[image ALT: Much of my site will be useless to you if you've got the images turned off!]
Bill Thayer

[image ALT: Cliccare qui per una pagina di aiuto in Italiano.]

[Link to a series of help pages]
[Link to the next level up]
[Link to my homepage]

This webpage reproduces an article in the
English Historical Review
Vol. 19 (1903), pp625‑631

The text is in the public domain:
F. J. Haverfield died in 1919.

This page has been carefully proofread
and I believe it to be free of errors.
If you find a mistake though,
please let me know!

 p625  The Last Days of Silchester

In the Roman period the corner of North Hampshire which we now call Silchester was a Romano-British town, Calleva Atrebatum. At the end of that period, early in the fifth century, this town was still occupied and inhabited, as the coins discovered in it sufficiently prove.​1 Then we lose sight of it in the general gloom. Somewhere in that dark age in which the whole Romano-British civilisation passed away Calleva also met its end. When next its site is mentioned, in Domesday and in the literature of the twelfth century, its Romano-British name has been utterly forgotten and it has ceased to be a dwelling-place of men.​2 Only its city walls must have stood then, as they stand to‑day, the enduring monument of a vanished world.

Historians have endeavoured by conjecture to pierce the obscurity which thus surrounds the last days of Calleva. Generally and very naturally they have imagined that the town was stormed and burnt by invading English, and various dates have been suggested for the catastrophe. In particular Mr. J. R. Green, arguing partly from the general course of the English  p626 conquest (as he conceived it) and partly from supposed archaeological evidence, placed the destruction of Calleva about the middle of the sixth century, probably between 552 and 568. It resisted longer (he thought) than any other British town of the Hampshire area, and its fall opened the way for a West-Saxon invasion of Surrey about 568 and of Bedfordshire in 571.​3 But the facts on which he relied are neither chronologically nor archaeologically sound, and his theory must be rejected as in part wrong and in part unproved. I propose here to summarise the evidence available for the solution of the problem and to suggest a different answer. This answer may illustrate a new feature in the process by which Romano-British gave way to English.

Literary evidence is naturally wanting. Calleva is mentioned in no Roman or Romano-British literature, except in one or two itineraries and topographical lists; Silchester is mentioned in no English treatise earlier than 1066. Nor can we fill the gap by a priori theory. The history of the English conquest of Britain in its initial stages is imperfectly known. The dates and facts assigned by the Chronicle to the fifth and sixth centuries are few; they are also much less certain than Mr. Green assumed. We possess no general evidence which is minute enough to justify an assertion that Silchester 'must have' fallen at such and such a time or under such and such circumstances.

But it may be desirable in passing to notice one medieval author. Geoffrey of Monmouth, the first writer in the twelfth century who mentions the site, makes it the scene of the coronation of Constantine, Uther's father, and of the consecration of Arthur; he also enriches it with an Arthurian bishop, Mauganius.​4 Had he any warrant for this? Historians much more recent than Aaron Thompson have thought so. He may (in their opinion) have used some authority now lost, who preserved in one fashion or another a direct record, British and not English, of Roman and of post-Roman history. It is not likely. Any such older authority would have called Calleva by its ancient name, and Geoffrey, true to his custom, would have adopted it. But he calls it by its English name of Silchester and by nothing else. Moreover his references to the place are very meagre; he does not include it among the chief cities of Britain, and he plainly knew next to nothing about it beyond its English name. Perhaps another suggestion may explain better how he came to mention it. He wrote in an antiquarian age, when Roman remains were eagerly noted and recognised as Roman in many parts of England — at Bath, for example, and Caerleon and Castor, and Carlisle and Pevensey. Silchester seems to have been noted with the rest as an ancient and therefore presumably a Roman site. It is mentioned  p627 not only by Geoffrey, but also by his contemporary Henry of Huntingdon. Henry was influenced by Geoffrey to include it in his list of British cities,​5 but he knows a little more about the spot than Geoffrey actually mentions, and he had probably heard of it independently. Here, perhaps, we touch the region of unrecorded current knowledge, and we may well believe that Geoffrey thus learnt of the ruined city walls and picked up the item as convenient to his purpose. That is why he both mentions it and yet calls it only by its English name and shows such ignorance about it. He goes on to invent a bishop for it, but that need surprise no one. Mauganius, prelate of Silchester, is kin to Boso, consul of Oxford, and Micipsa, king of Babylon, and Lucius Tiberius, procurator of the commonwealth, and several score others. They form the natural garniture of the medieval tale. We need pay no further attention to Geoffrey in our present quest.

Historical evidence often fails the historian; there remains archaeology. This, for Silchester, is a recently acquired assistance. The first serious excavation of the spot was started by Mr. Joyce in 1864. The systematic exploration began in 1890. The work is now five-sixths done: we may ask its results.

First, it appears certain that the Romano-British town came actually and completely to an end. The area within the walls is waste and uninhabited to‑day, save for a little church and farmhouse close to the eastern gate, and the excavations show that it has always remained uninhabited since the close of the Romano-British period. No trace of English dwellings or graves or other occupation has been found within it, or even in its neighbourhood. The church itself is not especially ancient, and it is natural to conclude that the site for many centuries lay practically desert.

Secondly, it appears certain that the end of Calleva did not come by fire and sword. It was not cut off and burnt by English enemies. Had it thus ended, the excavators would have discovered frequent traces of general conflagration and skeletons of townsfolk slain in fight or flight. Such have been found at Wroxeter, which we have reason to believe was stormed and destroyed; such also in numerous villas. At Silchester we meet with none of this; its end came otherwise.6

 p628  Thirdly, we have some slight evidence that the town passed through a period of decay before it ceased to exist. Some, if not all, of its gates were partly walled up — presumably because they could thus be more easily defended — and the material employed for the purpose includes worked stones from large buildings in Calleva. Such blocking of gateways has been found in other places — in the town of Caerwent (Venta Silurum), for example, and in the forts on Hadrian's Wall — and everywhere it seems to signify increasing danger or decreasing strength. The employment of worked stones from earlier buildings does not, however, necessarily imply that the town was decayed within as well as threatened from without. The Roman walls of places like Arlon and Sens are largely built with carved or worked stone torn from large and handsome structures, but it does not appear that these structures were in ruins when the walls were built. They were more probably dismantled in the hour of bitter need.​7 And at Silchester, so far as our present evidence goes, the amount of dismantlement need not have been very great.

Lastly, a strange object has been found which must be ascribed to the interval between the end of the Roman period (strictly so called) and the end of Calleva. In 1893 the excavators came upon a well or pit sunk rudely through the floor and outer wall of a corridor in a dwelling-house. In this pit, at a depth of five or six feet, lay a broken pillar bearing an ogam inscription, and below it a pewter vessel flattened out by its weight. The pit must have been dug after the corridor and its wall had fallen into ruin; the mouldings on the base of the pillar seems to be very late Roman; the occurrence of pewter harmonises with, if it does not demand, a late date. The ogam itself, according to Professor Rhys, might belong to the fifth or sixth century. It is imperfect, but in formula Celtic and sepulchral, and it might be translated 'the (grave) of Ebicatu‑s . . . son of the kin of . . .', though it seems uncertain whether it is actually an epitaph. In any case it is a Celtic and indeed Goidelic monument, with no Latin associations, since even the name Ebicatu‑s is taken to be Celtic and not the Latin Evocatus. It is the only ogam yet found in England east of Severn and Exe. It is the only important object found in the Romano-British town which can be attributed with probability to the post-Roman British period.8

 p629  Such is the evidence yielded by Calleva — slight but noteworthy. It is not perhaps discordant with the general history of the age in question. We know, generally, that the barbarians began seriously to menace the prosperity of Britain about the middle of the fourth century.​9 The assaults continued for a hundred years, and here, as elsewhere, the plunderers were superseded by immigrants invited or invading. Meanwhile Roman rule in northern Gaul had ceased, and Britain had been isolated from the empire. A Celtic revival followed. The native language, which had probably never wholly died out in the country districts, began again to spread, aided no doubt by the influences of Celtic Ireland; and with the language must have come a growth of native customs.​10 How fast the change progressed we cannot tell. It must have begun before the year 450, if a Vortigern then ruled Kent. It can be traced distinctly a century later in the pages of Gildas, though it had not then advanced so far as to obliterate in the minds of the British the notion that they belonged to the Roman empire. Later on the process was completed. Latin became merely the learned language of a Celtic-speaking people.

With these facts we can harmonise the details supplied by Calleva. The ogam falls into its place as a bit of Celtic revival. Some one in the fifth or sixth century set up this Celtic pillar at Calleva; then in the last days of the town it was thrown aside — or perhaps rather hidden out of sight and safe from insult, just like the Roman altars found in pits and wells in many Roman forts in northern Britain. The final extinction of town life also becomes intelligible, though it is not due to fire and fighting. It came rather by simple evacuation. As the English advanced, first as rough allies and then as rougher enemies, life became less and less attractive, not only in the forest region round Calleva,​11 but even within the shelter of its massive walls. At some moment or moments which we cannot fix the gateways were narrowed. At last the whole population arose and departed to some western land where the English had not yet appeared. The British at  p630 Silchester were literally exterminated: they fled 'beyond the border.'

We can parallel this 'extermination' from another part of the western empire, where the Roman civilisation perished as completely as in England. The province of Noricum Ripense, the land between Passau and Vienna, was a well-romanised district. The population was still Roman and still considered itself part of the empire in the middle of the fifth century. Its frontier forts and some of its towns had been destroyed in the course of the barbarian invasions, but it still held on in its walled settlements, and, as the empire could not help, it accepted the protection of the Rugi, on the opposite bank of the Danube. This availed little. The Rugi, like Hengist's people, were dangerous friends: other barbarians were as dangerous enemies. Life was hardly safe inside the towns, and those who ventured outside were liable to be caught up by marauders. The burden became intolerable. One town after another was abandoned. The inhabitants of Quintana retired in a body to Batava, the inhabitants of Batava soon after to Lauriacum, and in turn the occupants of Lauriacum retired to Favianae. They left their old homes desolate and uninhabited; no man dwelt in them, no trader found there any one with whom to traffic. A very few here and there declined to leave their native soil and attempted to occupy still the deserted towns: their immediate fate was death or slavery at the hands of the barbarians. At last in 488 Odoacer, who ruled Italy in the name of the eastern emperor, came to the aid of the survivors and, as the only remedy, transported them in a body from Noricum to Italy. From that day the north of Noricum ceased to be Roman in civilisation as in government.12

A kindly biographer has told us how and when the romanised town-life ended on the Danube. We have no such written evidence for Britain. But the process was plainly similar. It remains only to ask the date. It were easy to accept Mr. Green's theory of the conquest and simply substitute evacuation for destruction by fire and sword. But that theory is not, in itself, very probable. The early dates of Saxon history are untrustworthy. Geographically it is more likely that an attack on Silchester would come from the Thames valley than from the Itchen. The known facts of early English history suggest an earlier period than 560. Gildas, for instance, wrote somewhere about A.D. 540‑550,​13 and no reader of Gildas would suppose that in his time the Britons held parts of Surrey and Hampshire within forty-five miles of London.  p631 Mr. Green, I think, has overrated 'the ring of fortresses' (as he calls them) 'which enclosed the Gwent.'​14 Calleva may have ranked as a fortified place. The other two, Sorbiodunum (Sarum) and Cunetio (near Marlborough), are to the student of Roman Britain mere villages or post-stations. We cannot, with our present evidence, decide the time when the Callevan Britons lost heart and fled; now we can only perceive that at some date or other the town thus ceased to exist. To complete the tale we need other evidence, not yet discovered. It may be that when archaeologists have at last scientifically studied the chronology of English fibulae and burials the historian may learn from their conclusions another fragment of history.

F. Haverfield.

The Author's Notes:

1 The coins found at Silchester have not yet been adequately recorded in print. I have, however, been able to look through the Reading Museum collection, and its curator, Mr. Colyer, has supplied me with useful details. Coins of the late fourth century, of Honorius (gold, silver, and bronze) and of Arcadius (gold, bronze), seem fairly common at Silchester, but no later emperor is represented and no items occur (except, perhaps, minims) which can be attributed with any special probability to post-Roman British minting.

2 Geoffrey of Monmouth, VI.5, IX.1 and 15; Henry of Huntingdon, I.3 (following Geoffrey); Alfred of Beverley, I (following Henry). Compare Eulogium Historiarum, IV.170 (vol. II p148, ed. F. S. Haydon): Caersegent, Silecestre nominata, modo fere devastata.

3 Making of England, p113.

4 Geoffrey, loc. cit.

5 This I pointed out in the Athenaeum, 6 April 1901. Henry, however, adds a vague indication of where Silchester is, which is not given by Geoffrey. I may add here that, so far as I can at present judge, Geoffrey's book contains nothing to suggest that he had anywhere before him any direct British record of Roman Britain which could be called historical.

6 This was dimly recognised by Mr. Joyce (Archaeologia, XLVI.362‑3) and abundantly confirmed by the recent excavations (Victoria Hist. of Hampshire, I.371). Green, misreading Joyce, quotes 'a legionary eagle found beneath a charred wreck.' But this, according to Joyce, is débris dating from a fire long anterior to the time when the town ceased to be inhabited. For the Wroxeter evidence see J. C. Anderson, Uriconium, pp21‑2; Thomas Wright, Uriconium, pp68, 114; Guest, Origines Celticae, I.290 foll.

7 Fox (Silchester Report, 1895, p29) adduces another item which he thinks significant of decay — a fine gallery in a house (XIV.2) where masons had mixed their mortar as in a workshop and some one had lighted a fire on a costly mosaic floor. But this might occur without the town as a whole being in a state of decay.

8 For the ogam see the Silchester Report for 1893 (Archaeologia, vol. LIV), the Victoria Hist. of Hampshire, I.279, and Rhys and Brynmor Jones, The Welsh People, pp55‑65. The curious Colchester tablet, which I published in the Archaeological Journal, XLIX.215, with its concluding phrase, nepos Vepogeni Caledo, may show that the formula used on the Silchester ogam ('son of the kin of') was not unknown to Latin-speaking Britons as early as circa A.D. 235. But this may obviously be due to a stray Caledonian. In any case it is difficult to put the Silchester stone into the third century, though M. d'Arbois de Jubainville does seem to date ogams to the third century.

9 Hence the expedition of Constans (Ammian. XX.1), the first of many. In not a few Roman villas the latest coin finds date from about 350 A.D.

10 Precisely the same revival can be seen still progressing in many parts of Austria, where German used to be the language of the towns and Ruthene or Slovene or the like the language of the peasants. Perhaps I may refer in this context to what I wrote in the Edinburgh Review, April 1899, pp387‑8.

11 The name Calleva means 'the town in the forest.' The country round it was very thinly populated in Roman times, and is heavily wooded to this day.

12 Eugippii Vita Severini (ed. Mommsen, 1898). Eugippius distinctly implies that the evacuation was general and not confined to the rich. The south of Noricum, of course, retained a form of the Roman language, and was no doubt not evacuated.

13 So Mommsen and Zimmer. Compare W. H. Stevenson, Academy, 26 Oct. 1895.

14 It should be added that the term Gwent, as used by Mr. Green, has no proper authority. It seems to have been invented in recent times out of the place-name Venta, which was used of three little towns in Roman Britain. The etymology and meaning of this name Venta seem quite unknown.

[image ALT: Valid HTML 4.01.]

Page updated: 25 Mar 08