The voice of archaeology in Britain and beyond

Cover of British Archaeology 87

Issue 87

March/April 2006



Unique Roman tombstone may leave UK

"No synthesis of British prehistory is right"

Neolithic road is unique

Bottle message is dry

The dead make way for iron age warrior

In Brief


When Rome left Britain: the Bosnian perspective
Buckles and Bosnia - Stuart Laycock has a dark vision of early historic Britain.

Telling the story of the people who made London
Archaeology in London: Peter Rowsome reviews a year of new publications.

Boudica: a queen in search of a husband
Finding Prasutagus: Amanda Chadburn deciphers iron age coins.

on the web

Recommended websites


Views and responses

CBA news

Headlines from the CBA office.


ISSN 1357-4442

Editor Mike Pitts


Britannia: the threat within

Why did British society collapse after 350 years of Roman rule? Chasing an old mystery, Stuart Laycock finds answers in ditches, belt buckles and modern Bosnia.

In the 6th century Gildas blamed the Saxons. History indicates that Germanic immigrants came in small, separate groups, which the archaeological evidence does not contradict. These arrivals could cause severe disruption near the coast, but it is hard to see how they might precipitate structural collapse across 5th century ad Britain. The Anglo-Saxon chronicle says Aelle founds the territory of the south Saxons with just three keels, perhaps 60 men. By contrast, Verulamium (later St Albans) is reckoned to have had 5,000 or more inhabitants. The 4th–5th century cemetery at Poundbury, outside Dorchester, Dorset contained almost 4,000 graves; estimates for the entire population of Roman Britain range from 1.5m to 3.6m at its peak.

Romanised society in Gaul and Spain survived relatively intact. The new "owners", Germanic rulers, took over wholesale the structure of the Roman state, rather than build kingdoms from scratch. Compared to Britain, disruption was minor.

Here there seem to have been 10–20 years early in the 5th century between economic collapse and the Saxons' arrival. In some areas, particularly the west, the gap is much longer. The causes of the decline of Roman British society must be sought elsewhere.

There has been much debate over the function of linear ditch-and-bank earthworks scattered across the country. They can be seen as the ancient equivalent of barbed wire, not manned constantly like a parapet, but perhaps patrolled, and marking a territorial claim that hindered traffic, particularly wheeled or horse-borne. A modern parallel might be the British Army's cratering of unguarded border roads between Northern Ireland and the Republic during the Troubles.

Archaeology throws little light on their origins. Some dykes are pre-Roman, but a significant number date from towards the end of, or after Roman rule. In the past Offa's Dyke, the one historically documented post-Roman linear earthwork, has led other examples to be attributed to the Saxons. Recently, however, dykes have instead been linked to late Roman or post-Roman British societies, such as Wat's Dyke, near Offa's Dyke at its north end, and Wansdyke, which runs, with interruptions, from near Bristol to Marlborough, Wilts.

Some features of the post-Roman earthworks suggest they descend from a plentiful pre-Roman tradition. Ian Burrow has shown that Wansdyke incorporates hillforts in its layout, something alien to Saxon defences. Offa's Dyke has been named after its builder: others, as John Morris pointed out, tend to carry generic or legendary names, implying the Saxons knew little of their real origins.

Even where a Saxon date is still accepted, a dyke may be Britishinspired. Devil's Dyke (Cambs) is thought to be Saxon, but the nearby and similarly aligned Fleam Dyke, with a different V-shaped profile (often seen on Roman ditches) and a radiocarbon date of AD330–510 is more likely to be Roman or post-Roman British. Recent research suggests that Wat's Dyke, close to Offa's, is not, as previously thought, 8th century, but some 300 years earlier.

So if many post-Roman dykes are British or British-inspired, what is their function? It has been suggested that some were to hold back advancing Saxons. However, as Morris points out, if Saxons had known the British built them, they could have used names linked to Wealh (foreigner, the origin of the modern name Wales). Some of the dykes face away from the Saxon advance. Fleam Dyke, for instance, bars the Icknield Way to the south-west. Grim's Bank, cutting the Roman road north of Silchester (Hants), halts a potential Saxon advance south from Dorchester (Oxon). Yet the same ditch stretches west to encompass the road to Cirencester (Glos) which, according to the Anglo-Saxon chronicle, did not fall to the Saxons until 577.

So perhaps the ditches were built to counter an internal threat, rather than an external one.

At some stage after the end of Roman rule, Britain fragments. It is unlikely potters and traders ceased work without very compelling reasons. It has been suggested that a form of Roman provincial structure survived, attachment to Rome discouraging an independent coinage. Why such respect for imperial power after 410, when the British had spent much of the previous three decades in armed rebellion against it? Power is not exercised in a vacuum. When the carrot (imperial funding) and the stick (the Roman army) ceased to be relevant, power would naturally devolve, in the absence of any strong, central British power, to local authorities.

The process is clearest in the west, where by Gildas's time there are independent states like Dumnonia and Dyfed. But fragmentation highly likely occurs during economic collapse a century before. As it was no longer possible to trade goods across the country, how could a single political structure have been maintained? If stories about Vortigern have any credibility, then something may have survived above the local (or perhaps been reimposed); but even here there are indications of civil strife, with the battle of Guoloph (possibly in Hampshire), and Kent described as having had a separate British king (even if Vortigern has some kind of overlordship).

Empires or other multinational political entities that collapse, rather than being taken over, are comparatively rare. Yet in such cases extensive friction and conflict between the emerging fragments are recurring features.

Think of Alexander the Great's empire after his death, the Ottoman Balkans empire in the 19th century, Tsarist Russia at the beginning of the 20th and even former Yugoslavia at the end. Different emerging power bases vying for exclusive control of populations and natural resources, create insecurity, hostility (from closed borders to full scale war) and economic decline for a substantial period. As in former Yugoslavia, the process is often exacerbated by old conflicts and divisions previously checked by a central power. Explosive fragmentation can be avoided only by control, communication and, often, external pressure of the sort highly unlikely to have applied in 5th century Britain: it is arguable that without 130,000 US troops, Iraq, assembled by Britain after World War One from three separate Ottoman provinces – Basra, Baghdad and Mosul – would be the latest example of this process.

Gildas concentrates on the coming Saxon invasion, but on several occasions he refers to long term political instability and the regular overthrow of British rulers after 410. One passage may well describe the effects of political fragmentation:
For they took to looting from each other, since there was only a very small stock of food to give nourishment to the desperate people; and the calamities from abroad were made worse by internal conflict, and consequently, the whole area became almost devoid of food, except for what hunters could find.

A landscape divided by linear earthworks aimed apparently more at neighbours than an external threat, would sit well with such a scenario. A map of linear minefields would reveal confrontation lines in former Yugoslavia: maybe a map of linear earthworks can do the same for 5th century Britain.

If most post-Roman ditches represent confrontation lines between opposing British political units, then they reveal the scale of fragmentation and damage to the economy. Foss Ditch cuts off much of Norfolk to the west. Fleam Dyke cuts the Icknield Way leading into East Anglia. Faesten Ditch cuts the London–Kent road. Grim's Dyke (possibly repaired in the 4th century) cuts London off from countryside to its north. Grim's Bank cuts off Silchester from Dorchester in the north and Cirencester in the west. Bokerley Dyke (Dorset) divides Salisbury and Dorchester, and Wansdyke Cirencester and territory to the south. Still to consider are other possible post-Roman dykes such as Nico Ditch outside Manchester, Scot's Dyke near Catterick (n Yorks), King Lud's Bank in Lincolnshire, or natural barriers such as rivers or existing pre-Roman dykes.

Extensive disruption to the British economy could be achieved merely by such dykes preventing trade and contact between neighbouring political areas. Fragmentation and war reduced significant parts of former Yugoslavia from a modern integrated European economy to little more than subsistence farming within a few months. Without large-scale warfare, the process might have taken up to a decade, but the same is likely to have been the fate of post-Roman Britain. While archaeology cannot yet explain this process, there are fascinating hints.

One curious aspect of the time has been the apparent militarisation of the area where the Thames cuts through the Chilterns, far from areas most vulnerable to foreign attack. The small town of Mildenhall (Wilts) is heavily fortified in the late 4th century. A ballista is mounted on defences at Dorchester on Thames, and in the cemetery a body has been found with full Roman military belt fittings. There are significant numbers of late Roman hoards and military buckles in the region. These might represent general insecurity across Britain, but the regional level of militarisation seems unusual.

George Lambrick's work offers a possible explanation. He has convincingly linked the high number of pre-Roman defensive works to the fact that the territories of the Dobunni, Atrebates and Catuvellauni all meet here. Late 4th century militarisation may represent a recurrence of tribal tensions.

Pre-Roman inter-tribal rivalry seems to have been particularly extensive in areas of south-east Britain where British decline appears fastest and most complete in the early 5th century. The Catuvellauni appear to have occupied the territory of the Trinovantes and parts of the Atrebates (including Silchester) and possibly the Cantiaci and Corieltauvi.

This pattern of inter-tribal war may have persisted after the Roman invasion. Boudica and her Iceni attacked not just Roman targets, but Catuvellaunian as well, including the capital at Verulamium. It is notoriously hard to identify conflicts from archaeological evidence alone. Apart from the famous discoveries at Maiden Castle and Hod Hill (Dorset), there is little direct evidence of the death and destruction caused by three Roman invasions. Not every armed confrontation leaves burned buildings or a mass grave: nor is every headless skeleton or fire a sign of war. Nevertheless, fires do occur across Trinovantes territory in the late 2nd century ad, in the area of a main Roman road running south from Iceni territory. It is a very early date for foreign raiders. Tribal conflict seems a reasonable explanation.

Such activity could be expected to leave disputed tribal frontiers and a hostile legacy for many generations. Consider our own relations with the French. It is almost 200 years since we last fought, yet there is a persistent level of mistrust unhealed by alliances and extensive commercial and cultural ties.

Serbian oral epics kept alive the story of their defeat by the Turks in the 14th century, feeding the revival of Serbian nationalism in the 19th. It is pure speculation, but epics could easily have played a similar role in Roman Britain. Early Irish poetry tells of conflict, as does early Saxon (the Finnsburg episode in Beowulf and the Finnsburh fragment), and early British poetry (y Gododdin) tells of the defeat at Catraeth (Catterick). Pre-Roman epic poetry could have kept bitter memories fresh during the Roman period.

To suggest that tribal loyalties reasserted themselves after 350 years of Roman rule is controversial. Yet there is supportive evidence. The Cantiaci gave their name to the Saxon kingdom of Kent. Fleam Dyke cuts the Icknield Way close to, perhaps on, the eastern edge of the Catuvellauni. In the west, Ken Dark has identified the tribal origins of kingdoms such as Dumnonia and Dyfed. Long-lasting tribal and civitas boundaries have been demonstrated in France, where it is accepted that medieval dioceses reflect civitas boundaries. Arecent study has shown that the civitasAtrebatum in northern Gaul is almost identical to the bishopric of Arras.

For more evidence in the centre and east of Britain, we can look to belt buckles. Forty five years ago, Sonia Hawkes and Gerald Dunning (in Medieval Archaeology vol 5) associated a category of buckles and belt fittings with late Roman and post-Roman military activity.

Some have disputed this, as such items appear on civilian sites. In continental Europe, however, they occur almost exclusively along the imperial borders. The constant recurrence of basic designs, despite manufacture in a variety of locations, suggests it is still safe to ascribe a military or paramilitary role to the Hawkes and Dunning fittings. Any militia drawn from the civilian population or billeted on them, would leave buckles in civilian locations.

The horsehead buckles (Hawkes & Dunning type 1b) have proved particularly controversial. They are small (20–30mm across) and the few found in Anglo-Saxon graves are usually with women: it has been suggested these may be women's buckles.

However, the evidence still favours some military or political significance. There seems little evidence for non-Germanic women wearing belts even in the late Roman empire. The horsehead buckles evolve from dolphin buckles, with their military distribution in continental Europe. Stylistically they appear later than the dolphin buckles; if they were a female version of the latter one might expect them to be contemporary. Horsehead buckles seem to be made across Britain, yet retain a uniform size and style suggesting an official design. Size need not be an issue. Some waist buckles found in 5th century male graves are small. Horsehead buckles may be baldric buckles for sword belts worn over the shoulder: baldrics shown on Trajan's column in Rome are of a similar breadth to the horsehead buckles. The Vergilius Romanus, which Ken Dark has persuasively identified as a 5th century British manuscript, shows soldiers with baldrics of the same size.

I have been reviewing the Hawkes and Dunning buckles and belt fittings (see They may give valuable insights into the process of British fragmentation in the late 4th and early 5th centuries. Some of the emerging political units and their military forces probably do have tribal origins.

In the east there is evidence of former tribal groupings developing individual buckle styles. In Corieltauvi territory a "protrusion aesthetic" develops, with extremely exaggerated lips on the dolphins, projecting dolphin fins, and, most noticeably, birds perched on the dolphins or buckle plates. In Iceni territory are buckles with a central human head at the top of the buckle loop. In Catuvellauni territory, buckles and belt fittings appear decorated with lines of small dots.

However, the picture is not a simple reproduction of the tribal map of pre-Roman Britain. Again the ditches seem to play a role.

In the west horsehead Ib buckles are common in the Dobunni territory north of Wansdyke, but rare to the south (formerly Dobunni, but possibly under separate leadership in the immediate pre-Roman period and then taken by the Romans for the Belgae civitas), suggesting a separation between the areas each side. Could the hoards containing silver and/or gold just to the south of Wansdyke, and its design to defend against a threat from the north, suggest resistance by the Belgae civitas against an attempt by the northern Dobunni to reunify the territory?

A similar link between horsehead buckles and ditches may occur in northern East Anglia. While the buckles appear on the western fringes of former Iceni territory, they are almost absent beyond the late Roman or post-Roman Foss Ditch in the heart of Norfolk.

Further south three tribal areas, the Iceni, Trinovantes and Catuvellauni, meet near Cambridge. As near Dorchester on Thames, the archaeological evidence suggests this may have been an area of tension at the end of Roman rule. Many late Roman hoards containing silver and/or gold have been found here. Fleam Dyke was built between 330 and 510 to defend against a threat from the Catuvellauni direction. There are a few pieces of metalwork with Catuvellauni dots in the western parts of Iceni territory. The spread of pre-Roman Catuvellauni currency in this area is widely linked to Cunobelin's expansionist policies. Perhaps the spread of the dots indicates something similar in late Roman and post-Roman times.

Much further work is needed and more buckles need to be published. Already, however, we may be seeing evidence for the catastrophic fragmentation of Roman Britain – the Bosnian option. A mixture of old and new rivalries carved up British society, destroying trade and communication most notably in the east. The main beneficiaries were new Germanic immigrants. These arrived either with the acquiescence of British political units needing fresh support, or they took advantage of the British: who were simply too weak and scattered to resist.

Thanks for assistance with this article and with the buckle project to Kevin Leahy, Brian Cavill, Chris Marshall and Robert Vermaat (

CBA web:

British Archaeology

Jan/Feb 2005
Mar/Apr 2005
May/Jun 2005
Jul/Aug 2005
Sep/Oct 2005
Nov/Dec 2005
Jan/Feb 2006
Mar/Apr 2006
May/Jun 2006
Jul/Aug 2006
Sep/Oct 2006
Nov/Dec 2006
Jan/Feb 2007
Mar/Apr 2007
May/Jun 2007
Jul/Aug 2007
Sep/Oct 2007
Nov/Dec 2007
Jan/Feb 2008
Mar/Apr 2008
May/Jun 2008
Jul/Aug 2008
Sep/Oct 2008
Nov/Dec 2008
Jan/Feb 2009
Mar/Apr 2009
May/Jun 2009
Jul/Aug 2009
Sep/Oct 2009
Nov/Dec 2009
Jan/Feb 2010
Mar/Apr 2010
May/Jun 2010
Jul/Aug 2010
Sep/Oct 2010
Nov/Dec 2010
Jan/Feb 2011
Mar/Apr 2011
May/Jun 2011
Jul/Aug 2011
Sep/Oct 2011
Nov/Dec 2011
Jan/Feb 2012
Mar/Apr 2012

CBA Briefing

Courses & lectures
CBA Network
Grants & awards

CBA homepage