 115-20, and again about 155-63 and 175-80, the northern part of the province was vexed by serious risings, and the civilian area was doubtless kept somewhat in disturbance.[5] Probably it was at some point in this period that the flourishing country town of Isurium (Aldborough), fifteen miles from York, had to shield itself by a stone wall and ditch.[6]

[Footnote 1: Tac. Agr. 21, quoted in note 3 to p. 13.]

[Footnote 2: Silchester was plainly laid out in Roman fashion all at once on a definite street plan, and though some few of its houses may be older, the town as a whole seems to have taken its rise from this event. The evidence of coins implies that the development of the place began in the Flavian period (Athenaeum, Dec. 15, 1904). At Bath the earliest datable stones belong to the same time (Victoria Hist. of Somerset, vol. i, Roman Bath), the first being a fragmentary inscription of A.D. 76. At Caerwent the evidence is confined to coins and fibulae, none of which seem earlier than Vespasian or Domitian: for the coins see Clifton Antiq. Club's Proceedings, v. 170-82.]

[Footnote 3: A. von Domaszewski, Rhein. Mus., xlvi. 599; C. ix. 5533 (as completed by Domaszewski), inscription of Salvius Liberalis; C. iii. 2864=9960, inscription of Iavolenus Priscus. Both these belong to the Flavian period. Other instances are known from the second century.]

[Footnote 4: Classical Review, xviii. (1904) 458; xix. (1905) 58, withdrawal of Batavian cohorts. The withdrawal of Legio ii Adiutrix is well known.]

[Footnote 5: See my papers in Archaeologia Aeliana, xxv. (1904) 142-7, and Proceedings of Soc. of Antiq. of Scotland, xxxviii. 454.]

[Footnote 6: The town wall of Isurium, partly visible to-day in Mr. A.S. Lawson's garden, is constructed in a fashion which suggests rather the second century than the later date when most of the town walls in Britain and Gaul were probably built, the end of the third or even the fourth century. Thus, its stones show the 'diamond broaching' which occurs on the Vallum of Pius, and which must therefore have been in use during the second century.]

Peace hardly set in till the opening of the third century. It was then, I think, that country-houses and farms first became common in all parts of the civilized area. The statistics of datable objects discovered in these buildings seem conclusive on this point. Except in Kent and the south-eastern region generally, not only coins, but also pottery of the first century are infrequent, and many sites have yielded nothing earlier than about A.D. 250. Despite the ill name that attaches to the third and fourth centuries, they were perhaps for Britain, as for parts of Gaul,[1] a period of progressive prosperity. Certainly, the number of British country-houses and farms inhabited during the years A.D. 280-350 must have been very large. Prosperity culminated, perhaps, in the Constantinian Age. Then, as Eumenius tells us, skilled artisans abounded in Britain far more than in Gaul, and were fetched from the island to build public and private edifices as far south as Autun.[2] Then also, and, indeed, as late as 360, British corn was largely exported to the Rhine Valley,[3] and British cloth earned a notice in the eastern Edict of Diocletian.[4] The province at that time was a prosperous and civilized region, where Latin speech and culture might be expected to prevail widely.

[Footnote 1: Mommsen, Röm. Gesch., v. 97, 106, and Ausonius, passim.]

[Footnote 2: Eumenius, Paneg. Constantio Caesari, 21 civitas Aeduorum … plurimos quibus illae provinciae (Britain) redundabant accepit artifices, et nunc exstructione veterum domorum et refectione operum publicorum et templorum instauratione consurgit.]

[Footnote 3: Ammianus, xviii. 2,3, annona a Brittaniis sueta transferri; Zosimus, iii. 5.]

[Footnote 4: Edict. Diocl. xix. 36. Compare Eumenius, Paneg. Constantino Aug., 9 pecorum innumerabilis multitudo … onusta velleribus, and Constantio Caesari, 11 tanto laeta munere pastionum. Traces of dyeing works have been discovered at Silchester (Archaeologia, liv. 460, &c.) and of fulling in rural dwellings at Chedworth in Gloucestershire, Darenth in Kent, and Titsey in Surrey (Fox, Archaeologia, lix. 207).]

No golden age lasts long. Before 350, probably in 343, Constans had to cross the Channel and repel the Picts and other assailants.[1] After 368 such aid was more often and more urgently required. Significantly enough, the lists of coins found in some country-houses close about 350-60, while others remained occupied till about 385 or even later. The rural districts, it is plain, began then to be no longer safe; some houses were burnt by marauding bands, and some abandoned by their owners.[2] Therewith came necessarily, as in many other provinces, a decline of Roman influences and a rise of barbarism. Men took the lead who were not polished and civilized Romans of Italy or of the provinces, but warriors and captains of warrior bands. The Menapian Carausius, whatever his birthplace,[3] was the forerunner of a numerous class. Finally, the great raid of 406-7 and its sequel severed Britain from Rome. A wedge of barbarism was driven in between the two, and the central government, itself in bitter need, ceased to send officers to rule the province and to command its troops. Britain was left to itself. Yet even now it did not seek separation from Rome. All that we know supports the view of Mommsen. It was not Britain which broke loose from the Empire, but the Empire which gave up Britain.[4]

[Footnote 1: Ammianus, xx. 1. The expedition was important enough to be recorded—unless I am mistaken—on coins such as those which show victorious Constans on a galley, recrossing the Channel after his success (Cohen, 9-13, &c.). On the history of the whole period for Britain see Cambridge Medieval History, i. 378, 379.]

[Footnote 2: See, for example, the coin-finds of the country-houses at Thruxton, Abbots Ann, Clanville, Holbury, Carisbrooke, &c., in Hampshire (Victoria Hist. of Hants, i. 294 foll.). The Croydon hoard deposited about A.D. 351 (Numismatic Chronicle, 1905, p. 37) may be assigned to the same cause.]

[Footnote 3: It is hard to believe him an Irishman, though Professor Rhys supports the idea (Cambrian Archaeol. Assoc., Kerry Meeting, 1891). The one ancient authority, Aurelius Victor (xxxix. 20), describes him simply as Menapiae civis. The Gaulish Menapii were well known; the Irish Menapii were very obscure, and the brief reference can only refer to the former.]

[Footnote 4: Mommsen, Röm. Gesch., v. 177. Zosimus, vi. 5 (A.D. 408), in a puzzling passage describes Britain as revolting from Rome when Constantine was tyrant (A.D. 407-11). It is generally assumed that when Constantine failed to protect these regions, they set up for themselves, and in that troubled time such a step would be natural enough. But Zosimus, a little later on (vi. 10, A.D. 410), casually states that Honorius wrote to Britain, bidding the provincials defend themselves, so that the act of 408 cannot have been final—unless, indeed, as the context of Zosimus suggests and as Gothofredus and others have thought, the name 'Britain' is here a copyist's mistake for 'Bruttii' or some other Italian name. In any case the 'groans of the Britons' recorded by Gildas show that the island looked to Rome long after 410. On Constantine see Freeman, Western Europe in the Fifth Century, pp. 48, 148 and Bury, Life of St. Patrick, p. 329.]

Such is, in brief, the positive evidence, archaeological, linguistic, and historical, which illustrates the Romanization of Britain. The conclusions which it allows seem to be two. First, and mainly: the Empire did its work in our island as it did generally on the western continent. It Romanized the province, introducing Roman speech and thought and culture. Secondly, this Romanization was perhaps not uniform throughout all sections of the population. Within the lowlands the result was on the whole achieved. In the towns and among the upper class in the country Romanization was substantially complete—as complete as in northern Gaul, and possibly indeed even more complete. But both the lack of definite evidence and the probabilities of the case require us to admit that the peasantry may have been less thoroughly Romanized. It was covered with a superimposed layer of Roman civilization. But beneath this layer the native element may have remained potentially, if not actually, Celtic, and in the remoter districts the native speech may have lingered on, like Erse or Manx to-day, as a rival to the more fashionable Latin. How far this happened actually within the civilized lowland area we cannot tell. But we may be sure that the military region, Wales and the north, never became thoroughly Romanized, and Cornwall and western Devon also lie beyond the pale (p. 21). Here the Britons must have remained Celtic, or at least capable of a reversion to the Celtic tradition. Here, at any rate, a Celtic revival was possible.

CHAPTER VIII

THE SEQUEL, THE CELTIC REVIVAL IN THE LATER EMPIRE

So far we have considered the province of Britain as it was while it still remained in real fact a province. Let us now turn to the sequel and ask how it fits in with its antecedents. The Romanization, we find, held its own for a while. The sense of belonging to the Empire had not quite died out even in sixth-century Britain. Roman names continued to be used, not exclusively but freely enough, by Britons. Roman 'culture words' seem to occur in the later British language, and some at least of these may be traceable to the Roman occupation of the island. Roman military terms appear, if scantily. Roman inscriptions are occasionally set up. The Romanization of Britain was plainly no mere interlude, which passed without leaving a mark behind.[1] But it was crossed by two hostile forges, a Celtic revival and an English invasion.

[Footnote 1: Much of the ornamentation used by post-Roman Celtic art comes from Roman sources, in particular the interlaced or plaitwork, which has been well studied by Mr. Romilly Allen. But how far it was borrowed from Romano-British originals and how far from similar Roman-provincial work on the Continent, is not very clear. (See p. 36.)]

The Celtic revival was due to many influences. We may find one cause for it in the Celtic environment of the province. After 407 the Romanized area was cut off from Rome. Its nearest neighbours were now the less-Romanized Britons of districts like Cornwall and the foreign Celts of Ireland and the north. These were weighty influences in favour of a Celtic revival. And they were all the more potent because, in or even before the period under discussion, the opening of the fifth century, a Celtic migration seems to have set in from the Irish coasts. The details of this migration are unknown, and the few traces which survive of it are faint and not altogether intelligible. The principal movement was that of the Scotti from North Ireland into Caledonia, with the result that, once settled there, or perhaps rather in the course of settling there, they went on to pillage Roman Britain. There were also movements in the south, but apparently on a smaller scale and a more peaceful plan.[1] At a date given commonly as A.D. 265-70—though there does not seem to be any very good reason for it—the Dessi or Déisi were expelled from Meath and a part of them settled in the south-west of Wales, in the land then called Demetia. This was a region which was both thinly inhabited and imperfectly Romanized. In it fugitives from Ireland might easily find room. The settlement may have been formed, as Professor Bury suggests, with the consent of the Imperial Government and under conditions of service. But we are entirely ignorant whether these exiles from Ireland numbered tens or scores or hundreds, and this uncertainty renders speculation dangerous. If the newcomers were few and their new homes were in the remote west beyond Carmarthen (Maridunum), formal consent would hardly have been required. Other Irish immigrants probably followed. Their settlements were apparently confined to Cornwall and the south-west coast of Wales, and their influence may easily be overrated. Some, indeed, came as enemies, though perhaps rather as enemies to the Roman than to the Celtic elements in the province. Such must have been Niall of the Nine Hostages, who was killed—according to the traditional chronology—about A.D. 405 on the British coast and perhaps in the Channel itself.

[Footnote 1: Professor Rhys, Cambrian Archaeol. Assoc. Kerry Meeting, 1891, and Celtic Britain (ed. 3, 1904, p. 247), is inclined to minimize the invasions of southern Britain (Cornwall and Wales). Professor Bury (Life of St. Patrick, p. 288) tends to emphasize them; see also Zimmer, Nennius Vindicatus, pp. 84 foll., and Kuno Meyer, Cymmrodorion Transactions, 1895-6, pp. 55 foll. The decision of the question seems to depend upon whether we should regard the Goidelic elements visible in western Britain as due in part to an original Goidelic population or ascribe them wholly to Irish immigrants. At present philologists do not seem able to speak with certainty on this point. But the evidence for some amount of invasion seems adequate.]

All this must have contributed to the reintroduction of Celtic national feeling and culture. A Celtic immigrant, it may be, was the man who set up the Ogam pillar at Silchester (Fig. 21), which was discovered in the excavations of 1893.[1] The circumstances of the discovery show that this pillar belongs to the very latest period in the history of Calleva. Its inscription is Goidelic: that is, it does not belong to the ordinary Callevan population, which was presumably Brythonic. It may be best explained as the work of some western Celt who reached Silchester before its British citizens abandoned it in despair. We do not know the date of that event, though we may conjecturally put it before, and perhaps a good many years before, A.D. 500. In any case, an Ogam monument had been set up before it occurred, and the presence of such an object would seem to prove that Celtic things had made their way even into this eastern Romanized town.

[Footnote 1: Archaeologia, liv. 233, 441; Rhys and Brynmor Jones, Welsh People, pp. 45, 65; Victoria Hist. of Hampshire, i. 279; English Hist. Review, xix. 628. Whether the man who wrote was Irish or British depends on the answer to the question set forth in the preceding note. Unfortunately, we do not know when the Ogam script came first into use. Professor Rhys tells me that the Silchester example may quite conceivably belong to the fifth century.]

[Illustration: FIG. 21. OGAM INSCRIPTION FROM SILCHESTER.]

But a more powerful aid to the revival may be found in another fact—that is the destruction of the Romanized part of Britain by the invading Saxons. War, and especially defensive war against invaders, must always weaken the higher forms of any country's civilization. Here the agony was long, and the assailants cruel and powerful, and the country itself was somewhat weak. Its wealth was easily exhausted. Its towns were small. Its fortresses were not impregnable. Its leaders were divided and disloyal. Moreover, the assault fell on the very parts of Britain which were the seats of Roman culture. Even in the early years of the fourth century it had been found necessary to defend the coasts of East Anglia, Kent, and Sussex, some of the most thickly populated and highly civilized parts of Britain, against the pirates by a series of forts which extended from the Wash to Spithead, and were known as the forts of the Saxon Shore. Fifty or seventy years later the raiders, whether English seamen or Picts and Scots from Caledonia and Ireland, devastated the coasts of the province and perhaps reached even the midlands.[1] When, seventy years later still, the English came, no longer to plunder but to settle, they occupied first the Romanized area of the island. As the Romano-Britons retired from the south and east, as Silchester was evacuated in despair[2] and Bath and Wroxeter were stormed and left desolate, the very centres of Romanized life were extinguished. Not a single one remained an inhabited town. Destruction fell even on Canterbury, where the legends tell of intercourse between Briton or Saxon, and on London, where ecclesiastical writers fondly place fifth- and sixth-century bishops. Both sites lay empty and untenanted for many years. Only in the far west, at Exeter or at Caerwent, does our evidence allow us to guess at a continuing Romano-British life.

[Footnote 1: About A.D. 405 Patrick was carried off from Bannavem Taberniae. If this represents the Romano-British village on Watling Street called Bannaventa, near Daventry in Northants (Victoria Hist. i. 186), the raids must have covered all the midlands: see Engl. Hist. Review, 1895, p. 711; Zimmer, Realenc. für protestantische Theol. x. (1901), Art. 'Keltische Kirche'; Bury, Life of St. Patrick, p. 322. There are, however, too many uncertainties surrounding this question to let us derive much help from it.]

[Footnote 2: Engl. Hist. Review, xix. 625; Fox, Victoria Hist. of
Hampshire, i. 371-2.]

The same destruction came also on the population. During the long series of disasters, many of the Romanized inhabitants of the lowland regions must have perished. Many must have fallen into slavery, and may have been sold into foreign lands. The remnant, such as it was, doubtless retired to the west. But, in doing so, it exchanged the region of walled cities and civilized houses, of city life and Roman culture, for a Celtic land. No doubt it attempted to keep up its Roman fashions. The writers may well be correct who speak of two conflicting parties, Roman and Celtic, among the Britons of the sixth century. But the Celtic element triumphed. Gildas, about A.D. 540, describes a Britain confined to the west of our island, which is very largely Celtic and not Roman.[1] Had the English invaded the island from the Atlantic, we might have seen a different spectacle. The Celtic element would have perished utterly: the Roman would have survived. As it was, the attack fell on the east and south of the island—that is, on the lowlands of Britain. Safe in its western hills, the Celtic revival had full course.

[Footnote 1: How much of Britain was still British when Gildas wrote, he does not tell us. But he mentions only the extreme west (Damnonii, Demetae); his general atmosphere is Celtic, and his rhetoric contains no references to a flourishing civilization. We may conclude that the Romanized part of Britain had been lost by his time, or that, if some part was still held by the British, long war had destroyed its civilization. Unfortunately we cannot trust the traditional English chronology of the period. As to the date of Gildas, cf. W.H. Stevenson, Academy, October 26, 1895, &c.; I see no reason to put either Gildas or any part of the Epistula later than about 540.]

It is this Celtic revival which can best explain the history of Britannia minor, Brittany across the seas in the western extremity of Gaul. How far this region had been Romanized during the first four centuries seems uncertain. Towns were scarce in it, and country-houses, though not altogether infrequent or insignificant, were unevenly distributed. At some period not precisely known, perhaps in the first half or the middle of the third century, it was in open rebellion, and the commander of the Sixth Legion (at York), one Artorius Justus, was sent with a part of the British garrison to reduce it to obedience.[1] It may therefore have been, as Mommsen suggests, one of the least Romanized corners of Gaul, and in it the native idiom may have retained unusual vitality. Yet that native speech was not strong enough to live on permanently. The Celtic which is spoken to-day in Brittany is not a Gaulish but a British Celtic; it is the result of British influences. Brittany would have sooner or later become assimilated to the general Romano-Gaulish civilization, had not its Celtic elements won fresh strength from immigrant Britons. This immigration is usually described as an influx of refugees fleeing from Britain before the English advance. That, no doubt, was one side of it. But the principal immigrants, so far as we know their names, came from Devon and Cornwall,[2] and some certainly did not come as fugitives. The King Riotamus who (as Jordanes tells us) brought 12,000 Britons in A.D. 470 to aid the Roman cause in Gaul, was plainly not seeking shelter from the English.[3] We must connect him, and indeed the whole fifth-century movement of Britons into Gaul, with the Celtic revival and with the same causes that produced for instance, the Scotic invasion of Caledonia.

[Footnote 1: C. iii. 1919=Dessau 2770. The inscription must be later than (about) A.D. 200, and it somewhat resembles another inscription (C. iii. 3228) of the reign of Gallienus, which mentions milites vexill. leg. Germanicar. et Britannicin. cum auxiliis earum. Presumably it is either earlier than the Gallic Empire of 258-73, or falls between that and the revolt of Carausius in 287. The notion of O. Fiebiger (De classium Italicarum historia, in Leipziger Studien, xv. 304) that it belongs to the Aremoric revolts of the fifth century is, I think, wrong. Such an expedition from Britain at such a date is incredible.]

[Footnote 2: The attempt to find eastern British names in Brittany seems a failure. M. de la Borderie, for instance, thinks that Corisopitum (or whatever the exact form of the name is) was colonized from Corstopitum (Corbridge on the Tyne, near Hadrian's Wall). But the latter, always to some extent a military site, can hardly have sent out ordinary émigrés, while the former has hardly an historical existence at all, and may be an ancient error for civitas Coriosolitum (C. xiii (I), i. p. 491).]

[Footnote 3: Freeman (Western Europe in the Fifth Century, p. 164) suggests that a migration of Britons into Gaul had been in progress, perhaps since the days of Magnus Maximus, and that by 470 there was a regular British state on the Loire, from which Riotamus led his 12,000 men. Hodgkin (Cornwall and Brittany, Penryn, 1911) suggests that the soldiers of Maximus settled on the Loire about 388, and that Riotamus was one of their descendants. He quotes Gildas as saying that the British troops of Maximus went abroad with him and never returned. That, however, is an entirely different thing from saying that they settled in a definite part of Gaul. For this latter statement I can find no evidence, and the Celtic revival in our island seems to provide a better setting for the whole incident of Riotamus.

If Professor Bury is right (Life of Patrick, p. 354), Riotamus had a predecessor in Dathi, who is said to have gone from Ireland to Gaul about A.D. 428 to help the Romans and Aetius. Zimmer (Nennius Vind., p. 85) rejects the tale. But it fits in well with the Celtic revival.]

This destruction of Romano-British life produced a curious result which would be difficult to explain if we could not assign it to this cause. There is a marked and unmistakable gap between the Romano-British and the Later Celtic periods. However numerous may be the Latin personal names and 'culture words' in Welsh, it is beyond question that the tradition of Roman days was lost in Britain during the fifth or early sixth century. That is seen plainly in the scanty literature of the age. Gildas wrote about A.D. 540, three generations after the Saxon settlements had begun. He was a priest, well educated, and well acquainted with Latin, which he once calls nostra lingua. He was also not unfriendly to the Roman party among the Britons, and not unaware of the relation of Britain to the Empire.[1] Yet he knew substantially nothing of the history of Britain as a Roman province. He drew from some source now lost to us—possibly an ecclesiastical or semi-ecclesiastical writer—some details of the persecution of Diocletian and of the career of Magnus Maximus.[2] For the rest, his ideas of Roman history may be judged by his statement that the two Walls which defended the north of the province—the Walls of Hadrian and Pius—were built somewhere between A.D. 388 and 440. He had some tradition of the coming of the English about 450, and of the reason why they came. But his knowledge of anything previous to that event was plainly most imperfect.

[Footnote 1: Mommsen, Preface to Gildas (Mon. Germ. Hist.), pp. 9-10. Gildas is, however, rather more Celtic in tone than Mommsen seems to allow. Such a phrase as ita ut non Britannia sed Romania censeretur implies a consciousness of contrast between Briton and Roman. Freeman (Western Europe, p. 155) puts the case too strongly the other way.]

[Footnote 2: Magnus Maximus, as the opponent of Theodosius, seems to have been damned by the Church writers. Compare the phrases of Orosius, vii. 35 (Theodosius) posuit in Deo spem suam seseque adversus Maximum tyrannum sola fide maior proripuit and ineffabili iudicio Dei and Theodosius victoriam Deo procurante suscipit.]

The Historia Brittonum, compiled a century or two later, preserves even less memory of things Roman. There is some hint of a vetus traditio seniorum. But the narrative which professes to be based on it bears little relation to the actual facts; the growth of legend is perceptible, and even those details that are borrowed from literary sources like Gildas, Jerome, Prosper, betray great ignorance on the part of the borrower.[1] On the other hand, the native Celtic instinct is more definitely alive and comes into sharper contrast with the idea of Rome. Throughout, no detail occurs which enlarges our knowledge of Roman or of early post-Roman Britain. The same features recur in later writers who might be or have been supposed to have had access to British sources. Geoffrey of Monmouth—to take only the most famous—asserts that he used a Breton book which told him all manner of facts otherwise unknown. The statement is by no means improbable. But, for all that, the pages of Geoffrey contain no new fact about the first five centuries which is also true.[2] From first to last, the Celtic tradition preserves no real remnant of recollections dating from the Romano-British age. Those who might have handed down such memories had either perished in wars with the English or sunk back into the native environment of the west.[3]

[Footnote 1: The story of Vortigern and Hengist now first occurs and is obvious tradition or legend. A prince with a Celtic name may have ruled Kent in 450. There were, indeed, plenty of rulers with barbaric names in the fourth and fifth centuries of the Empire. But the tale cannot be called certain history.]

[Footnote 2: Thus, he refers to Silchester, and so good a judge as Stubbs once suggested that for this he had some authority now lost to us. Yet the mere fact that Geoffrey knows only the English name Silchester disproves this idea. Had he used a genuinely ancient authority, he would have (as elsewhere) employed the Roman name. Another explanation may be given. Geoffrey wrote in an antiquarian age, when the ruins of Roman towns were being noted. Both he and Henry of Huntingdon seem to have heard of the Silchester ruins, and both accordingly inserted the place into their pages.]

[Footnote 3: The English mediaeval chronicles have sometimes been supposed to preserve facts otherwise forgotten about Roman times. So far as I can judge, this is not the case, even with Henry of Huntingdon. Henry, in the later editions of his work, borrowed a few facts from Geoffrey of Monmouth, which are wanting in his first edition (see the All Souls MS.; the truth is obscured in the Rolls Series text). He also preserves one local tradition from Colchester: otherwise he contains nothing which need puzzle any inquirer. Giraldus Cambrensis, when at Rome, saw some manuscript which contained a list of the five provinces of fourth-century Britain—otherwise unknown throughout the Middle Ages (Archaeol. Oxoniensis, 1894, p. 224).]

But we are moving in a dim land of doubts and shadows. He who wanders here, wanders at his peril, for certainties are few, and that which at one moment seems a fact, is only too likely, as the quest advances, to prove a phantom. It is, too, a borderland, and its explorers need to know something of the regions on both sides of the frontier. I make no claim to that double knowledge. I have merely tried, using such evidence as I can, to sketch the character of one region, that of the Romano-British civilization.

INDEX

Aldborough (Isurium), 56.

Arretine pottery, 15.

Avot, 27.

Bannavem Taberniae, 63.

Bath, 42, 56.

Brittany, migration to, 65.

Bury, Prof., 66.

Caerwent, 50, 56.

Canterbury, deserted after the Roman period, 64.

Cantonal system in Roman Britain, 50.

Carausius, birthplace, 58.

Castor pottery, 40.

Celtic language in Gaul and Britain, 14, 26.

Celto-Roman temples, 30; houses, 31.

Christianity as affecting language, 15.

Cloth, British, 57.

Corbridge Lion, 43.

Corn, exported from Britain, 57.

Cornwall, Roman remains in, 22.

Demetrius of Tarsus in Britain, 28.

Devonshire, Roman remains in, 22.

Din Lligwy, village at, 37.

Frilford, Romano-British house at, 32.

Gaulish kingdom of A.D. 258, 17.

Geoffrey of Monmouth, 68.

Gildas, 64, 66.

Glastonbury village, 45.

Gorgon at Bath, 41.

Henry of Huntingdon, All Souls MS. of, 68 note.

Hesione and Hercules, 41.

Historia Brittonum, 67.

Houses in Roman Britain, their relation to houses in Gaul and Italy, 31, 34.

Icinos, 51.

Imperial domains in Britain, 49

Kent, origin of name, 29.

Late Celtic art, 39.

Latin, used in the provinces, 11, 14; in Britain, 24.

London, deserted after the Roman period, 64.

Magnus Maximus, army of, 66.

New Forest pottery, 39.

Northleigh, Romano-British house at, 33.

Ogam at Silchester, 62.

Pergamene style in Roman Gaul, 44.

Pitt-Rivers, 45.

Plaxtol, inscribed tile from, 27.

Punic language in Africa, 14.

Ravenna geographer, 52.

Riotamus, 65.

Samian pottery, 16, 44 note.

Seebohm, 53.

Silchester—
  Ancient names of, 53, 68.
  Date of development, 56.
  Dyeing works, 57.
  Houses of, 34.
  Imperial domains at, 49.
  Inscribed tiles from, 25.
  Latin used in, 24.
  Ogam, 62.
  Street plan of, 34, 56.
  Temples of, 31.
  Abandoned, 63.

Temples in Britain and north Gaul, 30.

Towns in Roman Britain, 48 foll.

Villages in Roman Britain, 37, 45.

Vinogradoff, Prof., 36, 46.

Warwickshire, Roman, 22.

York, Roman remains at, 48.