Why a problem? There are all sorts of answers. It may be to make the subject more interesting, to introduce the air of mystery which attracts readers. Or it may be those academics wanting to complicate what ought to be a nice simple subject, just to make work for themselves, and justify their existence. Or it may be that conjuring a word out of the air and applying it to a section of archaeology really does cause problems.

Is it a matter of Us and Them? Pleasant, sensible people who are interested in the past as they find it on a local dig, or as it is relayed to them in books and articles, and nit-picking professionals who would be out of a job if they dropped their complicated and theoretical defences. Unfortunately the lines are not drawn in quite that way. There are many paid diggers in the trench, many amateurs, and even a number of university archaeologists who dislike theory and prefer simple names, like Dobunni, for simple things. And there are amateurs and professionals who think that archaeology is really a matter of digging up material, doing fieldwork, and thinking about what they have found without dragging in ancient names and texts which never quite seem to fit.

"But look, all the sites up on the hills have pot with a soapy feel, and all the sites down here in the vale have sandy pots".
"Agreed, but is it because they had different kings or bought from different potters?"

Did that exchange take place in a Department of Archaeology, or a Unit, or a Local Society? It could perfectly well have been any of these or none. But what is being argued over is important. Are the Dobunni up in the hills the soapy-pot-people? If so who are the sandy-pot-people?

Now, there is a really dangerous question. Why are you happier (if you are) naming the soapy-pot-people Dobunni, and why do you think you would feel happier if you could name the sandy-pot-people?

"Well, we know who they are then".
"Do you? Why is the label Dobunni a value-adding thing?"
"Well it tells us that they belong to the people ruled from Cirencester. And the Romans knew about them. And the historians talk about them. It is like knowing their web-site, we can look them up".

But a few wires have got crossed here. The sites and pottery we are talking about are Iron Age so whatever Roman writers said, they said it a few hundred years too late. No, you cannot assume that what was written in AD 150 applies to 150 BC. One of the most obvious things about pottery fashions and settlements patterns in prehistory is that they change. People move. The concept of territory may well be important, but the concept of home in the sense of this few tens of square feet of land with a permanent dwelling on it just does not work in the context of earth fast wooden houses and basic farming. Posts in post-holes rot quite quickly, and land can quickly become worn out.

Is it better or worse in the Roman period? Well, in the second century the geographer Ptolemy wrote of the tribe of the Dobuni or Dobunni, and gave them the tribal centre of Corinium. Diocassius writing about the Roman invasion talked about a part of the tribe of the Bodunni who surrendered to Claudius. The sixth century writer from Ravenna mentioned Cironium Dobunnorum. There is a military diploma of AD 105 of Lucco son of Trenus, a Dobunnian who was serving with the First Britannic Cohort in Pannonia, modern West Hungary. A milestone of the emperor Numerian (AD 283-4) from Kenchester, Herefordshire is inscribed RPC (or G) D (first seen as B). If, that is IF, it reads RPCD you could expand it to read Res Publica Civitatis Dobunnorum - the administrative unit of the civitas of the Dobunni. If it reads RPG B it could be expanded Res Publica Glevensis B. Finally a mid-sixth century tombstone from Tavistock in Devon reads DOBUNNI FABRI FILII ENABARRI. Which should mean something like 'Of Dobunnus Fabrus son of Enabarrus'. And that is your lot. Or at least as much as A L F Rivet and Colin Smith could find (Rivet and Smith 1979).

You might find this comforting. For instance you could plot on the map the other tribal capitals and then draw a line round each of them to fill up the map, and so allocate a tract of countryside to each tribe. If the map of Britain filled an A4 page the lines would probably not be more than 20 or 30 miles out in any particular case. But you have to remember that the lines themselves would be many fields thick. The problem here is that of tidiness. This seems to me to be a very recent obsession. It is almost as if the people in the past delighted in exceptions and oddities and did their best to preserve them. So the detached portions of parishes have had to be tidied up. The bits of Gloucestershire swimming in a sea of Wiltshire or Worcestershire have had to be returned. And that one field on the other side of the river in the middle of Old Jack's land is often landholder Ted's pride and joy. Nice smooth lines on A4 maps may be comforting, but they are nonsense the moment you move into the field or ask about a particular site. And in the archaeological past I think they are an illusion.

Where then does our problem come from? It is the problem of not having an adapter. Most people know the disruption caused by being away from home with electrical gadgets like razors and hair-dryers but without an adapter. He cannot go down to breakfast until he has shaved, she will not appear in public until her hair is dry. Life can almost come to a standstill. The reception desk of the hotel can be relied upon to take one look at the plug on your gadget and say, after a pause, 'No, I have never seen one like that before'. The electricity is there inside the socket, your gadget is in prime working order, but how can the two be got to work together?

This is the same problem with Dobunni and pots. I am not excited by the amount of information that the label Dobunni carries, but at least it is a label. But to what? What is the adapter that connects it to the material culture of an area. Is it soapy-pot or sandy-pot, or both, or neither? And if this particular adapter of pottery is used, what happens when the Roman arrive and both type of pots go out of production. Were the Dobunni disbanded? Clearly not. How then can yet another adapter be fitted? We would have to do a survey of all sites with their pottery, and then a survey of early Roman sites and try to find a Roman type of pottery which only occurred on the defined-by-pottery-Dobunnic earlier sites. This is almost certain to be impossible because Roman pottery seems to be distributed by market area rather than tribe. This makes sense. Gloucester is supplied from kilns in the vale and from goods brought up the river, both of which are sources foreign to Cirencester with the 300m scarp edge of the Cotswolds in the way.

This is getting too complicated. We ought to go back to the question of whether the Dobunni can be studied. The answer of course is yes. The word Dobunni is a label used in written documents to describe a particular set of people who the author had otherwise never heard of, and had certainly never seen. The proper study of the Dobunni is therefore the study of the occasions on which the word was used, what the author might have meant by it, and how the word has been handed down in the manuscript tradition. No one has yet suggested that the name has been handed on into any modern context, though I do remember during the digging of Bagendon Mrs Clifford being very excited when a local farmer told her that the field in which she was digging was known as Bonny Meadow (Dobunni - Dubonni?). The name first appears on the 1792 enclosure map and has not been followed up. Whether it is worth doing so I do not know, but I strongly suspect that Bonny Meadows might be spread far wider than any area which could be called Dobunnic, or even Dubonnic.

Since the coinage appears to be a tempting adapter it might be worth considering it a bit further. The British coinage starts by inspiration from across the Channel and gets going in the 1st century BC. Early issues all seem to me to radiate out from the Gallo-Belgic issues, and then gradually differentiate into regional clumps. To what extent is there a regional grouping in the Cotswold area before the issues labelled Dobunnic start up? I ask this because the maps that are usually drawn of 'Tribal' issues all seem to belong to a century between 50 BC and AD 50, or less. They form a snap-shot of people and politics at a certain stage, and even within that century there are re-arrangements which show shifts of people or allegiances or influences. Is there an earlier regional grouping, or could we even expect one? What are the stages by which we assume coinage as an idea was taken up in different parts of the country?

Given that Cotswold coins were produced by an authority that had access to gold and silver and the technique of minting coins, and so quite a powerful entity which might be a chief or magistrate or king, what does the distribution of those coins mean? Let us call the coins Dobunnic because we assume they were struck by a tribal authority; is their use restricted to members of the tribe, or can they circulate more widely. If they can then they might as well circulate and be used in some form right up to the point where they run into an area where other coins circulate. If they are gold or silver then circulate might not be the right word because they have an intrinsic value and even if not accepted in transactions they will make nice bracelets or brooches.

On the other hand it might be safer to stick with coins you know, because some tribes have a nasty habit of manipulating the gold content, together with silver and copper, to produce new coins which look the same, but have far less gold in them. And the Cotswold locals have a habit of silver plating copper instead of producing pure silver. So there might be no compulsion to keep the coins in the tribal area, it may simply be that outside the area people look at them doubtfully and prefer not to risk taking them. The big difficulty is that while it may be reasonable to say these particular coins were produced by the Dobunni, it is a much more dangerous step to say that they were used and lost only by Dobunni. And that is the assumption that is made when the coins' findspots are plotted on a map and the area is labelled 'Territory of the Dobunni'.

And what happens when different sorts of material evidence collide? Should Dobunnic (defined by the coin area) pottery be uniform? That is probably too much to ask. The coin influence is almost certainly much wider than the pottery transport area. In other words coins can travel a long way without any difficulty whereas pottery travels over land only with difficulty. So there might be Dobunnic pottery a, b, and c. That causes no trouble. The trouble might come if some popular pottery were distributed to the edge of the Dobunni and the edge of the Catuvellauni - say around modern Oxford. If this were the case we could only plot the boundary from sites yielding coins

Like so much of archaeology the coin definition is perfect so long as it is virtually circular: the Dobunni are the users of this type of coins because the coins occur in the area where we expect the Dobunni to be. This site was inhabited by Dobunni because it is in the area in which Dobunnic coins were lost. That sounds hardly worth saying, yet it is tight enough to stand up. It is when we start using or interpreting such a definition that the trouble starts.

Which has not mentioned archaeology; for the simple reason that you cannot study historical concepts through excavated material unless you have the adapter. What then is the Iron Age and early Roman material of the area to which an adapter could be fitted if the right one were ever found? Because this involves some hard work I leave that to Tom Moore in the second part of this article.

Part 2: Studying the Dobunni, the Roman texts and the Iron Age archaeology - Tom Moore

Introduction

It has often been assumed that we know who the Dobunni were and where they lived. Studying the archaeology of the later Iron Age, however, reveals a picture of complex and varied societies undergoing considerable change in this period. Archaeological study is beginning to suggest that conventional views of the Dobunni may mask a far more complex and sophisticated reality than has been considered. This paper will discuss the orthodox view of the Dobunni and outline why this model of 'Dobunni' society may be flawed. It will then discuss recent developments in Iron Age archaeology of the 'Dobunni' region and suggest a different view of what society may have been like. It will suggest that present views of the Dobunni rely too heavily on text and coin evidence and that they try to place archaeological evidence into a straitjacket without examining it on its own merits.

How the Dobunni have been defined in the past

Despite the common use of the term 'Dobunni' in the literature on the Roman and Iron Age periods they have rarely been defined in archaeological terms. The current view of the Dobunni relies essentially on references in later Roman texts dating from between the 2nd and 4th century AD. Richard Reece has summarised these and demonstrated that they all come from the Roman period but appear to be referring to the political situation at around the time of the conquest. From these later sources a number of assumptions about the Dobunni have been made. These include assumptions that Ptolemy's (2nd century AD) reference to Corinium as the capital of the Dobunni refers to the pre-Roman situation as well as that after the conquest and that Hawkes (Clifford 1961) was correct in seeing Dio's reference (3rd century AD) to the 'Bodunni' as a misspelling for Dobunni thus implying that the tribe was divided, with part subject to the Catuvellauni.

In order to define the Dobunni more closely Allen (1961) claimed that Iron Age coinage found in the Gloucestershire area could be equated with this tribal group and therefore termed this coinage 'Dobunnic'. Plots of this coinage have subsequently been used by a number of authors to draw the tribal boundaries of the Dobunni (eg Selwood 1984, Cunliffe 1991, Van Arsdell 1994). In addition, the coinage also appeared to provide the names of individuals, which Hawkes (1961) attempted to mould into a political history for the region.

In the 1950s, excavations at Bagendon (Clifford 1961) appeared to confirm Ptolemy's text, by providing an immediate pre-Roman centre close to the Roman town, suggesting that the 'Dobunni' were a centralised tribe with a capital at 'Corinion'. In the 1980s this view of the Dobunni was moulded into the core-periphery model with the Dobunni trading with more developed groups to the east and indirectly with the Romans (Cunliffe 1984). This model saw the middle Iron Age as being a period when an inherently warlike society coalesced into ever more powerful groups based at important hillforts. The stimuli of external trade from the south east and the Roman Empire then provided the impetus for the most powerful groups to gain power by controlling trade with the more developed tribes in the east (Cunliffe 1984). This view regarded Bagendon and other 'oppida' as trade centres through which the kings based their power by controlling trade. It was assumed by some that the Roman civitas fossilised the late Iron Age tribal area replacing the capital at Bagendon with Corinium.

There are a number of problems with this seemingly straightforward view of the Dobunni. The rest of the paper will discuss some of the problems with this view and outline other ways of looking at the Dobunni of the later Iron Age.

The Roman texts

The traditional view of the Dobunni places direct emphasis on the Roman texts as evidence and assumes that the Roman civitas reflected earlier tribal boundaries. As Richard Reece has indicated all the Roman texts discussed are from far later time points than the period we are discussing and even if they do relate to the last decades of the 1st century AD there is no reason why they should refer to the later centuries BC. Basing a discussion of the later Iron Age on these texts tends to view the Dobunni of the later Iron Age backwards from the Roman period using the Roman's view of the world to illuminate the archaeology of the Iron Age. This has the danger of seeing the society before the Roman conquest as static and unchanging. Subsequent analysis of settlement evidence of the later Iron Age (discussed below) suggests this was not the case.

It is possible that the Romans may have categorised the area in terms they could understand. There is evidence from both the archaeology and texts, such as Caesar's Gallic Wars, that Iron Age societies were varied and that there was not a uniform social model based on a single capital and king. There is evidence for example that society in the Durotrigian area in Dorset had a dispersed power structure, with hillforts of the early period still seemingly in use. Despite these differences the Romans treated these groups in a similar way by dividing areas into convenient units and creating one single civitas capital. This may have borne little relationship to preceding systems of power. It is important to note, therefore, that even if the Romans believed, or claimed, that pre-Roman society was organised in a certain way it may in fact have been radically different. Consequently, we cannot assume that circumstances described by Ptolemy accurately reflect the pre-Roman situation. It may have been that only one small group called themselves the Dobunni, and the Romans assumed or deliberately chose to use this name for the entire region.

We must also be aware that the Romans, may in fact, have created the situation of one dominant group in an attempt to aid their own control and conquest of the territory. It seems for instance that areas with diffused power, such as Amorica and Dorset were more difficult to conquer and subsume into the empire (e.g. see Suetonius on Vespasian's campaign in Dorset). There is growing evidence that the Romans aided client kings in order to create a stable, unified unit which could be more easily subsumed into the Empire (Creighton 2000), thus the 'Dobunni' may have been created in the last decades before the conquest in such a way.

Coinage and Kings

Much of the debate over the Dobunni rests on coinage. Distribution of this coinage has frequently been used to draw up the political boundaries of the Dobunni (Selwood 1984, Cunliffe 1991). There are a number of problems in using the coinage in this way without explicitly discussing what the coins were used for, when, and how they arrived in the archaeological record. A main problem with late Iron Age coinage is that we cannot be entirely sure what it was used for. Most 'Dobunnic' coins actually derive from Roman contexts rather than Iron Age ones, and it seems that the roles of these coins may not only have been complex, but that they may have changed over time. For instance, coinage may have started life as gift exchange, perhaps from kings of chiefs to denote loyalty or acts of service. In addition, they may also have held ritual connotations as some sort of offering to the gods, later (or at the same time) they may have been used to buy goods and services. In the Roman period it appears that they may have been utilised as money in the absence of enough Roman money in the Claudian era, perhaps explaining why so many coins turn up in Roman contexts.

With all these different possible roles for coinage it is perhaps misleading to see a simple correlation between a Dobunnic coin find, and the user and location as being in 'Dobunnic' territory. Such assumptions form a circular argument where by the coins of the region are identified as being Dobunnic because they come from the Dobunnic territory and the boundaries of the territory are defined on basis of whether a site contains 'Dobunnic' coinage. Therefore, rather than viewing simple plots of coins on a map we need to ask- where was that coin found, why was it deposited there and how? For example, two Dobunnic coins are claimed to derive from Hailey Wood camp, a possible Roman temple in southern Gloucestershire, but do they signify a late Iron Age, 'Dobunnic' settlement or were they deposited in the Roman period. The varied use and deposition of the Dobunnic coinage in the Roman period will surely distort the picture of coin loss, reflecting Roman deposition and use of the coins rather than that of the Iron Age people.

Such plots of Dobunnic territory are also highly reliant on which sites have been excavated and where metal detectorists have been active. In this way, excavated sites such as Bagendon inevitably distort the maps whilst areas with an apparent lack of coinage, such as the Severn Estuary and the Forest of Dean, have been interpreted as being a different tribal area or 'no-mans' land (e.g. Van Arsdell 1994). However, this is likely to relate more to the lack of archaeological fieldwork in these areas rather than an indication of political affiliation or density of settlement.

Political history from the coins

'Dobunnic' coins have also been used to reconstruct a political history of the Dobunni. The coins appear to provide named individuals who it is assumed represent kings controlling the coin minting. Hawkes (1961) and later Van Arsdell (1994) have attempted to use these coins to construct a political history of Dobunnic Kings and suggest that the area was dominated by a single or possibly two kings. There are a number of problems with such an approach.

One problem with using coinage to recreate the social make up of the later Iron Age is that it appears to only have been in use from around 50 BC into the early Roman period. We are therefore provided only with clues as to leaders and politics in that specific period and we cannot assume that similar rulers existed before those dates. In addition, the coinage appears to reflect a rather fluid situation and it may be that many of these 'kings' were ruling at the same time as was possibly the case with Bodvoc and Corio (Cunliffe 1984).

Can we at least say that the individuals inscribed on the coins were kings? We certainly know that kings of some sort existed in many later Iron Age societies in Gaul and Southern Britain, such as Cogidubnos of the Catuvellaunni. However, there were many other forms of rule including magistrates and regional chiefs and we should not assume that all Iron Age societies were the same. It is possible an area such as this may have had a number of different individuals striking coinage at the same time and naming them is a matter of terminology depending on how one defines a 'King' or 'Chief'. It seems likely that there were important individuals such as Bodvoc and Corio but to say that they were kings of the Dobunni means that we must assume that the Roman texts are referring to a unified tribe.

The problem with the dating of the coinage means that these names may refer to short political snapshots when someone or a group was able to mint coins but does not necessarily mean that they were rulers over the entire region. There is evidence of minting from Bagendon, Ditches and possibly Salmonsbury (Dunning 1976) but it may have also have been taking place at other late Iron Age sites, where coins appear to be relatively prolific, such as Weston under Penyard (Jackson forthcoming) and Camerton, but which have yet to be fully excavated. It is possible that these 'kings' may have been short-lived and existed as a result of immediate stress and needs such as the Roman invasion (Creighton 2000, 13) as appear to have been the case with Vercingetorix in Gaul.

The existence of the rich burial at Birdlip (Staelens 1982) has also been claimed to represent a queen of the Dobunni. This burial is unusually rich and rare in a period when burials are relatively uncommon. However, she need not be a queen although she is likely to be someone of high status. The fact that this burial is not related to Bagendon also suggests that various individuals of high status existed across the region. Its late date in the mid 1st century AD may also imply influences of external, perhaps Gallo-Belgic fashion and may imply that the nature of the burial has as much to do with fashion as new-found status. In preceeding periods and elsewhere in the region other high status individuals probably expressed their status in different ways, other than lavish burials.

The Later Iron Age in the Cotswolds

It becomes apparent that in order to study the 'Dobunni' we need to look in more detail at archaeological evidence rather than assume that Roman texts accurately refer to the society of the later Iron Age.

Firstly, we need to be explicit about the chronology of the later Iron Age and when the 'Dobunni' may have fitted into this time frame. As discussed, above all historical references for the Dobunni are from much later dates than the Iron Age. It has often been assumed that these refer to a society that existed for a period of time before the Roman invasion. However, if these texts are correct at all they may only refer to the last decades of the 1st century AD. In order to establish whether conventional views of the Dobunni bear any relevance to the rest of the later Iron Age we must compare these opinions with other archaeological evidence.

The middle Iron Age (c400 BC-1st century BC) appears to have comprised an apparently densely settled landscape with a variety of settlement types including; enclosed farmsteads, such as at Frocester (Price 2000) and Birdlip (Parry 1998), open villages in the upper Thames valley, and hillforts such as Uley (Saville 1983), Bredon (Henecken 1938) and Conderton. Recent studies of this period have suggested that the traditional view of the hillforts as elite residences supported by smaller settlements may not have been accurate. Morris' analysis of pottery assemblages (1994) has shown there was apparently little difference in status between smaller enclosures and hillforts. In addition, it seems that both types of settlements were involved in similar farming and industrial activities and it seems unlikely that hillforts acted as consumers with smaller settlements as producers. This suggests that there may not have been chiefs ruling from the hillforts but that a more diverse society existed of independent groups co-operating at some levels of trade and exchange.

Around the 1st century BC a number of new settlements appeared at, for example, Ditches (Trow 1988) and Salmonsbury (Dunning 1976). Some earlier hillforts, such as Conderton, appear to have gone out of use, although others, such as Uley, may have continued to be occupied. Elsewhere similar shifts in settlement appear to have been taking place. At The Bowsings a new enclosure was built, seemingly replacing the more open settlement at The Park (Marshall 1995). In some cases there is no evidence of a shift and the earlier settlement continued to be occupied in the late Iron Age. This can be seen at Frocester where the settlement appears to have continued in use into the early Roman period. Slightly later, in the early 1st century AD, a new type of settlement, often referred to as 'oppida', seems to have emerged comprised of massive discontinuous dyke systems at Bagendon and possibly Grims Ditch (Oxfordshire). These sites, like Ditches, appear to have represented an early development in a move to new types and locations of settlement, which were then developed in the final phase of the Iron Age. Bagendon appear to have been occupied until around AD 60 and then abandoned whilst other sites, such as Ditches and Frocester, developed further into Roman villas.

It is apparent that there were major changes taking place in the settlement patterns of the 1st century BC and 1st century AD. Dating of the changes from many sites suggest that these changes were not uniform but that sites and regions adapted differently and that these changes were not all contemporary. For example, the shift seen from The Park to the more ostentatious enclosure at The Bowsings (Marshall 1995) seems to take place earlier than most, in the late 2nd century BC, whereas sites like the Duntisbourne enclosure seem to emerge late, in the 1st century AD (Mudd 1999). In other cases, some sites, such as Stokeleigh, near Bristol (Haldane 1975), do not show this shift and occupation seems to continue from the middle Iron Age into the Roman period.

Settlement evidence suggests that society in the later Iron Age was in a state of flux, and that settlement changes were not uniform. This lack of uniformity may suggest that settlement changes were not simply ascribable to influences of external trade, marking a sudden shift to a new settlement pattern. Instead it appears to show that whilst changes were taking place it was over a long time frame and that changes were regional with even individual sites reacting differently to new circumstances. Therefore, if we can see that these changes were different on a local level and at different times, this means that the texts may only be referring to a very localised and short time frame in this changing and fluid picture. It may be that these texts refer only to the last few decades before the conquest when the latest sites, such as Bagendon and Duntisbourne emerged, and perhaps just to the situation in the southern Cotswolds. It also suggests that, contrary to the orthodox opinion, society may not have already coalesced into hillfort dominated small 'tribal' groups in the middle Iron Age which were then subsumed into a larger tribal unit in the 1st century BC. Instead, settlements appear to have been relatively independent or dealt with each other on more or less equal terms. Thus the idea of a central king may have been anathema to people for most of the later Iron Age.

The role of late Iron Age sites

What then was the role of the new sites that emerged in the latest phase of the Iron Age? Traditional views regard Bagendon as the tribal capital from which the Dobunnic kings ruled. However, the new sites at Bagendon and Ditches appear to have been located in areas where middle Iron Age settlement is harder to find. This may suggest that the later sites were positioned in areas that had previously been peripheral. This is supported by evidence that the area may have been predominantly woodland until quite late in the Iron Age (Mudd 1999). The nature of occupation at Bagendon indicates that it was not densely settled in the manner of a hillfort or town and that much of the enclosed area may have been farming land or stock corrals. In addition, the area excavated by Clifford suggests it was involved in a number of industrial activities.

The lack of dense settlement and the industrial activity may suggest that Bagendon functioned as a market for groups to meet, trade and exchange. In addition new studies of oppida have suggested that their association with 'wet places' may also suggest a ritual function. Both of these suggestions could explain the large number of coins found and its location in a seemingly peripheral area that may have previously been 'no-mans' land. In addition, there is the suggestion from elsewhere in Britain that some of these late sites may have been heavily involved in new activities such as horse rearing and stock corralling. There is not enough evidence in the form of horse gear from the Cotswolds but this may explain the form of the sites, such as The Ditches and the banjo enclosures, which have large antenna ditches. If true this might also explain the predominance of horse imagery on Dobunnic coinage.

The idea that Bagendon was the only such site may also be false. In addition to Salmonsbury, it appears that there may have been a number of other sites with similar status to Bagendon. For instance we know very little about the role of the Banjo enclosures, such as those at Ashton Keynes and Northleach, and of complex enclosures, such as that at Frampton Mansell (RCHM 1976). Late Iron Age coinage from these sites and dating from elsewhere in southern Britain (eg Corney 1989) suggest that these are likely to be late Iron Age and they may have been involved in similar activities to Bagendon, with similar status. The location of an apparent banjo enclosure at the heart of the Bagendon complex may also suggest that there were similarities between it and the banjo enclosures.

The presence of relatively large number of imports on these sites had led to the claim that these must represent status and thus the centres for kings. However, we should also consider the possibility that only certain groups may have adopted these imports, perhaps those who were part of a new elite or people who were marginal to the established exchange networks. Other high status groups however, such as those already firmly established within the trade networks of Malvern or Glastonbury pottery had no need or desire to change and adopt these imports. It may be no coincidence that Bagendon is on the periphery of the older pottery networks and perhaps more open to influence from outside. Therefore, it is conceivably wrong to assume that a lack of imports signifies a lack of status.

The relationship between late Iron Age settlements and Roman settlement is also problematic. It has been considered that the Romans took over important Iron Age centres and placed their own towns on them, in effect fossilising the pre-Roman settlement pattern. However, it is important not to instantly interpret any Iron Age evidence from beneath Roman towns as indicating that the site may have had a similar status in the Iron Age. Often the location of Roman towns in close proximity to Iron Age settlements may purely be one of coincidence. For example, the growing evidence of Iron Age activity from the Gloucester (Hurst 1999) area does not necessarily imply a pre-Roman 'oppidum' but more likely various farmsteads of varying date and status. In interpreting back from the status of Roman sites we should be wary of viewing Iron Age sites from a Roman perspective rather than assessing them in relation to their role in the complex Iron Age settlement patterns.

At present our understanding of the nature, role and even dating of many late Iron Age sites in the region is limited, making any discussion of late Iron Age society difficult. However, we may be able to conclude that Bagendon was not necessarily alone and that there may have been a number of such sites fulfilling many roles, as elite centres, markets and even as ritual foci. Further large-scale study is needed to define the dating and role of many types of sites including the banjo enclosures and many of the hillforts in the area. In addition, further study is needed to understand the role of the Bagendon complex and its relationship with the Duntisbourne enclosures. Further work might involve a comparative analysis of seed and faunal assemblages from smaller enclosures and larger sites such as Salmonsbury and Bagendon, in order to examine the question of whether smaller enclosures were producing for elite consumers. Fieldwork is also required to determine whether the picture of later Iron Age settlement in areas previously peripheral in the middle Iron Age is real or a construct of variable recording. It is hoped that a forthcoming fieldwork project will answers some of these questions.

Iron Age Pottery and the Dobunni

Another way of seeing the Dobunni that has been suggested is to detect what pottery they were using and identify settlements on that basis. Identifying an ethnic group in such a way is always problematic. Pottery of the middle and late Iron Age varied across the region and between sites. The northern part of the region is dominated by stamped ware derived from the Malverns (called Croft Ambrey-Bredon Hill type by Cunliffe 1991). Further south, two regional types have been identified; Glastonbury-Blaise Castle style and the similar South Western decorated (Glastonbury) ware. In the west, on the apparent edges of the Dobunni, sites like Lydney were using what Spencer has called class B limestone tempered pottery (Saville 1984, Spencer 1983, Cunliffe 1991). These types of pottery appear to have been in use from roughly the 4/3rd centuries BC, and in some cases use may have continued on some sites in to the 1st century AD and possibly even the Roman period. Elsewhere, however, they went out of use in the 1st century BC and some sites adopted wheel thrown wares or 'Belgic' pottery and began to obtain Gallo-Belgic imports and later Samian. This indicates some sites adopting new sources of pottery as well as new styles and presumably uses for their pottery.

Some archaeologists have seen the pottery as relating to tribal affiliations. Cunliffe (1984) for example has claimed that political division in the Dobunni is reflected in the pottery evidence. This division, between Malvernian sourced pottery in the north and Glastonbury ware in the south, is claimed to reflect the apparent division in the Coinage between Bodvoc coinage in the north and Corio in the south.

However, the concept of a link between pottery and ethnic affiliation is questionable. Pottery is unlikely to have ethnic affiliations and is usually related to factors such as availability of material and where the contents of the pottery were being traded. Studies of the pottery from the region by Morris (1985, 1994) have shown that pottery distribution in the middle Iron Age was related to availability and traded within certain market areas. The various types of pottery being used in the region do not neatly coincide with the tribal area identified from the coinage but instead appear to have different market areas. In addition, the discovery of Glastonbury ware pottery as far north as Abbeymead, Gloucester (Atkins 1987) suggests that this division may be less clear cut and that pottery distribution probably has more to do with ability to trade along rivers rather than an indication of tribal allegiance.

Chronology is also important. The division in coinage (if one existed) may only relate to a short time frame in the 1st century AD. However, the pottery to which it is being compared, dates back as early as the 3rd/4th century BC and there may well be no relationship between them. Cunliffe's argument attempts to suggest that the region was already firmly divided politically in the middle Iron Age but, as the settlement evidence shows, there is little evidence for such a division. Pottery cannot therefore be used to identify tribal groups as its presence on sites was related to many other factors such as availability, status, trade networks and a multitude of cultural choices.

Conclusions and the way forward:

This paper has attempted to present an alternative view of the 'Dobunni' and to question established views that they were a coherent tribal group. Perhaps I have not identified the adapter between the archaeology and the Roman text based version of the Dobunni, which Richard Reece has asked for. However, it is my view that when studying the Dobunni we are in fact studying the nature of society in the latest phase of the Iron Age and early Roman period but we are not necessarily studying one tribe or people. The archaeology appears to show a complex world where different areas and settlements had different and perhaps changing roles and allegiances. It appears that the old idea of a centralised tribe with a single king is too simplistic and that there may have been various groups and chiefs vying for power, some adapting and dealing with Roman trade, others continuing to use pottery and settlements of the middle Iron Age. The archaeology suggests that each site and region may have reacted to the changes taking place in the 1st century BC / AD in different ways and we should not just accept the idea, from what is essentially Roman propaganda, that late Iron Age society merged seamlessly into the Roman civitas. In such a period of change the whole concept of identity and ethnicity may have been in turmoil and the concepts of Dobunnic and even Roman identity were probably more fluid than we often imagine. The term Dobunnic may only have been a useful label for the Roman establishment to use. Similarly, for modern archaeologists the term Dobunnic can only be used as a label to describe the last decades of the Iron Age.

How then can we move forward in studying the later Iron Age? As I have suggested above, a greater understanding of the nature of the settlement patterns is needed to determine to what extent sites like Ditches are exceptional or if other similar sites existed. Such studies may determine the extent to which there were settlement hierarchies and whether these sites were indeed elite residences. At present our understanding of the chronology is poor and needs to be refined in order to understand the dating of the changes evident in the settlement record and to determine to what extent supposedly 'middle Iron Age sites' continued to be occupied in the late Iron Age. There is considerable need to understand the use and role of coinage as well as why and how it was deposited, to determine what it can tell us of later Iron Age society. Such studies would enhance our knowledge of the nature of society and how it changed over the later Iron Age. It is only by understanding how these groups were developing throughout the Iron Age and Roman period that we can understand where the text based concept of the Dobunni fits into this narrative.

Part 3: Afterword - Richard Reece

Part 1 was demolition, or at least scepticism. Part 2 was rather like reconstructive surgery - cutting out the bad bits and trying to build on the good bits. Where are we now? The answer to that must depend on each reader.

If you want nice neat parcels of the past then you will probably have to keep on believing in the coinage as the Dobunnic adapter, or definer. Provided you remember what you have done there need be no problem. The Dobunni were the people who used 'those' coins; the Dobunnic area is the distribution area of 'those' coins; the Dobunni are therefore a phenomenon of the late 1st century BC and 1st century AD when 'those' coins were produced, used, and deposited. There are names on the coins and they probably belonged to important people but, in spite of many claims there is no fine dating possible for the coins. Any Dobunnic Family Saga constructed is total fantasy, but none the worse for that so long as it is admitted.

You could consider adding pottery as another Dobunnic definer, but Tom Moore clearly does not recommend this. There is the problem of the dates of the northern (Malvern) and southern (Glastonbury) pottery types which are centuries away from the dates of the northern and southern coins. I would add to this that it would be very odd to define the Dobunnic region as the southerly part of the northern pottery and the northerly part of the southern pottery thus leaving out all the rest of the distribution areas of the two types of pottery.

Why are we both unhappy about definition by coins? The main point seems to be one of unnecessary restriction. For instance: can we define Dobunnic pottery? Yes, but only when it is the same date as the coins and has the same distribution as the coins. Wider dates, or wider distributions cannot be called Dobunnic. You could say that some pottery was used in the Dobunnic area, but if it is also used elsewhere the whole process becomes too complicated and the label Dobunnic is pointless.

Can we study the Dobunni in the Roman period? Yes, in textual terms; no in archaeological or material terms. The definition of the area may be valid for the period 50 BC to AD 50, but is questionable for the second century AD and almost certainly useless for the 4th century AD. Does this invalidate all study? Of course not, but the study, to be valid, must define itself in valid terms. So the area of distribution of pottery produced near Corinium can perfectly well be studied at any date and this may give us an idea of the likely market area of Roman Corinium. If we add the distribution of tiles, mosaics, bronze objects and so on we could get a reasonable idea of market regions in the Roman period. But the moment the label Dobunnic appears the whole study has to be constricted into whatever form has been used for the definition of the Dobunni.

Which brings us back to the beginning. Why do people want to confuse modern archaeology by using ancient labels? I think it is a matter of power, and behind that, the need to be in control. If we can define an ancient people we can limit them and have control over them. If we do not impose names and boundaries the people of the past, rather like unwanted aliens, may get through to us on their own terms rather than ours.

And we don't want that, do we?

References