You are on page 1of 186

South Wales in the Iron Age and Roman

Periods

Jerrad Lancaster
Submitted to Swansea University in fulfilment of the requirements for the Degree of
Master of Philosophy
Swansea University
2012
i

Summary

The aim of this thesis is to examine archaeological and documentary evidence of


South Wales during the Iron Age and Roman periods with the intention of better
understanding the settlement, economic and socio-political structures of the Silures before
and during the Roman occupation of the area. In the first part, a study of Iron Age defended
enclosures with regard to shape, defences, settlement location and internal size indicates an
overall settlement structure of multiple independent groupings situated throughout South
Wales. Through the consideration of the roles of hillforts within economic, social and
political interaction, these individual groupings most likely indicate a hierarchical system of
decentralised clans acting independently of one another, but under a single, shared cultural
identity.
The second part of this thesis analyses the impact of the Roman conquest of South
Wales and the extent to which the native population was transformed by the Roman military
and civil establishments. The Roman military strategy of conquest and control over South
Wales proves further the decentralised political structure of the Silures in the Iron Age.
Roman and Romano-British settlement patterns indicate a changing lifestyle influenced by the
Roman occupation with a native population choosing at what degree to embrace the new
culture. Evidence suggests this change was immediate providing a new Roman way of life
yet keeping much from the previous Iron Age existence.
ii

Table of Contents
Acknowledgements...................................................................................................................iii
List of Maps..............................................................................................................................iv
List of Figures...........................................................................................................................iv
List of Illustrations....................................................................................................................iv

Part 1: South Wales before the Romans

Chapter 1: Introduction..............................................................................................................1
Chapter 2: Defining the Territory of the Silures......................................................................13
Chapter 3: Silurian Settlement Patterns....................................................................................22
Chapter 4: The Role of the Settlements within the South Wales Society................................43
Chapter 5: The Trade Network of South Wales.......................................................................53
Chapter 6: Silurian Social Politics............................................................................................62
Chapter 7: Defining a Silurian Political System......................................................................77

Part 2: Roman South Wales

Chapter 8: From AD 43 to Southern Wales.............................................................................84


Chapter 9: The Roman Military Network in South Wales.......................................................92
Chapter 10: Urban Life...........................................................................................................105
Chapter 11: The Rural Society...............................................................................................125
Chapter 12: Silurian South Wales Conclusion.......................................................................146

Appendix 1: Updated List of Defended Enclosures in South Wales......................................148


Appendix 2: Maps: Roman Military Occupation in South Wales..........................................157
Illustrations.............................................................................................................................159
Works Cited............................................................................................................................161
iii

Acknowledgements
Maps of South Wales used herein are © Joanne Edwards, designed through the use of
Ordnance Survey data (©Crown Copyright Ordnance Survey. All rights reserved) and used
with permission. Additions to the maps, including the location placement and labelling, were
made by the author. Other images used are © The National Museum of Wales, Cardiff, The
Glamorgan-Gwent Archaeological Trust and/or Cadw, used with permission and cited as
such.

The author wishes to thank Eddie Owens and everyone at Swansea University, Cardiff
University, Cadw and The Glamorgan-Gwent Archaeological Trust for their help throughout
research. Support from Joanne Edwards in the illustrations and at the beginning stages of
work was essential. Lastly, thanks to my mother for the hours spent proofreading, as well as
the neverceasing support from my friends and family without which none of this would have
been possible.
iv

List of Maps

Map 1: South Wales with rivers and major modern locations mentioned herein.......................1
Map 2: The preserved counties of South Wales.........................................................................4
Map 3: Definite, probable and possible settlements in South Wales.......................................23
Map 4: Hillforts of 2 ha in size or greater within the South Wales settlement pattern............38
Map 5: Possible Iron Age Clan Groupings...............................................................................79
Map 6: Major Roman sites in South Wales..............................................................................92
Map 7: Venta Silurum (Caerwent)..........................................................................................106
Map 8: Romano-British farmsteads and Roman villas...........................................................127
Map 9: Pre-Flavian Roman sites to AD 60.............................................................................157
Map 10: Roman sites to AD 80..............................................................................................157
Map 11: Roman sites to AD 150............................................................................................158
Map 12: Roman sites c. AD 300.............................................................................................158

List of Figures
Figure 1: Prehistoric Defended Enclosures of Known Size in Gwent......................................28
Figure 2: Prehistoric Defended Enclosures of Known Size in Glamorgan..............................31
Figure 3: Prehistoric Defended Enclosures of Known Size in Brecon.....................................32

List of Illustrations

Illustration 1: Seven Sisters Hoard (c. AD 1-150)..................................................................159


Illustration 2: Terrets – Chariot Rein-Guides (c. 300 BC – AD 100).....................................159
Illustration 3: Bronze Strap-Union with Horse Bell of La Tène or Celtic Art style
(c. AD 1-120)...................................................................................................
Illustration 4: Bronze Moulded Tankard Handles (c. AD 1-150)...........................................159
Illustration 5: The Goldcliff Stone..........................................................................................159
Illustration 6: A dupondius of Antoninus Pius (c. AD 140-4)................................................159
Illustration 7: Light Poldon Fibula (c. first century AD)........................................................160
Illustration 8: T-Shaped fibula (c. late first to middle second century AD)...........................160
Illustration 9: Iron Intaglio Ring of Achilles (c. first half of the second century AD)...........160
Illustration 10: Iron Ring with intaglio of a pastoral scene
(c. first or early second century AD)...............................................................
1

Part 1: South Wales before the Romans

Chapter 1: Introduction

Map 1: South Wales with rivers and major modern locations mentioned herein
©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved

The Silures were a tribal group situated on the southern coast of modern day Wales
(Map 1). The tribe is first mentioned in Tacitus’ biography of his father-in-law, Gnaeus
Julius Agricola. In Agricola, Tacitus writes of the curly hair and tinted complexion of the
Silures as possible descendants from the Iberian Peninsula. 1 Tacitus also mentions the Silures
in his account of the Roman conquest of Britain in Annals and the problems coming from the
rebellion of the Silures in South Wales. 2 Unfortunately, not much more is known from the
written history of the tribe inhabiting southern Wales. Any further indication of the Silures
needs to be taken from the interpretation of archaeological evidence.
The aim of this thesis is to examine archaeological evidence of South Wales during
the Iron Age and Roman periods with the intention of better understanding the settlement,
economic and socio-political structures of the Silures before and during the Roman
occupation of the area. Initially, this will be shown chiefly through the settlement patterns
1
Tacitus. Agricola. 11.
2
Tacitus. Annals. 12.32-3.
2

provided by known and probable defended enclosures identified within the territory of the
Silures. Archaeological evidence illustrating the economic and social systems in place during
the Iron Age will bring about a further understanding of the political arrangement in South
Wales at the time. An examination of the Roman invasion and subsequent occupation period
will then demonstrate the development of the Silures into a more Romanised lifestyle while
still incorporating native aspects. Chronologically, this dissertation will focus upon Iron Age
and Roman South Wales, but occasional references to the Bronze Age or earlier will be made
to further the central theme.
It must be mentioned that the recognition of the Iron Age population in South Wales
as the Silures is a later Roman identification. There is no evidence to confirm that the natives
of the area labelled themselves as such. Yet equally, there is nothing to suggest otherwise.
As mentioned, Tacitus identified the Silures when describing the native resistance during the
Roman conquest of South Wales. During the Roman period, the civitas-capital, Venta
Silurum, and the Paulinus inscription both refer to the population of South Wales as the
Silures. Conversely, Roman descriptions of other British tribes have proven problematic, as
can be seen in Caesar’s list of surrendered British tribes during his second invasion of Britain
in 54 BC.3 The Cenimagni, Segontiaci, Ancalites, Bibroci and Cassi are all listed as tribes
who sent embassies surrendering to Caesar, yet none of those can be linked with currently
identified tribes known from the Claudian Invasion of AD 43. Ultimately, to continue with
modern scholarly work, the classification of the population and tribal territory designation of
the tribe of South Wales will remain that of the Silures.4
Academic work regarding Iron Age South Wales and the Silures has not been absent
by any means. However, one view of the Silurian territory as suffering from a ‘paucity of
survey, excavation, and research’ is not far from the truth; work on the Iron Age Silures is
impaired by a lack of available evidence. 5 Gwilt recognized this need and addressed it with
great fervour, presenting a collection of evidence from hillforts to material culture artefacts. 6
His effort has proved to be invaluable for all future work on the area and will be a major
source of evidence supporting the ideas given in this thesis. It is the aim of this dissertation to
use what Gwilt has collected as an avenue towards furthering what is known and understood
about Silurian culture. The work herein is an effort in inference and interpretation to use all
that is available in the archaeological record to present a possible understanding of how the
Silures lived. The Glamorgan-Gwent Archaeological Trust (GGAT) is another important
source of information used within, most notably its work on prehistoric defended enclosures
3
Caesar. Gallic Wars. 5.21.
4
For more on the tribal boundaries of the Silures, see below, page 13.
5
Gwilt, A. (2007). ‘Silent Silures? Locating People and Places in the Iron Age of South Wales’, in
Haselgrove, C. and Moore, T. (eds.) (2007). The Later Iron Age in Britain and Beyond. Oxford: Oxbow
Books, 297.
6
Gwilt (2007), 297-328.
3

in the preserved counties of Glamorgan and Gwent. 7 The compilation of defended enclosures
will further Gwilt’s work, as well as provide detailed evidence of each location, allowing
arguments focused upon the importance of specific sites to each other and within the society.
In the last quarter of the twentieth century, research on Iron Age southern Wales was
predominately focused upon the area of South Wales known until recently as Glamorgan.
This was in large part due to The Royal Commission on Ancient and Historical Monuments in
Wales (RCAHMW) in its inventory of the Iron Age and Roman occupation ancient
monuments of Glamorgan in 1976.8 As well as this, volumes of work on the Glamorgan
county history were published in the 1980s by Savory, including one volume focused on pre-
history and early history.9 Subsequent publications on excavations in Glamorgan have
themselves been predominately from sites in the modern Vale of Glamorgan (e.g. Biglis,
Caldicot, and Llandough).10
However, recent focus in the last decade has shifted toward the area of South-east
Wales known as Gwent. Howell has provided much of the work and in 2006 with Pollard,
Chadwick and Leaver released an excavation report on Lodge Wood Camp, 11 and in 2004
edited with Aldhouse-Green volume 1 of The Gwent County History.12 Much of the Iron Age
research featured in the latter publication was expanded upon in 2006 in Searching for the
Silures, the first such publication solely on the Silures. 13 In this work Howell does mention
the possible Silurian evidence found in western Glamorgan, yet focuses much of the research
on Gwent and the Vale of Glamorgan. Despite this, his publication is invaluable towards any
research on the Silures. Most recently, Gwilt expanded upon Howell’s studies, and in doing
so purposefully focused upon the area of Gwent. Citing the RCAHMW and Savory volumes,
Gwilt ‘[aimed] to balance up this unevenness by giving prominence to the archaeology and
material culture of Gwent.’14 Thus, this ‘unevenness’ has seemingly been shifted in favour of
Gwent. Because of this, much of what will be presented is found in Gwent, which obviously
gives an impression of Gwent as more developed or ‘civilised’ than Glamorgan. However

7
Wiggins, H. (2006). Southeast Wales Prehistoric Defended Enclosures: Prehistoric Defended
Enclosures in Gwent. GGAT Unpublished Report No.2006/021; Gerrard C., Wiggins H. and Evans, E
M. (2006). Prehistoric Defended Enclosures in Glamorgan with Recommendations for Fieldwork.
GGAT Unpublished Report Nos. 2006/089.
8
RCAHMW. (1976b). An Inventory of the Ancient Monuments in Glamorgan, Vol. I: Pre-Norman. Pt.
II: The Iron Age and the Roman Occupation. Cardiff: HMSO.
9
Savory, H.N. (ed.) (1984). Glamorgan County History, Vol. II: Early Glamorgan: Pre-History and
Early History. Cardiff: Glamorgan County History Trust.
10
Robinson, D.M. (ed.) (1988). Biglis, Caldicot and Llandough: Three Late Iron Age and Romano-
British Sites in South-East Wales. Oxford: British Archaeological Reports British Series 188.
11
Pollard, J., Howell, R., Chadwick, A. and Leaver, A. (2006). Lodge Hill Camp, Caerleon and the
Hillforts of Gwent. Oxford: British Archaeological Reports British Series 407
12
Aldhouse-Green, M. and Howell, R. (eds.) (2004). The Gwent County History Vol. 1: Gwent in
Prehistory and Early History. Cardiff: University of Wales Press.
13
Howell, R. (2009). Searching for the Silures: An Iron Age Tribe in South-east Wales. The History
Press.
14
Gwilt (2007), 297.
4

this cannot be assumed, and only through further work in Glamorgan can this material bias be
eliminated. It is in light of the most recent work on Silurian South Wales that this dissertation
hopes to not shift away from Gwent, but spread focus onto the entire territory of the Silures,
and in doing this raise the importance of the hither-to mostly ignored West Glamorgan and
south Powys (formerly Brecknock) regions (Map. 2).

Map 2: The preserved counties of South Wales


©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved

Academic work in Brecon is not as prevalent as the rest of South Wales, but
nonetheless is not lacking. Savory dominates early work with overviews of the prehistoric
area and his subsequent excavations15, while around the end of the twentieth century
independent surveys were published on ancient features in the Brecon area. 16 The most
thorough work on hillforts in the area was published in 1986 by RCAHMW, similar to the
inventory covering Glamorgan in 1976.17 Subsequently, a new survey in 2007 by the Clwyd-
Powys Archaeological Trust (CPAT) added an additional nine defended enclosures in
Brecknock.18

15
Savory, H. N. (1952). ‘List of Hill-forts and Other Earthworks in Wales: III. Brecknockshire’,
Bulletin of the Board of Celtic Studies. 14, 69-75; Savory, H. N. (1955). ‘Prehistoric Brecknock’ ,
Brycheiniog. 1, 79-127; Savory, H. N. (1961). ‘Twyn Lechfaen hill-fort excavations 1959’, Bulletin of
the Board of Celtic Studies. 19, 173-176.
16
Burnham, H. (1995). A Guide to Ancient and Historic Wales: Clwyd and Powys. London: HMSO;
Children, G. and Nash, G. (2001). Prehistoric Sites of Breconshire: Ideology, Power and Monument
Symbolism. Monuments in the Landscape Vol. 9. Hereford: Logasten Press.
17
RCAHMW. (1986). An Inventory of the Ancient Monuments in Brecknock (Brycheiniog). Vol. I: The
Prehistoric and Roman Monuments. Pt. II: Hillforts and Roman Remains. London: HMSO.
18
Silvester, R.J. (2007). Defended Enclosures in Brecknock. CPAT Report 859, 8.
5

Initially, the archaeological record of Gwent and Glamorgan, shown through the
GGAT reports and other publications, will be explained here separately. This modern
separation has been maintained over time in research done on Iron Age South Wales, and
although it can be seen as an arbitrary distinction, it seems best to keep the same format. As
will be shown, the two preserved counties do exhibit numerous differences, and furthermore,
for the sake of ease in demonstrating what is known about South Wales, discussing Gwent
and Glamorgan counties individually allows for a better understanding. Afterward, further
examination, including implementing theoretical views of the Silures through inferences of
the archaeological record, will be placed over all of South Wales collectively. It is the
intention of this thesis that in doing this an understanding of the Silures will be given showing
a loose decentralised federation and the complex social system at the time of the Roman
conquest.
This dissertation includes a discussion of the boundaries of the Silures in an effort to
identify the extent of their territory and cultural domination. Contrasting archaeological
findings with those of the surrounding areas will be shown to be strong indicators of different
cultural entities at work. Another focus on defended enclosures will show the settlement
arrangement of the Silures from which ideas of the cultural reasoning behind the settlement
pattern will be discussed. Throughout this dissertation, the discussion of defended enclosures
will follow the idea of larger sites being of greater importance, whether in displays of wealth
or status or in symbolic or communal purposes within the community, and involving an elite
aristocracy which places importance upon outward displays of wealth and status. Larger
hillfort enclosures with multiple elaborate defences are just one avenue of demonstrating such
affluence.19 Other possible purposes for building large hillforts may be social, political,
religious, agricultural/pastoral or even as a source of refuge during hard times. 20 Additionally,
it could also be that defended enclosures built on such a large scale may have incorporated
many or all of the above mentioned functions. Equally, small settlements probably sustained
only small families or groups, most likely dependent upon the aristocracy within the larger
hillforts for resources (e.g. elite goods) which are harder to obtain.21
19
Cunliffe, B. (2005). Iron Age Communities in Britain: An Account of England, Scotland and Wales
from the Seventh Century BC until the Roman Conquest. 4th Edition. London: Routledge, 252, 594;
Davies, J.L. and Lynch, F. (2000). ‘The Late Bronze Age and Iron Age’, in Lynch, F., Aldhouse-
Green, S. and Davies, J.L. (2000). Prehistoric Wales. Stroud: Sutton, 158; Children and Nash (2001),
129.
20
Cunliffe, B. (1971). ‘Some Aspects of Hill-forts and Their Cultural Environments’, in Hill, D. and
Jesson, M. (eds.) (1971). The Iron Age and its Hill-forts: Papers Presented to Sir Mortimer Wheeler on
the Occasion of his Eightieth Year, at a Conference Held by the Southampton University
Archaeological Society, 5th-7th March, 1971. Southampton: Millbrook Press Ltd., 62-3; Jackson, D.
(1999). ‘Variation in the Size Distribution of Hillforts in the Welsh Marches and the Implication for
Social Distribution’, in Bevan, B. (ed.) (1999). Northern Exposure: Interpretive Devolution and the
Iron Ages in Britain. Leicester: Leicester Archaeology Monograph 4, 211; Davies and Lynch (2000),
162; Howell (2009), 36-7.
21
Davies and Lynch (2000), 148.
6

Furthermore, artefacts of the Silurian material culture, as seen in the archaeological


record, will be shown with regard to the sites at which they were discovered. The relationship
of these cultural materials to the location in which they were found, as well as with other
artefacts, will be discussed to explain the importance, implications and impact that the items
had upon the area and the society as a whole.
During the course of this thesis religion will be briefly discussed, but there will be no
attempt here to recreate any structure, concepts, or beliefs held by the Silures. Other scholars
have already attempted to place religious artefacts found in South Wales within an existential
concept of how the Silures may have worshipped.22
One short coming of archaeology is that the only definitive answers found are in the
existence of the materials themselves. The purpose for and reasoning behind the existence of
such findings cannot always be deduced by their context. Conversely, a lack of
archaeological finds can be misleading as it may suggest little to no historical activity, but
such a point of view should be dismissed. For example, Iron Age and Roman archaeology in
the areas of modern Cardiff, Newport and Swansea has been lost overtime to the eventual
expansion of the population centres, but the relative scarcity of finds should not indicate a
lack of historical relevance. Moreover, anthropological findings elsewhere may provide a
better understanding by looking beyond the physical evidence of archaeological remains into
an understanding of the possible uses for such findings, and still further into the reasoning
behind them. Most importantly, however, any theories of Silurian culture and lifestyle gained
through anthropological studies must not be taken as definitive for South Wales. Rather, the
purpose of incorporating anthropological work, and this dissertation as a whole, is merely to
further the overall study of the Iron Age tribe inhabiting the area.
Silurian social systems will then be discussed to deduce plausible social, economic
and political organisations while delving into the entire material culture of the Iron Age in
South Wales. Through this a better picture will emerge of the Silurian upper class and their
position within the entire cultural territory as not only leaders, but providers of resources and
material goods. Decentralised tribal organisations have been recognised in other British Iron
Age tribes (i.e. the Iceni.23). It will be shown that a similar political system was most likely in
place in Silurian South Wales, and an examination into the settlement pattern reflecting upon
the elite social situation will optimistically provide groundwork towards surmising the layout
upon which the decentralised political system adhered.

22
Davies and Lynch (2000), 211-6; Aldhouse-Green, M. (2004). ‘The Iron Age: Art, Ritual and
Society’, in Aldhouse-Green and Howell (2004), 160-77; Howell (2009), 21-34; Gwilt (2007), 297-
328.
23
Warner, P. (1996). The Origins of Suffolk. Manchester: Manchester University Press, 29-31; Sealey,
P.R. (2004). The Boudican Revolt Against Rome. 2nd Edition. Princes Risborough: Shire, 10.
7

During the course of this thesis, evidence will be presented pointing towards a
settlement pattern of a clan-based association. Political connections within a possible Silurian
association can only be inferred and cannot be demonstrated conclusively. However, political
relationships between regional leaders are a natural step that most likely occurred within the
local groupings which will be demonstrated herein. Natural boundaries along the borders of
preserved counties continue to show accepted territorial divisions, whether intentional or
accidental. Modern boundaries cannot be viewed as definite territorial edges, tribal or clan,
but the preserved counties of Glamorgan and Gwent along with parts of the Brecon area seem
to show definite differences within all of South Wales. As mentioned, hillfort enclosures are
often viewed as monuments signifying status and ownership of land. 24 This ownership of
land could even be extended to an entire region reaching multiple settlement groups and
creating a clan connected through a ruling family or class.
Throughout South Wales a high number of settlements can be found of varying sizes
representing pre-Roman occupation. It is obvious through the archaeological record that an
Iron Age population had already been established in southern Wales. However, subtle yet
distinct differences can be found within the tribal group. As mentioned, recent work done by
the Glamorgan-Gwent Archaeological Trust (GGAT) has managed to bring about an updated
record of prehistoric defended enclosures in both Glamorgan and Gwent. 25 Unfortunately,
much of the record is through field-walking, survey and aerial photography, uninvasive
methods that do well in leaving land undisturbed, but provide little beyond what can be seen
on the surface. Furthermore, prehistoric sites recorded are not necessarily of the pre-Roman
Iron Age, but could have been abandoned centuries before Roman occupation. Caution,
therefore must be placed in using undated sites for the purpose of deducing the Iron Age tribal
formation. Yet, what has been excavated in the area can start to bring in an idea of how the
Silures lived, and through this archaeological evidence, sites still unexplored can be of value.
Even sites which are of the Bronze Age or earlier can be helpful in demonstrating continuity
of settlement patterns.
Unenclosed settlements have been discovered in South Wales although not enough
evidence has been found to provide many definitive answers to that aspect of the Iron Age
Silures.26 Twelve unenclosed roundhouse platforms were documented recently in western
South Wales, all of which of elevation over 300 metres. 27 These were most likely part of a
seasonal pastoral occupation.28 Two unenclosed sites have been dated to the Late Iron Age:

24
Children and Nash (2001), 129.
25
Wiggins (2006); Gerrard et al (2006).
26
RCAHMW (1976b), 72-9; Howell (2009), 48; Egloff, S. (2009) Later Prehistoric Undefended and
Roman Rural Settlements in Southeast Wales. GGAT Unpublished Report No.2009/013; Egloff, S. and
Evans, E. M. (2010) GGAT Unpublished Report No.2010/019.
27
Egloff (2009); Egloff and Evans (2010).
28
See below, page 11.
8

Biglis in the Vale of Glamorgan, and Caldicot in Gwent. 29 Both were later enclosed with
palisades and banks during the Romano-British period typical of the other defended
enclosures in the area. This indicates a continual occupation through the Roman conquest,
leaving the possibility of other unenclosed settlements remaining in use. As well, an
unenclosed settlement exists near Merthyr Mawr Warren, a part of which may prove to be the
most important of those in the western area of the Silures. 30 Unfortunately, unenclosed
settlements leave less behind in the archaeological record, and are more likely to be lost to the
effects of time and subsequent development. Thus, what is presented here will be focused
upon defended enclosures; the remains of which are more readily available.
Part two of this thesis will discuss the impact the Roman conquest had upon the
Silures. Evidence of the natural intraregional separation between the Silures and
neighbouring tribes will be furthered through the Roman army’s invasion and occupation
strategy in the first century AD, specifically with regards to the placement of Roman forts.
Settlement patterns changed drastically with Roman influence, yet Iron Age sites were not
wholly abandoned. How the Silures adapted to the changing social and political lifestyles
will be another focus of this dissertation. Using artefacts found of Roman period date and
within Romano-British settings, the impact from the influx of Roman control in the area will
be shown.
Scholarly work on the Roman period in Britain is by far more readily available than
its Iron Age counterpart. Of the most modern published material, Salway’s Roman Britain,
Millet’s The Romanization of Britain, and most recently Mattingly’s An Imperial Possession
are all key texts for the entire history of the British Isles and the archaeological
interpretations.31 The Roman Inscriptions of Britain by Collingwood and Wright and the
multiple edited publications based on the text which have followed are also important
archaeological texts for the epigraphic record of Roman Britain. 32 The study of the towns of
Roman Britain have benefited greatly from The Place-Names of Roman Britain by Rivet and
Smith and Wacher’s The Towns of Roman Britain.33 Annual publications including Britannia
and The Journal of Roman Studies continually provide essential material on the most
contemporary topics.

29
Parkhouse, J. (1988). ‘Excavations at Biglis, South Glamorgan’, in Robinson (1988); Vyner, B.E.
and Allen, D.W.H. (1988). ‘A Romano-British Settlement at Caldicot, Gwent’, in Robinson (1988);
Gwilt (2007), 302.
30
Davies and Lynch (2000), 213.
31
Salway, P. (1981). Roman Britain. Oxford: Oxford University Press; Millet, M. (1990). The
Romanization of Britain. Cambridge: Cambridge University Press; Mattingly, D. (2006). An Imperial
Possession: Britain in the Roman Empire. London: Penguin Books.
32
Collingwood, R.G. and Wright, R.P. (1965). The Roman Inscriptions of Britain I: Inscriptions on
Stone. Oxford: Clarendon Press.
33
Rivet, A.L.F. and Smith, C. (1979). The Place-Names of Roman Britain. London: Batsford; Wacher,
J. S. (1995). The Towns of Roman Britain. 2nd Edition. London: BCA.
9

In reference to Roman Wales, and more specifically South Wales, numerous


publications prove invaluable. Roman and Early Medieval Wales by Arnold and Davies
covers well the entirety of Wales during those periods. 34 Manning and Brewer’s works in The
Gwent County History are key for the preserved county of Gwent during the conquest and
entire Roman period, respectively.35 Roman Frontiers in Wales and the Marches edited by
Burnham and Davies is essentially the third edition of Nash-Williams’ 1954 The Roman
Frontier in Wales,36 and provides the most up to date work on Roman Wales. 37
Archaeological reports are also priceless in the benefit they provide to the advancement of
knowledge of the area. In the first half of the twentieth century, Nash-Williams consistently
published excavations throughout South Wales on a yearly basis, which can be found in the
journals Archaeologia, Archaeologia Cambrensis and Bulletin of the Board of Celtic Studies
(now titled Studia Celtica). The latter two annual publications, coupled with the journals
Archaeology in Wales and Morgannwg, which covers just the former county of Glamorgan,
continually strive to present the most recent work in all historical periods of Wales. Other
archaeological publications include the excavations at the pre-Flavian legionary fortress of
Usk by Manning, those inside the Flavian legionary fortress of Caerleon by Evans and
Metcalf, work in the Caerleon canabae by Evans and the Romano-British farmstead/villa at
Whitton by Jarrett and Wrathmell. 38 Much archaeological work, including that inside and
outside of Caerleon, was managed by, or at least in cooperation with, the Glamorgan-Gwent
Archaeological Trust, and contributions from them, both published and unpublished, prove to
be central to many of the discussions presented herein. A Research Framework for the
Archaeology of Wales released a list of key Roman sites within south-east Wales in 2003
which coincides with the initiative by local archaeologists to document, research and preserve
the history of Wales.39

34
Arnold, C.J. and Davies, J.L. (2000). Rome and Early Medieval Wales. Stroud: Sutton.
35
Manning, W.H. (2004). ‘The Romans: Conquest and the Army’, in Aldhouse-Green and Howell
(2004), 178-204; Brewer, R.J. (2004). ‘The Romans in Gwent’, in Aldhouse-Green and Howell (2004),
205-243.
36
The second edition edited by Jarrett. Nash-Williams, V.E. (1954a). The Roman Frontier in Wales.
Cardiff: University of Wales Press; Jarrett, M.G. (ed.) (1969). The Roman Frontier in Wales. 2nd
Edition. Cardiff: University of Wales Press.
37
Burnham, B.C. and Davies J.L. (eds.) (2010). Roman Frontiers in Wales and the Marches.
Aberystwyth: RCAHMW.
38
Manning, W.H. (1981). Report on the Excavations at Usk 1965-1976: The Fortress Excavations
1968-1971. Cardiff: University of Wales Press; Manning, W.H. (1989). Report on the Excavations at
Usk 1965-1976: The Fortress Excavations 1972-1974 and Minor Excavations on the Fortress and
Flavian Fort. Cardiff: University of Wales Press; Evans, D.R. and Metcalf, V.M. (1992). Roman Gates
Caerleon. Oxford: Oxbow Monograph 15; Evans, E. (2000). The Caerleon Canabae: Excavations in
the Civil Settlement 1984-90. London: Britannia Monograph Series No. 16; Jarrett, M.G. and
Wrathmell, S. (1981). Whitton: An Iron Age and Roman Farmstead in South Glamorgan. Cardiff:
University of Wales Press.
39
See http://www.archaeoleg.org.uk/pdf/roman/KEY SITES SE WALES ROMAN.pdf., Last Date of
Access: 4/17/11.
10

As a whole, this dissertation will attempt to illustrate not only the change the Silures
had to contend with after the Roman conquest in the first century AD, but also how the
natives of South Wales adapted to their new lifestyles, yet maintained much of their Iron Age
traditions.

A Geography of South Wales


The natural landscape of South Wales is bordered to the north by the Brecon Beacons
and Black Mountains and to the south by the Bristol Channel. In between, hills and valleys
follow the rivers that form in the mountains and drain at the coast. Upland areas, above 250
metres in elevation, and lowland regions provide stark contrasts to subsistence, where the
lowland’s more hospitable environment is shown in its long history of habitation. Eastern
South Wales (Gwent) contains a more lowland landscape between the Wye and Usk rivers.
Glamorgan is dominated by coastal plains and internal uplands with multiple valleys formed
by rivers flowing from the Brecon Beacons to the coast. The only exception to this is the
Gower peninsula which will be shown to maintain an isolated existence separate from the rest
of South Wales. The northernmost recesses of the Usk river valley, north of the Brecon
Beacons and incorporating modern Brecon, are located within upland and mountain regions
almost separating the area from the rest of South Wales. These varying landscapes will be
seen as major aspects of the differing settlement patterns not only between the east and west
of South Wales, but within local areas as well.

Climate Change
At the end of the Bronze Age, global factors were altering the lifestyles of peoples
around the world causing population shifts and cultural reorganization. To blame for this
worldwide phenomenon is a differentiation in galactic cosmic ray intensity and solar
irradiation resulting in, amongst other things, cooler sea temperatures, harsh winter-like
stormy weather with higher rainfall, and lower summer temperatures. 40 From approximately
1,200 to 500 BC, populations around the world were dealing with a new way of life: cooler,
wetter climate in the eastern Mediterranean, excessive drought in China, and increased
flooding in North America and Europe resulting in higher water levels. This drastic climate
change led to abandonment of coastal areas, disruptions in trade, increased instability and
decreased predictability of resources, and ultimately changes in social, economic and political
systems.41 This climatic deterioration effected Britain greatly, even being equated with other
population disasters of British history.42
40
Kidder, T.R. (2006). ‘Climate Change and the Archaic to Woodland Transition (3000-2500 Cal B.P.)
in the Mississippi River Basin’, American Antiquity. 71(2), 212; 214.
41
Cunliffe (2005), 33-4; Kidder (2006), 196.
42
Burgess, C. (1985). ‘Population, Climate and Upland Settlement’, in Spratt, D. and Burgess, C.
(1985). Upland Settlement in Britain: The Second Millennium B.C. and After. British Archaeological
Reports British Series 143, 202; Higham, N.J. (1987). ‘Brigantia Revisited’, Northern History. 23, 2.
11

Unfortunately, South Wales shared in the increased flooding of Europe, concentrating


in the Gwent Levels on the Severn Estuary coast. The hardest times for the area were from
1,220 to 900 BC and from 800-470 BC during which the climatic deterioration resulted in
abandonment of the uplands and coastal regions, and a fall in population. 43 Davies labels this
period ‘The Late Bronze Age Crisis’ as the massive loss of exploitable land eliminated much
of the area’s agricultural potential, leaving the uplands capable of only supporting pastoral
farming, which accounts for the change to a predominately pastoral subsistence. 44 Late
Bronze Age settlements were quickly abandoned in favour of more sheltered locations.
From the beginnings of sedentary life and agriculture, and into the turmoil of the
drastic climate change at the end of the Bronze Age, South Wales was in a constant
evolutionary trend. This new environment forced the Silures to adapt to a mixed subsistence
of agriculture and pastoralisation, while another response to the drastic cultural changes may
have been a desire for more personal security to protect themselves and what little resources
that may have been available. Thus, the settlement, economic, and socio-political patterns
presented below were the native population’s response to the uncertainties brought about by
the everchanging environment in which they lived.

Agro-pastoralism
Much of the local uplands were unable to sustain agriculture by the Iron Age, leaving
the lowlands to fulfil this need. Instead the uplands were reserved for pastoral farming, which
eventually would become a major part of the Silurian subsistence. Archaeological evidence
in South Wales portrays a pastorally dominated lifestyle where cattle, sheep and pig dominate
the diet but are supplemented by horse, dog, red deer and even domestic fowl. 45 Fence lines
can be seen through aerial photographs indicating stock control and dental evidence implies
animals are kept for longer than just as sources of meat. Bone remains compared with
farming evidence and an apparent paucity of querns from excavated Silurian sites also hint at
a more pastorally based subsistence. 46 Howell proposes a transhumance lifestyle where stock
follows a family group to summer houses usually at higher elevation, possibly in the
coalfields and Brecon Beacons. Yet, querns are found throughout South Wales in lowland
(i.e. Sudbrook Camp) and upland (i.e. Twyn-y-Gaer (Cwmyoy)) sites, suggesting that
agriculture was still a staple of the Silures’ diet. 47 This is furthered through cropmarks found
within field systems near settlement locations in South Wales. 48

43
Rippon (1996), 22; Davies and Lynch (2000), 140-2.
44
Davies (1995), 672.
45
Howell (2009), 50-3.
46
Gwilt (2007), 313.
47
Howell, R. and Pollard, J. (2004). ‘The Iron Age: Settlement and Material Culture’, in Aldhouse-
Green and Howell (2004), 151-3.
48
See below, below.
12

South Wales during the Iron Age was a landscape of mountains in the north leading
to upland and lowland settlements near the southern coasts. Eastern South Wales contains
more lowland landscapes allowing for more agriculture, while elsewhere west of the Usk
river, higher elevations dominate creating a need for pastoralism. Global climate change at
the end of the Bronze Age left the uplands unsustainable for agricultural purposes, which led
to an increase and eventual reliance upon pastoralism in the area. Lowland settlements,
mostly near the coasts, could rely more upon arable farming, but also maintained a
dependence upon livestock for resources. An overall agro-pastoral lifestyle thus emerged
amongst most of the South Wales inhabitants. Possibly along with this, defended enclosures
became more common methods of protecting families or communities and the local resources.
13

Chapter 2: Defining the Territory of the Silures


Of what can be inferred, the heartland of the Silures consisted roughly of the
preserved counties of Glamorgan and Gwent. Furthermore, an extension of Silurian influence
most likely existed in the Late Iron Age along the southern edge of Powys in the area of
Brecon from the third century BC.49 Glamorgan includes the modern divisions of Bridgend,
Cardiff, Merthyr Tydfil, Neath Port Talbot, Rhondda Cynon Taff, Swansea, Vale of
Glamorgan, and the western half of Caerphilly; while Gwent preserved county includes
Blaenau Gwent, Monmouthshire, Newport, Torfaen and the eastern half of Caerphilly county.
This area incorporated natural water boundaries of the Wye river in the east, Loughor estuary
in the west, the northern Usk river valley and the Brecon Beacons, and the Bristol Channel
and Severn Estuary in the south. Natural boundaries, whether mountainous terrain or river
systems, are common landmarks between territories. 50 For example, territorial divisions in
the Sussex Downs, Chilterns, and Wiltshire were closely related to river valleys, 51 while a
similar situation may have also existed, albeit on a smaller level, in nearby Herefordshire. 52
These boundaries are more noticable, as will be shown below, in settlement patterns creating
clusters across the Welsh Marches,53 and thus can be more easily suggested for South Wales.
It must be mentioned that modern boundaries obviously had no effect during the Iron
Age, and looking for evidence of a tribal community bounded by those limits is dangerously
narrow-sighted. Contact between the Silures and neighbouring tribes is apparent, and culture
can be presumed to have blurred the adjoining landscapes to some degree. When discussing
trade goods that extend across tribal boundaries, cultures can appear similar if exchange
networks have saturated different areas with the same items. In other words, just because a
site has a Dobunnic artefact, does not make it part of Dobunni territory. It is how an item was
or was not used within the cultural system that defines its importance and identity. 54 Amongst
other examples, this will be seen best in the atypical appearances of Dobunnic coins, and the
relative scarcity of Malvern ware within Gwent.

49
Nash-Williams, V.E. (1956). ‘The Roman Town of Venta Silurum and Its Defences’, in Swoboda, E.
(ed.) (1956). Carnuntina: Ergebnisse der Forschung über die Grenzprovinzen des römischen Reiches
Vorträge beim internationalen Kongreß der Altertumsforscher, 1955. Graz: H. Böhlaus Nachf., 105;
RCAHMW (1986), 1; Brewer, R.J. (1993a). Caerwent Roman Town. Cardiff: Cadw: Welsh Historic
Monuments, 2; See below, below.
50
Kimes, T., Haselgrove, C. and Hodder, I. (1982). ‘A Method for the Identification of the Location of
Regional Cultural Boundaries’, Journal of Anthropological Archaeology. 1, 127-8.
51
Cunliffe (1971), 59.
52
Richardson, R.E. (1992). ‘Iron Age and Romano-British Farmland in the Herefordshire Area’,
Transactions of the Woolhope Naturalists’ Field Club. 47, 147.
53
Jackson (1999), 197-216.
54
Stein, G.J. (2002). ‘From Passive Periphery to Active Agents: Emerging Perspectives in the
Archaeology of Interregional Interactions: Archaeology Division Distinguished Lecture AAA Annual
Meeting, Philadelphia, December 5, 1998’, American Anthropologist. New Series, 104(3), 908.
14

Definitive aspects of Silurian culture, independent of Dobunnian, Demetean or


Ordovician cultures, are difficult to distinguish, but evidence can be shown of a predominant
culture within the aforementioned boundaries that can then be distinguished from other
cultures near the natural borders. Therefore using natural boundaries within this landscape as
definite lines of separation between tribes becomes logical. Once again, this cannot limit
looking outside the perceived boundaries for evidence of settlement, but once the tribal
cultural centre within the boundaries becomes more defined, the cultural influence in the
blurred landscapes become easier to recognize. Thus, looking at the area defined during the
Iron Age will provide a better understanding of the Silures’ actual extent of tribal territory.

South Wales
The Wye river and its valleys as a natural eastern boundary of the Silures can be best
seen from the differing Iron Age evidence found on both sides of the river. Jackson sees a
separation in size distribution of hillforts between Gwent and its eastern neighbouring
counties.55 Iron Age hillforts found in Gloucestershire and Herefordshire tend to be on
average larger than those found in Gwent with the majority of sites being over 3 hectares (ha)
in size and relatively few sites under 1.2 ha. By contrast, of hillforts of known size in Gwent,
there are twice as many less than 1.2 ha as those over 3 ha (21 and 10 respectively). 56 Jackson
takes this archaeological evidence along with topographical and land capability analysis and
determines Gwent to be in a different ‘zone’ than Gloucestershire and Herefordshire. 57
Manning concurs citing the prevalence of Herefordshire hillforts being univallate, over 6 ha
in size, and with an entrance guard-chamber. 58 Although univallate hillforts are of the
majority in Gwent, only one is of more than 2 ha in size (Twm Barlwm: 4.2 ha).
Furthermore, only 4% of hillforts in Gwent with known sizes are over 6 ha in size, and only
one of which contains an entrance guard-chamber (Coed-y-Bwynydd Camp); this is the only
example of such an entrance in all of Gwent.59
In addition, regionally produced pottery distribution patterns also show a separation
between cultural territories. The Malverns, located in western Gloucestershire, became a
centre for Iron Age pottery around the fourth century BC. Malvern ware became well
distributed around the Malvern Hills, the Cotswolds and Herefordshire, yet very little remains
can be found within South Wales. With the Wye river as the only boundary between Gwent
and the Malvern Hills, one would expect a relatively fair amount of Malvern sherds, at least at
sites near the river, yet pottery assemblages in South Wales are quite lacking of the regional

55
Jackson (1999), 202. This distinction from those east of the Wye River also includes the southern
Powys, Brecon area: See below, below.
56
Wiggins (2006), 19: table 4.
57
Jackson (1999), 201-6.
58
Manning (1981), 17.
59
Wiggins (2006), 17: table 3; 19: table 4.
15

style.60 A few exceptions to this are found at Lodge Hill Camp 61, Twyn-y-Gaer (Cwmyoy)62
and a single sherd at Thornwell Farm. 63 Lodge Hill Camp had a surprisingly large
assemblage of Malvern ware for a site at such a distance from the Malvern Hills. Yet, as will
become the focus of this thesis, Lodge Hill Camp was one of a select group of prominent
defended enclosures within the South Wales settlement pattern, and such an important site
within the trade network of the eastern territory of the Silures would be expected to contain a
collection of a dominant regional ware. It is in the smaller settlements that the actual
saturation of the pottery within the region can be better attested. At Twyn-y-Gaer (Cwmyoy)
in northern Monmouthshire, a Malvernian ware vessel was found in fourth to early third
century BC context. As will be explained further, this site shows signs of a cultural change
less than a century later into a more southern, Silurian influence, and with this comes Lydney-
Llanmelin style pottery. Before that transition, Twyn-y-Gaer (Cwmyoy) would have been
more culturally Ordovician, which included trade in Malvernian ware. 64 A few other small
finds within the Usk river valley are likely associated simply with the trade connections of
Twyn-y-Gaer (Cwmyoy).65 Thus, although Malvern ware is found within South Wales, the
relative scarcity of the pottery as compared to the other side of the Wye river suggests
different cultural systems at work on either side of the water boundary.
While hillforts and pottery show markedly different lifestyles, coin finds can be less
definitive yet still support the idea of a change in culture at the Wye river. Dobunnic coin
distributions within eastern Gloucestershire show a scattering of finds bearing more silver
than gold, whereas on the west side of the Severn, there are far fewer finds, but mostly of
gold. The small number of finds suggests less participation in the monetary system, while the
frequency of gold implies the Silures placed a higher value on the metal itself. Possibly the
Silures viewed gold as a way of storing or conveying wealth, and although it was exchanged
in other forms, mostly coin types remain. 66 By contrast, Durotrigan territory, south of the

60
Moore, T. (2007). ‘Perceiving Communities: Exchange, Landscapes and Social Networks in the
Later Iron Age of Western Britain’, Oxford Journal of Archaeology. 26(1), 81, 88.
61
Peterson, R., Pollard, J. and Morris, E. (2006). ‘Prehistoric Pottery’, in Pollard, J., Howell, R.,
Chadwick, A. and Leaver, A. (2006). Lodge Hill Camp, Caerleon and the Hillforts of Gwent. Oxford:
British Archaeological Reports British Series 407, 33-37.
62
Probert, L.A. (1976). ‘Twyn-y-Gaer Hill-fort, Gwent: An Interim Assessment’, in Boon, G. C. and
Lewis, J. (eds.) (1976). Welsh Antiquity. Cardiff: National Museum of Wales, 118; See below, page 57.
63
Woodward, A. (1996). ‘The Prehistoric and Native Pottery’, in Hughes, G. (1996). The Excavation
of a Late Prehistoric and Romano-British Settlement at Thornwell Farm, Chepstow, Gwent, 1992.
Oxford: British Archaeological Reports British Series 244, 44.
64
See below, page 18.
65
Webster, P. V. (1993). ‘The Post-Fortress Coarsewares’, in Manning, W.H. (1993). Report on the
Excavations at Usk, 1965-1976: The Roman Pottery. Cardiff: University of Wales Press, 232; Peterson
et al (2006), 35.
66
Sellwood, L. (1984). ‘Tribal Boundaries Viewed from the Perspective of Numismatic Evidence’, in
Cunliffe, B. and Miles, D. (1984). Aspects of the Iron Age in Central Southern Britain. Oxford: Oxford
University Committee for Archaeology. Monograph 2, 197; Green, M. and Howell, R. (2000). A
Pocket Guide: Celtic Wales. Cardiff: University of Wales Press, 42.
16

Dobunni, are known to have operated on a silver monetary system, and as a result gold coins
are scarcely found compared to silver. 67 Therefore the lack of silver found in South Wales and
gold in the Durotrigan area show the contrasting use of the metals when compared to the
Dobunnic territory which encompassed both as forms of currency, giving rise to the Wye
river as a natural boundary from the Dobunnic territory to the east. Davies provides four
maps showing distribution patterns centred upon the Severn Estuary; among which include
Dobunnic coinage, Lydney-Llanmelin ware, Glastonbury ware, and Malvernian ware. 68 The
territorial differences shown by the cultural trade patterns become more apparent when
compared with natural river boundaries.
Another theory suggests the territory between the Severn and Wye as property of an
independent tribe. Ariconium, in the Forest of Dean, has one of the largest collections of
Dobunnic coins found at a single site, suggesting its place as an important trade centre,
although not necessarily connected with either the Silures or Dobunni. 69 Yet more evidence
may point towards the area being of Dobunnic influence. A Roman installation was found
near the site of Ariconium which Walters suggests could have been placed after a possible
treaty between the Dobunni and the Romans.70 He believes the Silures did not control the
area, and the inclusion, if any, of the site into a Roman treaty may have been through an
alliance with the Dobunni, since the tendency of forester communities to be independent and
isolationist implies an autonomous society.
Manning, on the other hand, sees a commonality between Gwent and western
Gloucestershire and Herefordshire.71 He suggests that the Forest of Dean was a part of
Silurian territory which subsequently became separated from the tribe by the territorium of
Caerleon during the Roman conquest. This view, as he admits, puts more weight in coin
distribution than settlement patterns; choosing to believe the coin finds as a Late Iron Age
unifying trait, but the contrary notion as merely a result of differing origins. However, this is
quite dismissive of earlier evidence. Viewing the area just east of the Wye river as possibly
separate from the Dobunni seems logical yet connecting the people within it to the Silures
seems less so. Manning does discuss the theory of an independent tribe, albeit with little
conviction.72 Therefore, an established boundary can be seen in place at the Wye river. This

67
Sellwood (1984), 197.
68
Davies, J.L. (1995). ‘The Early Celts in Wales’, in Green, M.J. (1995). The Celtic World. London:
Routledge, 690: fig. 35.10.
69
Van Ardsell, R. (1994). The Coinage of the Dobunni: Money Supply and Coin Circulation in
Dobunnic Territory. Studies in Celtic Coinage, 1. Oxford: Oxford University Committee for
Archaeology, Monograph 38, 25-6.
70
Walters, B. and Walters, M. (1989). ‘Ariconium Military Site Excavation Interim Report’, Dean
Archaeology. 2, 37-8.
71
Manning (1981), 19-23; Manning (2004), 178.
72
Manning (1981), 21.
17

territorial separation between two cultures differed not only in settlement patterns, but in the
production and exchange of cultural goods as well.
On the western side, the Silurian territory borders the Loughor river, north of the
Gower Peninsula. Around this natural boundary there is only scarce settlement evidence.
The western half of Gower itself is somewhat more densely populated by Iron Age
settlements, while on the eastern side of the peninsula fewer sites have been found. This
could be due to the subsequent modern occupations destroying any evidence over time, but
this could also be explained by a separation of neighbouring cultures and peoples. South-west
Wales differs in landscape from the South as there is substantially less upland territory, most
likely leading to the Demetae establishing different settlement styles, as well as subsistence
strategies, than the Silures, as can be seen in the ringforts of the area.
During the last quarter of the first millennium BC, a sudden change from hillforts to
ringforts takes place in South-west Wales. 73 From then, there is evidence of not only
continual use of ringforts into the Roman conquest period, but also continued construction
after the Roman occupation began. The change to ringforts, essentially defended farmsteads
rather than enclosures, may have been due to an increasing population choosing to fragment
into social groups as a result. This almost complete lack of large hillforts suggests a
settlement pattern is in place unlike that in Gower and South Wales on the other side of the
Loughor Estuary.74 Unfortunately, there has been not as much research done on the Demetae
in south-west Wales or as much evidence found there as in the tribes surrounding the Bristol
Channel, so comparative artefactual evidence cannot be found.

Brecon
Southern Powys, the area that follows the Usk river valley, may comprise the
northern extent of the territory of the Silures. Similarities found between the Brecon area and
Gwent, down the Usk river, may show a cultural connection or at least Silurian influence that
extends beyond the two preserved counties discussed so far. Mixed farming has long
dominated upland regions of Britain, and South Wales is no exception. With the Brecon area
situated mostly above 250 metres in elevation, it can be presupposed that a subsistence pattern
existed there similar to the rest of South Wales. 75 With the familiarity of the Silures to the
pastorally dominated mixed economy, it should be expected that the Silurian culture spread
into the Brecon area as well.

73
Williams, G. (1990). ‘Recent Work on Rural Settlement in South-west Wales’, in Burnham, B.C. and
Davies, J.L. (eds.) (1990). Conquest, Co-existence and Change: Recent Work in Roman Wales.
Lampeter: Trivium 25, 115-6; Davies and Lynch (2000), 157.
74
Davies and Lynch (2000), 218.
75
Briggs, C.S. (1985). ‘Problems of the Early Agricultural Landscape in Upland Wales, as Illustrated
by an Example from the Brecon Beacons’, in Spratt and Burgess (1985), 293-5.
18

Hillfort settlement patterns found within the Usk-Llynfi basin indicate similar models
as those found in Gwent and Glamorgan, where a few large hillforts in central positions are
surrounded by multiple smaller enclosures.76 Slwch Tump is the only large univallate (3.8 ha)
site in area, but the size does not demonstrate the importance of the enclosure as much as its
central location. Just east of modern Brecon town centre, the site holds a strategic position at
the confluence of the Afon Honddu77 and the river Usk. A similar situation exists just west of
Brecon town centre at the medium multivallate Pen-y-crug (1.86 ha), which also overlooks
the same confluence. The close proximity of these suggests different dates or a pairing, each
with distinct functions; the latter of which seems likely through the differing rampart types.
Another paring may exist with the Hillis and Pendre enclosures on opposite sides of the Afon
Llynfi, a tributary running north to the Wye river. Coincidently, the source of the Afon Llynfi
is within two miles of the river Usk, but the two never join.
Intervisibility between hilllforts may provide a plausible connection over the Wye
and Usk valleys in Gwent.78 In Gwent, hillforts along the Usk and Wye river valleys within
view of each other create a series of intervisible watch-points allowing the possibility of a
communication network along the rivers. From a single or even multiple points along the
valleys, a leader or ruling class could have access to the entire valley. This would be
advantageous for anyone hoping to gain control or defend the area as strategically placed
stations could be used to signal messages up or down the river. The Usk river valley
continues north of Gwent into southern Powys turning west; surrounding the northern edge of
the Brecon Beacons. Along the valley, hillforts are aligned facing towards the river. These
hillforts are also within viewpoint of each other and those following the Usk valley towards
the Bristol Channel. Intervisibility between hillforts may show cultural or possibly territorial
connections between sites.79 With the interconnected visibility of the hillforts along the edge
of the Gwent-Powys border apparent, it becomes plausible that at the end of the Iron Age
southern Powys was under Silurian cultural influence.
Sitting on the Fwthog ridge, 426 metres above sea level on the southern edge of the
Black Mountains, Twyn-y-Gaer (Cwmyoy) holds a commanding position overlooking the
surrounding grazing land and valleys. Twyn-y-Gaer (Cwmyoy)’s dominating position is not
only due to the long views over the grassland, but also down the valleys to the east and west
where, respectively, the Honddu river80 joins the Monnow river which meets the Wye river

76
RCAHMW (1986); Jackson (1999), 211. Brecon settlement patterns are discussed more below: see
page 21.
77
Not to be confused with the river Honddu which joins the river Monnow.
78
Thomas, B. (2000). Communication and Lines of Sight: An Interpretation of the Intervisibility of
Hillforts and Related Earthworks of Usk and Lower Wye River Valleys. MA Dissertation in Celto-
Roman Studies: University of Wales, Newport.
79
See below, page 81.
80
Not to be confused with the Afon Honddu which joins the River Usk.
19

before washing into the Bristol channel, and the Grwyne Fawr river joins the Grwyne Fachen
to meet the Usk river also eventually flowing into the Bristol Channel. This gives the hillfort
importance in its ability to watch over and possibly control the northern ends of these
waterways and the south edge of the Black Mountains. Twyn-y-Gaer (Cwmyoy) has also
been linked with Pen Twyn immediately to the east and a possible cultural ‘buffer zone’ to
the north-east, overall making the site very important, especially if placed near the boundary
lines of two tribes.81
As has been suggested, the modern Brecon area was most likely within Ordovician
territory during the Early Iron Age. The archaeology of Twyn-y-Gaer (Cwmyoy) suggests
the site was not originally Silurian (probably Ordovician, but possibly Deceangli), but became
so later, possibly through extreme circumstances such as war. Excavations began on the site
in 1965 in hopes of better assessing the forts of the Marches and the Bristol Channel and the
relations between them.82 Probert identifies six periods (I through VI) at the site, with a
drastic change occurring between Periods IV and V. Archaeological evidence most definitive
of Central Wales comes in Period II where the East Gate parallels similar designs at Croft
Ambrey and Midsummer Hill in Herefordshire, and Period III when Malvern ware is found.
As mentioned above, Malvern ware is not commonly found in South Wales, save Twyn-y-
Gaer (Cwmyoy) and Lodge Hill Camp. In Period V, the East Gate was extended and rebuilt
to provide better support, a style unseen in the previous gate forms. In the last two periods of
occupation, Malvern stamped ware remains cease to be found, replaced by Lydney-Llanmelin
ware which is not found in the Central Marches prior to this time. Typically Silurian beehive
rotary querns also appear as well as more ironwork, which fits in with a Silurian culture that
will be shown to have been increasingly producing local stone, metal and ceramic work. 83
Probert theorizes ‘a cultural, if not political, extension of Silurian power into this district.’ 84
A sword, a sword hilt guard and a spearhead, all iron, which date to the third or second
centuries BC were excavated at the site;85 contemporary with the change in the East Gate and
the beginning of the Silurian influence. These weapons could be unrelated to the Silures, or
conversely they could be remnants of a battle, and subsequent refortification of the gate, or
even an offering by the victorious leader of the Silures. How this change came about is still
unknown, but with a prevailing warrior elite class that is evident in the Silurian archaeological
record, explained further below,86 a switch of such a drastic nature coupled with the obvious
strategic advantages of the site would seem to suggest a warlike conflict.

81
Olding, F. (2000). The Prehistoric Landscapes of the Eastern Black Mountains. Oxford: British
Archaeological Reports British Series 297, 63.
82
Probert (1976).
83
See below, below.
84
Probert (1976), 118.
85
Gwilt (2007), 312.
86
See below, below.
20

Whatever the circumstances behind the change to Silurian cultural dominance at


Twyn-y-Gaer (Cwmyoy), it should not be hard to perceive an extension of such an influence
into the modern Brecon area. The scarcity of middle to late Iron Age pottery from the
Malverns in the Brecon area itself suggests a different social structure. 87 Since this same
scarcity is seen in Gwent at the same time, a correlation between the areas cannot be
dismissed. Controlling the entrance into the Usk river valley which cuts through the Brecon
Beacons then allows for further penetration up river. With the Usk river curving westward, a
convenient northern boundary exists closing the Brecon Beacons into the South Wales region.
A practical and even religious connection to these uplands may have existed through their
relationship with their own uplands, and thus led to a desire to secure the area from northern
influence. No matter the reason, evidence points towards this area of southern Powys as part
of Silurian territory by the end of the Iron Age; more likely beginning the cultural transition
by the third century BC. This position will be more apparent when the actions of the Romans
and their network of forts are discussed.88
Manning postulates that the northern boundary of the Silures was between the Teme
and Wye rivers. However this is through the identification of an Ordovician settlement on the
other side of the Teme.89 As has been shown, the Brecon area most likely underwent a
cultural change in the later Iron Age, which may support a much more northern limit up to the
Teme. This can only be speculation though, as archaeological evidence has yet to provide
Silurian cultural material that far north, and settlement patterns suggest otherwise. Therefore,
this thesis will suggest the Usk river Valley and nearby upland areas as the northern limit of
Silurian cultural influence.
Tribal boundaries can be difficult to distinguish if there are no obvious markers, and
if interregional interaction results in similar, shared cultural material. Evidence along the
Wye river indicates a tribal boundary as Dobunnic coin distribution becomes quite limited
with a preference to gold, as well as a paucity of Malvern ware. This demonstrates that the
Silures had dissimilar uses for the gold coins from their neighbours in addition to a lack of
need or desire for pottery from the Malverns. Out west, less of an archaeological record
makes distinction more difficult to deduce, but a look at settlement patterns, especially a
prominent lack of a pattern around Loughor, provides a noticeable boundary separating the
Demetae from the western Silures. Southern Powys follows similar settlement patterns as its
southern neighbours, as well as comparable artefacts from the third century BC. This leads
the Brecon area as most likely the northern boundary of the Silurian territory. Evidence
demonstrating the Silurian territory has focused upon the material cultural record, and for the
most part, this is all that is definitely known about the Iron Age tribe. The tribal boundary
87
Jackson (1999), 212-3.
88
See below, page 91.
89
Manning (2004), 178.
21

suggested herein constitutes the area in which the native population maintained a shared
cultural identity. The following discussions will suggest this cultural identity transcended
various social and political settlement patterns, and ultimately united the tribal population in
defence during the Roman invasion. Continuing research in Iron Age South Wales is
providing more knowledge about the inhabitants of Gwent, Glamorgan, and the Brecon area,
giving a clearer understanding of the tribe’s territory and cultural influence.
22

Chapter 3: Silurian Settlement Patterns


In the 2006 project conducted by the Glamorgan Gwent Archaeological Trust ninety-
nine sites were identified in the area of Gwent as ‘certain, probable, and possible prehistoric
defended enclosures’ (Map 3).90 Particular aspects of each site (entrance type, location
landscape, ramparts, shape and size in hectares) were ascertained, if possible, and placed into
the database. Using characteristics of each site, interpretations can be made about the
settlement pattern of the Silures in the area. Most prehistoric defended enclosures
documented in Gwent lie within modern areas of Monmouthshire and Newport; in fact only
six can be found outside these areas. Two sites are in the modern Caerphilly area, one of
which is close to the border of Torfaen which has only one documented site, while three sites
lie on the north edge of Blaenau Gwent above 250 metres. There is a relative lack of
evidence in southern Newport within the Gwent Levels that can be presumed to be due to the
coastal wet landscape which is susceptible to flooding and can easily destroy archaeological
remains over time. The Gwent Levels will be shown to have been quite important to the
Silures. Once again, the idea of the county of Gwent is irrelevant when considering
prehistoric settlement, but some distinctions will be shown that separate this area from the
sites contained in the preserved county of Glamorgan.
In the Glamorgan area, GGAT has listed 216 defended enclosures (Map 3). 91 The
same characteristics (entrance type, location landscape, ramparts, shape, and size in hectares)
were documented and can be once again used to deduce settlement patterns in the area.
Similar to sites found in Gwent, the vast majority of documented enclosures in Glamorgan are
in lowland areas of under 250 metres. Thus much of the work in the area will be focused
upon the modern coastal districts of Bridgend, Neath Port Talbot, Swansea, and Vale of
Glamorgan. There are only four documented sites within the Glamorgan half of the
Caerphilly county borough, and five listed in Cardiff, despite the latter being a lowland
county. This is probably due to the modern city and its encompassing settlement destroying
any evidence of prehistoric occupation. There is a relative lack of sites within the modern
Swansea city boundaries for probably the same reason. Myrthyr Tydfil and Rhondda Cynon
Taff also contain little in the way of prehistoric defended enclosures, five and nine
respectively, but they remain important due to many of the sites’ placements above 250
metres.
The Brecon area discussed herein includes only that within the valley constituted by
the Usk river and its tributaries. This valley is surrounded by uplands and mountains from the
Brecon Beacons and Black Mountains in the south and the east and Mynydd Epynt in the
90
Wiggins (2006).
91
Gerrard et al (2006).
23

north. Settlement patterns in the area follow the river valleys, as expected, and the higher
elevations, which slope into the Usk river valley, are dominated by larger hillforts in the
lowlands. Of the hillforts listed in the RCAHMW and CPAT inventories of Brecknock 92,
sixty are within the mentioned region (Map 3). From these, one site, although listed as a
hillfort, is doubtful as such, while two have been lost to subsequent quarrying. Thus fifty-
seven hillforts will be discussed below as evidence of a comparable settlement pattern as that
found in South Wales, which suggests a similar lifestyle, if not a cultural connection. As
mentioned, evidence towards a culture parallel to the Silures in the Brecon area originates at
Twyn-y-Gaer (Cymyoy) in the Middle Iron Age, therefore some hillforts referred to below
may have been in disuse by that period. However, not enough evidence is known to date and
exclude most of the defended enclosures around Brecon, and for that reason they will be
included in the discussion.

Map 3: Definite, probable and possible settlements in South Wales.


©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved. Settlement locations taken from Wiggins (2006),
Gerrard et al (2006), RCAHMW (1986) and Silvester (2007). © Annotations Copyright GGAT/Cadw

Therefore in South Wales, evidence provided by GGAT, CPAT and RCAHMW


shows a settlement pattern that, although varying slightly due to regional needs, is consistent
throughout the area. This suggests a likely connection amongst those living in South Wales.
The evidence presented here of settlement patterns in South Wales has been collected by
others through separate studies of Gwent, Glamorgan and Brecon. Although at times
comparisons between these areas prove useful, this in no way suggests the Iron Age

92
RCAHMW (1986), 8-124; Silvester (2007).
24

boundaries were as such. Once the entire settlement record of South Wales is presented, a
further study of the evidence will be exhibited collectively, and smaller groupings within the
territory of the Silures will be stressed. Locations identified by the GGAT, CPAT and
RCAHMW have been comprised into a single list presented below in Appendix 1comprising
the most up to date inventory of Iron Age defended enclosures in South Wales. Within the
following sections, certain characteristics will be stressed which highlight larger, more
strongly defended enclosures having more prominent roles in the South Wales settlement
pattern. Subsequent chapters will further this point.

Enclosure Shape
Of the physical aspects noted in the GGAT and CPAT reports, the most obvious
commonalities are found in enclosure shape, rampart type and landscape location. All three
will be discussed, in addition to internal size and structures, and presented as evidence of
patterns and connections within South Wales which may be part of local and regional
settlement structures. With regard to shape, sites classified as sub-circular or sub-rectangular
are most often found; however some sites are also classified as strictly circular or rectangular.
The differences between classifications of circular and sub-circular, along with rectangular
and sub-rectangular, exist only in how closely the structures conform to a strict circular or
rectangular shape. The classification of a site into a ‘sub’ category is highly relative, and can
be effected by either natural degradation over time or differing opinions from those
classifying the structure. As well as this, it should be expected that structures were built
shaped slightly different whether in length, width, or roundness of corners, but were treated
similarly by the inhabitants. Therefore, for the sake of this thesis and ease of comparison,
sub-circular and sub-rectangular enclosures will be grouped with circular and rectangular
structures respectively.
Of the ninety-nine sites documented by GGAT in Gwent, eighty-seven sites still
contain defined shapes (twelve site shapes are unknown), forty-four are circular and thirty-
two are rectangular. With only eleven sites of differing shapes (five complex, three
triangular, two polygonal, and one irregular), these two shapes are by far the most common,
and therefore can show a trend in tribal enclosure construction in the area. Enclosure shapes
are found for 201 (93%) of the 216 sites in Glamorgan with the majority, 122 sites (61%),
being circular. There are forty-six examples (23%) of rectangular enclosures, making them
the second most commonly found in Glamorgan. Although significantly less than circular
sites, rectangular shaped enclosures still account for more than all other documented
enclosure types (twenty-one irregular, six triangular, five polygonal, one complex). Thus, just
as eastern South Wales, circular and rectangular typed enclosures show a trend in preference
among settlement patterns. In the Brecon area, circular enclosures dominate with thirty-eight
25

(67%) of the fifty-seven examples, while rectangular enclosures are only found in ten
instances. Five triangular enclosures can also be found, while four sites shapes remain
unknown. Therefore overall, circular and rectangular shaped defended enclosures dominate
the Siluran settlement pattern which may indicate shared construction ideas, cultural
connections or settlement purpose.
Rampart shapes are often found in clusters showing more local preferences towards
one or a few shapes over others. This preference will be shown as often affected by the
landscape. However, other factors can influence a settlement type, such as tradition, social
standing and enclosure purpose.93 Similar aspects could have had bearing upon buildings
within enclosures as well.94 The majority of sites documented in eastern South Wales are in
the lowland areas comprising the central and south-central regions. Here can be found sixty-
seven (all but nine) of the sites classified as rectangular and circular. Circular sites can be
found throughout the lowlands with rectangular sites found mostly in the south-central area
north of the Gwent Levels. Within Glamorgan, preference for circular-type enclosures varies
regionally. West and Mid Glamorgan exhibit a circular based settlement pattern with thirty-
three and fifteen examples, respectively. South Glamorgan, closest to Gwent, only contains
eleven circular enclosures. Where the circular enclosures become less frequent, rectangular
enclosures appear more often. In the south of Glamorgan, eleven are of rectangular shape,
equal to the number of circular enclosures. West and Mid Glamorgan, conversely, only have
three instances of rectangular-type enclosures each. More prevalent is the number of
enclosures not of circular or rectangular shape. South Glamorgan circular and rectangular
enclosures account for 79% of all enclosures in the region, and Mid Glamorgan: 72%. West
Glamorgan shows the two enclosure types accounting for only 68% of the fifty-three known
sites in the area. This is due to a larger number of irregularly shaped enclosures on coastal
positions in the peninsular region. It must be noted that West Glamorgan has more than twice
as many irregular coastal sites as South Glamorgan, while none are found in Mid Glamorgan.
Therefore, individual settlement patterns may have existed within South Wales with different
preferences based upon and influenced by geographical location.

Enclosure Defences
To further promote the settlement landscape distinction of South Wales, an
examination of rampart types is necessary. Elaborate defences often suggest that an elite
aristocracy is present, and this idea will be stressed throughout this dissertation. 95
Multivallation, as opposed to univallation, has been viewed as providing a two-fold purpose:
increasing the degree of social isolation, and enhancing the prestige of the inhabitants and the

93
See below, page 43
94
See below, page 33.
95
Cunliffe (2005), 294.
26

enclosure itself.96 This latter aspect is especially true with stone-built ramparts as the large
labour supply needed for construction may be seen as a symbol of authority. Of the ninety-
nine sites noted in Gwent, sixty-two are univallate, containing only one line of defence.
Twenty-two sites have bivallate defences composed of two lines of banks and ditches, while
fifteen are multivallate enclosures with three or more lines of defence. The overwhelming
majority, almost two-thirds of all sites, with univallate ramparts shows a preference by the
natives to only single defences. This is probably due to most of the sites with only univallate
defences being of small single or extended family settlements with little need for strong
defences. This corresponds with a logical view that the complexity of defences corresponds
to the importance of the site, thus the monumental or symbolic areas need more security. 97
However, assumptions must be made that in some instances outer lines of defence have been
lost either to restructuring during settlement or over time due to human involvement such as
ploughing for agricultural use. An example of this may be found at Gaer hillfort on the south-
eastern edge of the Brecon Beacons where a univallate enclosure is found, but trace evidence
in the area may suggest a second enclosure boundary.98
Rampart types in Glamorgan also tend to show a similar pattern as is in the east,
although less complete archaeologically. Only 147 (68%) of the 216 sites in Glamorgan have
known ramparts compared with all documented sites in Gwent having a recorded rampart
type. Nevertheless, enough can be interpreted from what is known to give a clear
understanding of what happened in each area. Univallate enclosures still encompass the
majority of sites (57% of 147), yet higher percentages of more strongly defended sites are
found in Glamorgan with thirty-seven bivallate (25%) and twenty-seven multivallate (18%)
sites. Once again a tendency towards more protection at a limited number of sites shows
preferential treatment placed upon those sites.
Brecon enclosures are found most often with single, univallate defences (31
examples), while ten bivallate and eleven multivallate defended enclosures suggest a similar
pattern as in the rest of South Wales where certain sites gain importance and are locally
prominent.

Settlement Location
Where the defended enclosures are placed within the eastern South Wales landscape
of Gwent can also reveal more about the settlement pattern itself. Preference by the natives
was found in placing sites in raised elevation, with twenty-seven sites on hillslopes, twenty-
five on hilltops and fifteen on ridges. Other settlement sites varied greatly amongst eleven

96
Bowden, M. and McOmish, D. (1987). ‘The Required Barrier’, Scottish Archaeological Review. 4(2),
77-8.
97
Davies and Lynch (2000), 158.
98
Wiggins (2006), 13.
27

location types, but were no more commonly found than the six examples in a level position.
Four sites had unknown locations. Reasoning for the preference towards a raised elevation
can be implied from the obvious advantages awarded to higher ground: an increase in range
of visibility of the surrounding area and better defensive position on any would-be attackers
deterring such an event.
With known topographic locations at 213 of the 216 sites in Glamorgan, a similar
pattern emerges with hillslopes being the preferred location at sixty-six sites (31%). Hilltops
(thirty-two sites; 15%), however, are not as equally valued as hillslopes in Gwent; in fact,
topographical preference is given more so to level ground at thirty-five sites (16%).
Enclosures are still to be found on ridges (nineteen sites) as in Gwent, but with less frequency
(9%). Western South Wales stands out most in the number of coastal locations: fifteen sites
(7%) are situated on coastal promontories and eleven (5%) on coastal cliffs. Although these
are not overwhelming amounts, they do represent a different preference and use of coastal
locations than elsewhere in Silurian territory as promontory sites have been viewed as large-
scale community projects where substantial areas are enclosed. 99 Gower contains the majority
of promontory sited hillforts, often on cliff edges, due to the peninsula’s frequent high coastal
elevation. These placements are highly defended by the natural rock edge, creating vastly
different strategic situations for communities than those which are found at a lower elevation.
In the Usk river valley near Brecon, hillslope locations are found at 21 sites (37%)
out of fifty-seven settlements. Twelve can be found on promontories, while nine are on
hilltops, eight on ridges and six are located on spurs (1 is unknown). Hillslope locations are
the most prominent in South Wales, and although promontory sites are regular in Brecon,
hilltops are also frequent, paralleling the most common occurrences in the Gwent and
Glamorgan.

Internal Sizes
While looking at site shape, defence and location are useful in picking out a pattern
amongst the data. Distinguishing the sites further by size creates a better understanding of the
possible use of each site in the tribal landscape. Where a location may be viewed one way
due to external characteristics, it may be perceived differently when looking internally.
The internal sizes of the defended enclosures in Figure 1: Prehistoric Defended
South Wales demonstrate a preference for smaller Enclosures in Gwent in Order of
Known Size (size in ha)
enclosures, but usually with at least one hillfort of Twyn Bell Camp 0.06
a larger size nearby. This may suggest single Bishop Barnet’s Wood Camp 0.08
Twyn-yr-Allt 0.1
family units within smaller enclosures with some St. Julians Wood Camp 0.15
Tregare 0.2
type of relationship, possibly social or political, Newhouse (Llantrisant) 0.2

99
Davies and Lynch (2000), 148.
28

connecting the site with the larger hillfort. This Pierce Wood Camps 0.2
Castell Prin 0.22
will be discussed further in the next chapter in its Hendre Hafaidd 0.25
Gaer 0.3
possible correlation with social interaction, from Camp Hill 0.33
Coed-y-Caerau (Langstone) 0.4
reflecting the size of the settlement group to the
Pen-Toppen-Ash Camp 0.45
level of prestige desired by the inhabitants. Blackfield Wood Camp 0.46
Gaer Hill Camp 0.5
Research into defended enclosures is often Caerau 0.7
Y Gaer (Graig-Y-Saeson) 1.0
supplemented by classifications to simplify Golden Hill Earthwork 1.0
groupings of internal sizes. A 1962 Ordnance Coed-y-Caerau Roman Fort 1.0
Wilcrick Hill Camp 1.1
Survey publication created only two distinctions: Rhiwderin Camp 1.19
The Larches Camp 1.24
small and large hillforts split at 1.2 ha.100 For the Ysgyryd (Skirrid) Fawr Hillfort 1.4
White Castle Farm/Tredam 1.5
most part, this separation sits at the high end of
Enclosure
‘small’ enclosure groupings. However, as more The Mount (Pen-y-Lan) 1.56
Cae-Camp 1.59
work has been done on hillforts and defended Gwernesney Defended Enclosure 1.6
(Usk)
enclosures, further division has proved necessary. Llancayo Camp 2.0
A RCAHMW inventory of Glamorgan in 1976 The Bulwarks 2.06
Great House Camp 2.37
kept the size division, and chose to form seven Twyn-y-Gaer (Cwmyoy) 2.98
Sudbrook Camp 3.21
classes centred on rampart types. 101 Work done Gaer Fawr 3.5
Gaer 3.5
by Jackson on the Welsh Marches also followed
Priory Wood Camp 3.87
Ordnance Survey guidelines, with the addition of Pen Twyn 4.17
Twm Barlwm 4.2
‘medium’ (1.3 to 3 ha) and ‘very large’ (over 6 Coed-y-Bwnydd Camp 6.39
Llanmelin Wood Camps 6.56
ha) classifications. The ‘medium’ classification Tredegar Hillfort 9.28
was to compensate for a general absence of Lodge Wood 9.58
Figure 1: Wiggins (2006), 19: table 4.
hillforts around the 3 ha size in the study, which
included the Usk valley of
Gwent and Brecon.102 This class addition proves that simple standards cannot be applied to
all regions. Hogg followed this, asserting a different scale for Wales with a ‘small’ size
division capped at 0.7 ha, ‘medium’ up to 2 ha, and anything above: ‘large’. 103 Although
established many decades ago, Hogg’s classification will prove to still be the most useful for
South Wales. With these methods in consideration, a scale will be employed using Hogg’s
division of smaller enclosures supplemented by Jackson’s approach to ‘very large’ sites.
Therefore in this dissertation: ‘small’ enclosures are less than and including 0.7 ha, ‘medium’

100
Ordnance Survey. (1962). Map of Southern Britain in the Iron Age, 1:625,000. Chessington, Surrey:
Ordnance Survey, 13.
101
RCAHMW (1976b), 7.
102
Jackson (1999), 200.
103
Hogg, A.H.A. (1972). ‘The Size-Distribution of Hillforts in Wales and the Marches’, in Lynch, F.
and Burgess, C. (eds.) (1972). Prehistoric Man in Wales and the West: Essays in Honour of Lily F.
Chitty. Bath: Adams and Dart, 295.
29

are between 0.7 ha and 2 ha, ‘large’ enclosures are from 2 ha to 6 ha, and ‘very large’ are
greater than 6 ha in internal size.
In the east, forty-one of the ninety-nine sites listed are complete enough to provide an
approximate size (Figure 1). Although this is not a large enough number of sites to give a
more complete picture, ideas can still be deduced from the information and then placed into
the larger data set. The smallest site size found was 0.06 hectares (ha) with the largest being
9.58 ha. To test the divisions decided upon above, these forty-one sites in Gwent will be
placed in the previous conventional divisions used by Ordnance Survey and RCAHMW: a
separation at 1.2 ha. In doing this a ‘small’ group of twenty-one enclosures, and a ‘large’
group of twenty are found, effectively splitting the total data set in half. However, this
division separates two sites only 0.05 ha in size apart: Rhiwderin Camp (1.19 ha) and The
Larches Camp (1.24 ha). If size is a determinant, one would expect these two sites to be of
similar settlement use. If larger size determines different usage, then a more distinct
separation between size groups would be expected. This could take place at a few points. If
The Larches Camp could be considered ‘small’ then the closest ‘large’ site would be Ysgyryd
(Skirrid) Fawr Hillfort (1.40 ha): a 0.16 ha difference. As well, a 0.4 ha separation exists
between Gwernesney Defended Enclosure, Usk (1.6 ha) and Llancayo Camp (2 ha). This
latter separation is not only a stronger dividing line, but corresponds perfectly with Hogg’s
division scale. Thus Hogg’s dividing lines, fit well in the South Wales archaeological
evidence.
With the idea that larger sites assume more important, complex functions and smaller
sites being simpler,104 a separation following Hogg’s scale was created at two hectares.
Splitting the database at this size creates a division of sixteen ‘small’ sites (39%) of 0.7 ha or
below, eleven ‘medium’ sites (27%), and fourteen sites of ‘large’ size (34%). From this
patterns are already beginning to emerge.
Looking at these sites with known sizes of two hectares or less (‘small’ and ‘medium’
sizes), distinctions can be immediately found. Among twenty-eight sites which are within
these limitations, half are found with single, univallate enclosures. By contrast, the thirteen
sites (32% of forty-one with known sizes) over two hectares in the area include only one
univallate enclosure. Furthermore, through obvious inferring of more important sites needing
more complex defences, only three sites (11%) below two hectares have multivallate ramparts
while eight (62%) of the thirteen larger sites have such defences. Amongst this ‘large’ group
is included the largest sites with known sizes in Gwent (and all of South Wales): Lodge Wood
and Tredegar Hillfort at 9.58 and 9.28 ha respectively. The next two sites are of smaller sizes
yet are distinctively larger (over 2 hectares larger) than the rest: Llanmelin Wood at 6.56 ha
and Coed-y-Bwyndd Camp at 6.39 ha. This can only suggest that these four larger sites
104
Davies and Lynch (2000), 148.
30

served more complex purposes since a certain amount of space would be required for each
function. As it so happens, these four sites can be grouped separately into Jackson’s ‘very
large’ class, demonstrating firmly the need for this highest division.
Following a 0.7 ha division creates a 0.3 ha partition from the next three hillforts of 1
ha size; leaving sixteen ‘small’ sites containing nine univallate, seven bivallate, and no
multivallate defences. Incorporating Hogg’s approach makes a convenient initial break to
class those greater than 0.7 ha and less than 2 ha as ‘medium’: with four univallate, four
bivallate, and three multivallate defended enclosures. Further dividing the enclosures by size
is best fitting between Twm Barlwm (4.2 ha) and Coed-y-Bwnydd Camp (6.39 ha): ‘large’
and ‘very large’ respectively. ‘Large’ sites include only two univallate, four bivallate, and
four multivallate, while all four ‘very large’ sites are multivallate defended. The ‘very large’
category contains two groupings too obvious to ignore: Coed-y-Bwnydd Camp (6.39 ha) and
Llanmelin Wood Camps (6.56 ha) along with Tredegar Hillfort (9.28 ha) and Lodge Wood
Hillfort (9.58 ha).
Twm Barlwm, the largest (4.2 ha) of the ‘large’ sites, and Llancayo Camp, the
smallest (2 ha) of the ‘large’, both are of a univallate rampart type while all others above 2 ha
are bivallate or multivallate. This seems to contradict any correlation between greater size
and defence as indication of site importance. However, an examination at the location of each
site may provide understanding for the lack of defences. Llancayo Camp may not have been
considered by the Silures as similar in importance as others in the ‘large’ group, therefore size
may be misleading. Furthermore, close proximity to Coed-y-Bwnydd Camp (6.39 ha), the
largest site in the northern half of Gwent, may downplay the significance of Llancayo Camp.
With such a large site nearby, protection may have been less essential to the inhabitants of the
hillfort, which may have been of such a large size only due to a larger than normal group of
close or extended families. Twm Barlwm is situated on the western edge of Gwent, away
from most sites of the area. Situated on a ridge, natural defence may have superseded the
need for multiple ramparts. As well as this, the outlying nature of the hillfort away from most
other sites suggests the large size may have served another unknown purpose, and was
possibly connected with the nearby Lodge Wood, only twenty-two kilometres away.
As with sites in Gwent, an examination of shape, ramparts and location in western
South Wales is best supplemented by those with known internal area sizes (Figure 2).
Unfortunately, only half (108) of all the sites in Glamorgan have known internal sizes, yet this
still provides a much better perspective than that in the east. With such a larger data set,

Figure 2: Prehistoric Defended Enclosures in Glamorgan in Order of Known Size (size in ha)
Encl. In Ilston Parish; E. Of Parkmill 0.05 Pen-y-Castell, Kenfig Hill 0.4
Mynydd Pen-hydd (Nant Herbert Ringwork) 0.05 High Pennard Promontory Fort 0.4
Blue Pool Bay 0.07 Ty’n-y-Waun 0.4
The Knave Promontory Fort 0.1 Llantrithyd Camp, Llancarfan 0.4
31

450m Southwest of Llandewi Church 0.1 Pantypyllau: Parcnewydd 0.4


Yellow Top Promontory Fort 0.1 Nash Point Camp 0.4
Glyn-Neath Fort 0.1 Danish Fort; Sully Island 0.4
Hen Gastell (Dan-y-Lan Camp) 0.1 Castle Field Camp, Graig-Llwyn 0.4
Gron Gaer 0.1 Llwynda’ddu/Llwynda-Ddu Camp 0.4
Crawley Rocks Promontory Fort, Nicholson 0.1 Geulan Las 0.4
Burrows
Bishopston Valley Promontory Fort 0.1 Caer Cwmphilip or Moel Ton Mawr 0.4
Llanfythin Camp; Llanvithyn Camp 0.1 New Mill Farm, Monknash 0.4
Carn Caca 0.1 Flemings Down 0.45
Pencaerau 0.1 Graig Fawr 0.5
Blaen Cwm Bach 0.1 Earthwork near Fforest Newydd 0.5
Mynydd-y-Gaer (Gaer Fawr) Lower Camp 0.1 Lle’r Gaer 0.5
Cwm Ffairty 0.1 Rectilinear Enclosure: Kenson South 0.5
Cefn-yr-Argoed 0.1 Coed-y-Cymdda 0.6
Caer Blaen-y-Cwm 0.1 Hardings Down West Camp 0.6
Danish Camp; Cwm Philip West 0.1 Worms Head Promontory Fort 0.6
Concentric Ditched Fort; Kenson Wood 0.1 Parc Coed Machen 0.6
Unknown Name 0.12 Maiden Castle Promontory Fort, Oxwich 0.6
Point
Oval Enclosure, Tredogan NW 0.15 Llanbethery 0.6
Cefn Cilsanws 0.16 Carn Nicholas 0.6
Unknown Name 0.18 Old Castle Promontory Fort; Rhossili Cliffs 0.7
Druids Moor 0.2 Ogmore 0.8
Hardings Down Lower Camp; North Camp 0.2 Castell Moel 0.8
Maendy Camp 0.2 Warren Hill 0.8
Stembridge Camp 0.2 Castell Morlais 0.8
Reynoldston Camp 0.2 Hardings Down East 0.9
Bryn Sil 0.2 The Bulwark, Llanmadoc Hill 0.9
St. Mary Hill Down 0.2 Craig Tan-y-Lan 0.9
Cilonnen 0.2 Pen-y-Gaer 0.9
Cae Summerhouse Camp 0.2 Llanquian Wood 1.0
Redley or Caswell Camp 0.2 Y Gaer; Bonvilston Gaer 1.0
Gwersyll Enclosure 0.2 Mynydd-y-Gaer Hillfort; Coed Cae Gaer 1.0
Wenallt Camp 0.2 Buarth-y-Gaer 1.1
Craig-y-Parc 0.2 Burry Holms Hillfort 1.2
Craig Ty Isaf 0.2 Mynydd Twmpathyddaear 1.2
Pen-y-Castell, Cwmavon 0.2 Summerhouse Camp 1.3
Half Moon Camp 1 0.2 North Hill Tor; Nottle Tor Camp 1.85
Llwynheiernin 0.25 Craig Ruperra 2.0
Lewes Castle Promontory Fort 0.3 Coedymwstwr 2.5
Thurba Head Camp, Promontory Fort 0.3 Castle Ditches, Llantwit Major 2.5
Mynydd Bychan 0.3 Craig-y-Dinas 2.52
Mynydd-y-Fforest 0.3 Mynydd-y-Castell 2.7
Cwm Bach 0.3 Cil Ifor Top 2.9
Coed-y-Gaer, Homestead Enclosure 0.3 Caerau, Llantrisant 3.6
Whitton Lodge Roman Villa 0.3 Caer Dynnaf 3.8
Camp 530m E. of Tonmawr 0.3 Porthkerry Bulwarks 4.1
Fonmon Castle Wood 0.3 Castle Ditches (Llancarfan) 4.2
Tor-Gro 0.4 Caerau Camp, Ely 5.1
Horse Cliff Promontory Fort 0.4 Dunraven 6.5
Berry Wood 0.4 Y Bwlwarcau 7.2
Figure 2: Gerrard et al (2006), 34-6: Table 5.

separating the sites by size was necessary; however, with ninety-six sites (89%) of two
hectares in size or less, a smaller division was needed. Since separating the sites at one
hectare would only subtract six from the group, a further partition at 0.7 ha, once again
following Hogg’s assertion, seemed best for initial understanding. Of the 108 sites with
known sizes, seventy-nine (73%) are ‘small’ (0.7 ha or lower), representing a majority of the
32

group. Once again, univallate ramparts provide Figure 3: Prehistoric Defended


the preponderance (forty-five sites; 58%) of Enclosures in Brecon in Order of
Known Size (size in ha)
defences amongst the smaller enclosures. This Rhyd Uchaf 0.01
percentage is maintained when including Blaencamlais 0.03
Pen-y-Garn 0.06
‘medium’ enclosures (up to two hectares) Allt-yr-Esgair (West Enclosure) 0.085
Corn-y-Fan 0.09
showing a consistent settlement pattern (59% Mynydd Llangorse (Cockit Hill) 0.09
Llwyn Bedw 0.1
under 2 ha are univallate). Yet, amongst 0.7 ha or
Caeau 0.13
less in size, Glamorgan shows more use of Pantywenallt 0.18
Coed-y-Brenhin 0.2
stronger defences with nineteen bivallate (24%) Lan Fawr 0.22
Pen-yr-Allt 0.23
and thirteen multivallate (17%) sites, the latter of Ffinant Isaf 0.24
which including three instances at sites 0.1 ha in Nant Tarthwynni (Enclosure I) 0.26
Coed Cefn 0.29
size and one 0.05 ha. This variation from the Coed Mawr 0.33
Twyn Llechfaen 0.37
norm suggests more importance on security in the Nant Tarthwynni (Enclosure II) 0.38
Twyn-y-Gaer (Llanfihangel Fechan) 0.38
area possibly due to a more independently
Penffawyddog 0.4
orientated settlement culture. Crug-y-Gaer 0.43
Coed Gaer 0.45
Internal sizes are found at forty-nine (The) Gaer (Aberllynfi) 0.45
Twyn-y-Gaer (Trallong) 0.45
(86%) of the fifty-seven sites in the Brecon area Gaer Fach 0.48
(Figure 3). This allows for a far more complete Gaer Fawr 0.48
Coed-y-Gaer 0.57
representation of the northern hillforts than those St. Illtyd’s (Llanilltyd) Churchyard 0.57
Plas-y-Gaer 0.58
in the southern regions. Once again, following Crug Hywel 0.63
Tump Wood 0.63
the same scale detailed above, thirty-six sites are
Dostre 0.65
‘small’, ten are ‘medium’, and only three are Merthyr Cynog 0.66
Llys-wen 0.7
‘large’. There are no ‘very large’ enclosures, Twyn-y-Gaer (Llanspyddid) 0.7
Y Gaer (Twyn-y-Gaer (Defynnog)) 0.7
which may be attributed to the natural landscape Llwyncelyn-fawr 0.75
of the Usk river valley which does not allow for Cross Oak 0.8
Pwll-y-Cwrw 0.8
such extravagantly sized hillforts. Following the Coed-y-Caerau (Llanfrynach) 0.86
Coed Fennifach Camp 0.88
same defence patterns as the rest of South Wales, Coed Pen-twyn 0.89
Tyle Clydach 1.0
twenty-two sites are univallate (61%) amongst the
Castell Dinas 1.03
‘small’ enclosures, and they predictably become Pendre 1.57
Pen-y-crug 1.86
less frequent as internal size increases (4 of 10 Hillis 3.6
Slwch Tump 3.8
‘medium’ sites and 1 of 3 ‘large’ enclosures). Allt-yr-Esgair 5.45
This single univallate ‘large’ enclosure, Slwch Figure 3: RCAHMW (1986), 8-124; Silvester
(2007).
Tump,
holds a promontory position near the Usk river at the bottom of the valley, and as is discussed
below, the univallate defences may point to a communal purpose which would negate the
need for such an elaborate structure.105
105
See below, page 43
33

As one aim of this thesis is to discover and speculate upon the decentralised structure
of the Silures, comparing and contrasting the areas of South Wales will provide a basis for
this. Of the total, unfortunately, only a little over half of the defended enclosures (198 of 372)
discussed herein as evidence towards a Silurian settlement pattern have a known size.
Therefore an overall study of South Wales can only be speculative at best, yet this should not
preclude such a study as a model of the social structure can still be suggested.
In central and western South Wales, 73% of sites with known internal sizes are
‘small’. A similar picture exists in the north (71%), while only 39% are ‘small’ in the east.
Amongst these, univallate defences are present in 58% of enclosures in Glamorgan, 56% in
Gwent, and 61% in the Brecon area. If anything, this shows quite a close comparison
throughout the territory of the Silures; however, 17% of the Glamorgan ‘small’ enclosures
exhibit multivallate ramparts, and 16% in Brecon. This is not found in Gwent where the other
‘small’ sites are only bivallate or univallate, emphasising a greater desire for more ramparts at
enclosures in western and northern South Wales.
Differences in settlement pattern from Gwent may be more noticeable when
contrasting sites above two ha in size. With only twelve such sites in Glamorgan (11% of 106
with known sizes) and three in Brecon (6% of 45 with known sizes) compared to thirteen in
Gwent (32% of 41 with known sizes), obviously there is a necessity for larger sites in the
latter region. Furthermore, three of those twelve larger sites in Glamorgan are found with no
greater than a univallate defence (25%), while only five are multivallate (42%). This
contrasts with Gwent which defends their larger enclosures more heavily, having only one
univallate (8%) and eight multivallate (62%).
Overall, a common settlement structure pattern is found amongst all of South Wales:
a majority of smaller, univallate enclosures along with the larger, more strongly defended
sites. These initial groupings only exist through internal size comparisons, and further study
on the correlation between the sites within South Wales must be undertaken to better discern
the intra- and inter-grouping relationships. As well, a further look into settlement sites and
their usages will hopefully bring about a better understanding of the relationship between
structure size and function, as it has been suggested that hillforts overlooking fertile land are
centres for the surrounding agriculture or pastoral uses along with possibly many more
including social, political, religious or even as a source of refuge during poor economic
times.106

Internal Buildings
As previously mentioned, the perceived status of larger sites as more important in the
cultural and settlement pattern is evident through the archaeological record. Of the thirty sites

106
Cunliffe (1971), 62-3; Jackson (1999), 211; Davies and Lynch (2000), 162; Howell (2009), 36-7.
34

in South Wales which are two hectares in size or larger, twenty-four are of bivallate defences
or larger. This stresses the importance placed on security, yet this importance may become
more apparent when associated structures are considered. While timber built roundhouses are
typical of Later Bronze Age and Iron Age building constructions, rectangular building types
occasionally emerge in certain settings. Rectangular dwellings have been identified
throughout Britain with possible parallels on the continent. 107 In other parts of Wales, timber
built roundhouses and rectangular buildings are found within the same hillfort. 108 Moel
Hiraddug in Denbighshire shows roundhouses and four-post buildings in separate areas of the
fort. Moel-y-Gaer also in Denbighshire has a clear distinction between areas, as roundhouses
are constructed in the central area while four-post buildings are placed near the southern
rampart. Collfryn, in north-eastern Powys, had roundhouses and four-post structures
representing up to twenty different phases of construction, suggesting changing uses of the
site called for altering the structures. 109 These imply a zoning system with buildings required
for varying purposes in different parts of the enclosure, not unlike modern commercial,
industrial, and residential zoning systems. 110 There may have been similar systems involved
in larger hillforts where space is available and multiple purposes are assigned to the
enclosure.
Rectangular buildings have also been associated with a change to privatized
resources.111 Rectilinear structures are more easily adapted to serve specific needs than
circular units, and can be added on to or reduced in size without needing to reconstruct the
entire foundation. Therefore evidence for rectangular internal buildings may suggest
specialized storage of resources or any important good, and the various social, political or
even commercial systems that correspond to such privatization.
Nine sites in the county of Gwent have been found with a structure within an
enclosure.112 Three of these sites contain rectangular structures: Two multivallate enclosures
(Lodge Wood and Ysgyryd (Skirrid) Fawr Hillfort), and a bivallate defence at the other
(Twyn-y-Gaer (Cwmyoy)). Two sites, furthermore, are of over two hectares in size: Twyn-y-
Gaer (Cwmyoy) (2.98 ha) and Lodge Wood (9.58 ha: the largest of those with a known size).
The only exception to the correlation of size and importance is at Ysgyryd (Skirrid) Fawr
Hillfort where a rectangular structure is found within a 1.4 ha enclosure. However, this site’s
importance comes through its placement along with the complex settlements of Twyn-y-Gaer
107
Stanford, S.C. (1971). ‘Invention, Adoption and Imposition – The Evidence of the Hill-forts’, in Hill
and Jesson (1971), 47.
108
Davies and Lynch (2000), 160.
109
Wigley, A. (2007). ‘Rooted to the Spot: The ‘Smaller Enclosures’ of the Later First Millennium BC
in the Central Welsh Marches’, in Haselgrove and Moore (2007), 182.
110
Howell (2009), 27.
111
Flannery, K.V. (2002). ‘The Origins of the Village Revisited: From Nuclear to Extended
Households’, American Antiquity. 67(3), 417-8.
112
Wiggins (2006), 24: table 6.
35

(Cwmyoy) and Pen Twyn in the north of the county. This cluster of hillforts is northern
Gwent lies at the edge of the Black Mountains overlooking the river Honddu where it joins
the river Monnow. Thus, rectangular buildings within an enclosure stress the importance
upon the site amongst others in the area.
Adding to the GGAT report, at Goldcliff, on the shore of the Severn Estuary, eight
unenclosed rectangular timber buildings were uncovered dating from Early Iron Age to early
Roman period.113 Architecturally, these buildings are unique to the area with similarities only
found nearby in Redwick. There, five rectangular buildings of Middle to Late Bronze Age
date were found, possibly giving the architectural style this date of origin. 114 Greenmoor
Arch, north-east of Goldcliff, has three partially excavated Middle Iron Age rectangular
buildings, as well as another sub-rectangular structure excavated at Lodge Wood Camp. 115
These three sites suggest a building tradition in south Gwent possibly related to the trade
network already established in the Gwent Levels by the Later Iron Age.
At Goldcliff, rectangular building construction beginning in the Early Iron Age,
increasing near the end of the Middle Iron Age and continuing into the early Roman Period
suggests a consistent use of the site along the coast. The addition of each building through
the Iron Age, as well as the numerous trackways nearby, all point to the use of the area for a
specialized commercial lifestyle.116 This trend at Goldcliff peaked sometime between 410 and
40 BC, during which the construction of at least four rectangular buildings and four trackways
have been radiocarbon dated.117 If Goldcliff was a trading centre of South Wales, it would
more than account for the steady increase in construction. Dating of this site also corresponds
to the end of the second Gwent Levels inundation in the middle of the fifth century BC. Most
likely once merchants on either side of the Severn recognized the end of high sea levels, trade
flourished and subsequent construction was needed.
As mentioned above, the rectangular buildings are of similar architectural style to
another group at Redwick, also located in the Gwent Levels, 6.3 km to the east; possibly
showing a tradition originating in the Bronze Age. Redwick’s rectangular buildings dated to
the Middle to Late Bronze Age are contemporary with the first inundation phase in the Severn

113
Allen, S.J. and Bell, M. (2000). ‘Rectangular Buildings of the Later 1 st Millennium BC at Goldcliff:
Dating, Character and Reconstruction’, in Bell, M., Caseldine, A. and Neumann, H. (2000). Prehistoric
Intertidal Archaeology in the Welsh Severn Estuary. York: Council for British Archaeology Research
Report 120, 128-35; Gwilt (2007), 301.
114
Bell, M. and Neumann, H. (1999). ‘Intertidal Survey, Assessment and Excavation of a Bronze Age
Site at Redwick, Gwent, 1999’, Archaeology in the Severn Estuary. 10, 25-37; Bell, M. (2001).
‘Interim Report on the Excavation of a Middle Bronze Age Settlement at Redwick 2000-1’,
Archaeology in the Severn Estuary. 12, 99-117; Gwilt (2007), 301.
115
Locock, M. and Yates, A. (1999). ‘Greenmoor Arch, Bishton (ST 400 866)’, Archaeology in Wales.
39, 88; Howell and Pollard (2000), 82.
116
Davies and Lynch (2000), 170.
117
Bell, M and Neumann, H. (1997). ‘Prehistoric Intertidal Archaeology and Environments in the
Severn Estuary, Wales’, World Archaeology. 29(1): Riverine Archaeology, 98: table 1.
36

Estuary beginning around 1,200 BC and lasting into the end of the Bronze Age. 118
Construction at Redwick ends near the beginning of the Iron Age, while the earliest
radiocarbon dating at Goldcliff arises around 130 years later. In addition, the Goldcliff
buildings, although similar, are shorter, wider and better constructed. 119 Plausibly, the
building tradition migrated with a population which chose to abandon Redwick in hopes of
better accommodating the difficulties of a high sea level at Goldcliff. There they evolved the
building style to create more substantial rectangular buildings to outlast the effects of the
climatic deterioration. Although hoof prints at both sites suggest the same pastoral
community,120 if there was construction for intended trade success at Goldcliff, there is no
earlier corresponding evidence at Redwick for the same intentions. Three rectangular
structures in two groups comparable to Goldcliff are found at Greenmoor Arch. 121 There
cattle hoof prints are also found, as at Redwick and Goldcliff, and pottery suggests a Middle
Iron Age origin for the site. Possibly, Greenmoor Arch was built after Goldcliff arose as a
trading centre opportunistically with the lowering sea levels and the material culture evolution
which took place between the end of the Bronze Age and the Late Iron Age.
Evidence has proven that roundhouses and rectangular buildings were built
contemporaneously showing not a trend towards one and away from the other, but a
consistent use of both, probably depending upon the purpose of the building. The choice of
roundhouses as residential may coincide with a symbolic purpose, while rectangular buildings
were ancillary or for specialized ritual use, including those uses already mentioned. 122 In
Danebury and South Cadbury, rectilinear slot buildings support an argument for religious foci
of similar hillforts.123 Conversely, it has been hypothesised that in other cultures circular
structures gained a ritual purpose after the development of rectangular housing. 124 Above all,
no rectangular structures in British prehistory compare to the construction or size of those
found in the Seven Estuary, furthering the idea of a wetland building tradition specific to
South Wales.125
Rectangular buildings within settlements may show consistent use from the Bronze
Age, but in southern Wales, rectilinear-shaped enclosures themselves became more frequent
at a much later date in the Late Iron Age. 126 As mentioned above, in Gwent thirty-two (37%),
of the eighty-seven defended enclosures with documented enclosure shapes, are sub-
rectangular or rectangular. This is not a majority, but along with those forty-four (51%)
118
Bell and Neumann (1997), 98: table 1.
119
Bell and Neumann (1997), 104.
120
Bell and Neumann (1997), 108.
121
Locock and Yates (1999), 88.
122
Howell and Pollard (2004), 143-5; Howell (2009), 46-7
123
Jackson (1999), 198.
124
Flannery (2002), 433, footnote 1.
125
Bell and Neumann (1997), 103-4.
126
Gwilt (2007), 302.
37

enclosures of sub-circular or circular shape, the preferences among the Silures for enclosure
shape become apparent.127
More internal structures have been found in Glamorgan, twenty-nine in total, which
can be expected due to the larger set of enclosures, yet the prevalence of rectangular buildings
is noticeably lacking. Only two such structures have been found, both in the Vale of
Glamorgan. This suggests the cultural phenomenon of rectangular building had not yet
reached or been implemented in the western half of Glamorgan before the Romans arrived, or
even possibly was not deemed a necessary change of building type. Of course, more
excavation in western Glamorgan may change this situation.

Associated Annexes and Field Systems


Twenty-nine annexes are also mentioned in the GGAT reports as connected with
enclosures in South Wales, with an additional five in Brecon. This may prove important as
larger hillforts with attached areas, focused upon ‘stock management’, present the idea of
cattle as a marker of wealth and status. 128 In the east, Twyn-y-Gaer (Cwmyoy) has a
crossridge type of annexe believed to be for animal husbandry, 129 while elsewhere in South
Wales, four other annexes are detached from their corresponding site and six others are
classified as lobate. Llanmelin Wood stands separate with lobate and detached annexes.
Most notable in the west are three sites which are over 2 ha in size: Cil Ifor Top (2.9), Caer
Dynnaf (3.8) and Y Bwlwarcau (7.2). Cil Ifor Top is the largest site in Gower, and being
centrally located is best suited for a focal point of communal activity possibly with an
aristocratic family in power over the entire peninsula. An annexe there may have been for
such displays of wealth, or, as the area is naturally isolated, a more communal social
environment may have persisted where the annexe served to further this. Both Caer Dynnaf
and Y Bwlwarcau in south Neath Port Talbot are relatively centrally located amongst clusters
of smaller settlement sites, further substantiating the claim of these locations as centres for
ruling families or key communal points.
Of these thirty-four total annexes, twenty are associated with sites that have bivallate
ramparts or larger. Out of those twenty sites, eleven are of a size greater than two hectares
(one site has an unknown internal size). Three of these eleven sites are also complex in
structure: Twyn-y-Gaer (Cwmyoy) and Pen Twyn situated in the northern Gwent and
Llanmelin Wood in the south. The other eight larger, heavily defended enclosures are of
either rectangular or circular shape (four each).

127
Wiggins (2006), 20-3: table 5.
128
Cunliffe (1971), 62-3; Bowden and McOmish (1987), 83; Wiggins (2006), 25: table 7; Gerrard et al
(2006), 38: table 7.
129
Probert (1976), 109.
38

Three field systems were also identified in Gwent: Priory Wood Camp, Buckholt
Wood Hilltop Enclosure and Talaches Farm. 130 Priory Wood Camp and Talaches Farm both
contain bivallate ramparts, while the other has a multivallate defence. Talaches Farm’s field
system was identified by cropmarks, and it must be mentioned that an annexe can also be
found at the site. This suggests an agriculturally based settlement, possibly mixed with
pastoralism, with storage facilities for not only gathered resources, but possibly for all the
other materials involved with the production process. Of the three sites with associated field
systems, only Priory Wood Camp has a known internal size (3.87 ha), and as expected is a
larger sized enclosure. In the west, eight field systems have been found associated with
defended enclosures. Two fields have been identified through cropmarks which suggest an
agricultural practice, most notably at Whiton Lodge. Cil Ifor Top, the largest site in Gower,
also has an associated field, although it is only identifiable by earthworks which leave its use
in question.

Map 4: Hillforts of 2 ha in size or greater within the South Wales settlement pattern.
©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved. Settlement locations taken from Wiggins (2006),
Gerrard et al (2006), RCAHMW (1986) and Silvester (2007). © Annotations Copyright GGAT/Cadw

Possible Patterns
With all of the known features of defended enclosures considered, it is not surprising
that almost all of the more complex enclosures with larger defences and associated features
are greater than two hectares in internal size (Map 4). No other assumptions can be made

130
Wiggins (2006), 26: table 8.
39

except that these sites were important to those living in the area at the time. Functions such as
production, storage, elite housing or religious purposes give the enclosures more prominence
in daily life and in turn more necessity for the extra area and defences. Hillforts and groups
will be suggested here which may have played prominent roles within the Iron Age Silurian
society, at least locally. Possible larger geographical groupings may be suggested, such as an
independent Gower, but much more excavation and interpretation will be necessary to form
any conclusions.
One exception to the general theory presented in this thesis is found at Pen-Toppen-
Ash Camp (Coed-y-Caerau (Langstone) 131) (Map 4: site 1), where a bivallate rampart system
and an annexe are found at a site of only 0.45 ha in size. This could be due to the site being
located on a ridge and possibly losing some of the structure size over time, or, just as likely,
the site was a purpose built structure and did not need to be of a larger size. The existence of
two other sites, of similar size also with bivallate ramparts, quite nearby suggests correlation
among the three; possibly being built by the same family/group and used simultaneously for
different functions, or even built consecutively after one fell out of use.
In Gwent, the Usk river served a central role as many settlements were situated
nearby. In the north, Twyn-y-Gaer (Cwmyoy) (2.98 ha) (Map 4: site 2) and Pen Twyn (4.17
ha) (Map 4: site 3) have been mentioned as an important cluster, as well as some focusing
upon the lowland plains between the Wye and Usk rivers. Llanmelin (6.56 ha) (Map 4: site 4)
and Sudbrook (3.21 ha) (Map 4: site 5) have been explicitly shown as an important Silurian
cluster, exploiting the Gwent Levels, and likely serving as the economic and cultural centres
of Gwent, if not all of South Wales.
On the south Gwent coastline sit the Gwent Levels, about 111.2 km 2 of reclaimed
estuarine alluvium between the rivers Ely and Wye. 132 A section of the Levels near Cardiff
partially falls within Glamorgan, but the extent of the alluvium and the major impact from its
exploitation remain within Gwent. The largest section of this six kilometre coastal plain is
the Caldicot Level extending from the river Usk eastward to Sudbrook. With the Usk river
acting as a central route (e.g. communication 133) through much of Gwent and into Brecon, and
Sudbrook being such an important site to the Silures in the east, the Caldicot Level may have
served a greater position within the flood plain. This is echoed in the location of the most
prominent settlements on the Levels: Goldcliff, Greenmoor, Magor Pill and Redwick are all
situated on the Caldicot Level.

131
Coed-y-Caerau (Langstone) and Pen Toppen Ash Camp are often mentioned as a single site.
Wiggins (2006), however, refers to them individually, and as such they are listed separately in
Appendix 1.
132
Rippon, S. (1996). Gwent Levels: The Evolution of a Wetland Landscape. York: Council for British
Archaeology Research Report 105.
133
See below, page 81.
40

More structure to the Silurian settlement pattern may be found by examining the sites
within the modern boundaries of the Vale of Glamorgan. As mentioned above, rivers are not
only routes of communication but can also be viewed as boundary markers between
territories. The same may be seen here between the Taff and Ogmore rivers, east and west of
the area respectively. Half of the sites over 2 ha in size within Glamorgan can be found in the
Vale. These six sites are spread throughout, almost systematically covering the county with
three sites on the coast and three inland. The only exception to this is at Castle Ditches
(Llancarfan) (4.2 ha) (Map 4: site 6), where its close proximity to the coastal Porthkerry
Bulwarks (4.1 ha) (Map 4: site 7) suggests the pair were mutually connected; most likely in a
similar situation as Llanmelin and Sudbrook. Within the Vale of Glamorgan at least one
example of most enclosure shapes can also be found, yet the majority of those are circular-
and rectangular-like shape. This distinction shows a separate need for such styles of building
in this area over any other in Glamorgan, as less rectangular enclosures are built further west
of the area. The dominance of rectangular and sub-rectangular buildings in this area may also
coincide with this area being not far from Gwent and likely the next stop on a lowland-coastal
trade network.134 Only the Cardiff area lies between the Vale of Glamorgan and Gwent. As
mentioned, not much, if any, remains of Iron Age settlements in the area now encompassed
by the city of Cardiff. However, an influential connection with Gwent can be found through
excavations in the Cardiff/Vale of Glamorgan area. A recent excavation at RAF St. Athan
has provided a rectilinear enclosure of possible Early Iron Age construction yet much Late
Iron Age evidence suggests a heavier occupation in that period. 135 As well as these two sites,
the rural settlement at Whitton has also produced an enclosure possibly of Late Iron Age
date.136 These sites suggest a building pattern of rectilinear defended enclosures during the
Iron Age period, parallel with those in Gwent.137
Following the coast west is Bridgend County Borough. Most of this relatively small
area is covered by the modern city from which the area gets its name. Here is possibly the
most important Silurian settlement area outside of Gwent: Merthyr Mawr Warren (Map 4: site
8). The possible trade connections associated with the site may give prominence to this area
and provide links with the settlement groupings nearby. The settlement gap created by the
modern urbanisation of the area may disguise an Iron Age pattern connecting the local sites
with either Neath Port Talbot or the Vale of Glamorgan. Conversely, the area could have
been a seat of local rule with little to no ties with either county.
In the present area of Neath Port Talbot, a cluster of settlements is found. This
grouping is smaller than those in the Vale of Glamorgan and Gower, but is still too
134
Howell (2009), 42.
135
Barber et al. (2007), 56.
136
Jarrett and Wrathmell (1981), 84.
137
Gwilt (2007), 302.
41

predominant in the area to ignore. The largest hillfort in western South Wales, Y Bwlwarcau
(7.2 ha) (Map 4: site 9), is within this regional grouping, and was quite possibly paired with
the only other ‘large’ site in the area: Mynydd-y-Castell (2.7 ha) (Map 4: site 10). Both sites
may have asserted some form of hegemony upon the area, possibly extending toward and
including Merthyr Mawr Warren. Most noticeable in the area is the predominance amongst
the sites to be of sub-circular shape, following the river Neath. Although the area is a smaller
community, the prevalence of such a pattern shows at least an underlying preference due to
some factor, environmental or possibly cultural, found in the area. Settlement gaps at
Bridgend and Swansea, east and west of the county respectively, almost isolate the
community. This could have been the organisation of a local grouping which separated
themselves from Gower and the Vale of Glamorgan. Yet, it could in fact be the opposite, as
the little evidence found near these settlement breaks are of circular enclosures, which
suggests influential building practices if not direct association with those neighbouring
communities.
Gower in itself creates the opportunity for a more easily defined community. Being
surrounded on three sides by coast, it can be expected that any society living in the area will
be of a more sea-borne subsistence. The number of irregularly shaped sites located in coastal
cliff and coastal promontory positions only furthers this claim. There still can be seen the
usual circular-like enclosures within the interior, but the predominance of irregular sites, their
shape mostly due to the coastline and its erosion over time, is evident in comparison to other
sites not only in Glamorgan, but in all of South Wales (Gwent has only one irregular shaped
enclosure; located north of central Monmouthshire.) Thus the coastline most likely played a
prominent role in the settlement pattern of the Silures inhabiting the Gower peninsula.
Glamorgan exhibits over twice as many Iron Age defended enclosures as Gwent,
which should be expected with such a larger area. 138 However, the natural landscape provides
a different situation than that in Gwent. The prevalence of uplands above 250 metres in
elevation in the north leaves only lowland coastal areas for settlement. This results in a
largely condensed settlement pattern such as in the Vale of Glamorgan where a similar
number of sites as Gwent are found in a smaller area. Similar to Gwent, large hillforts are
found throughout, possibly in pairings or clusters (e.g. Castle Ditches (Llancarfan) and
Porthkerry Bulwarks) creating strong centrally located sites easily accessible to the
surrounding settlements. This may be most apparent in Gower where only one hillfort is
above 2 ha in size, Cil Ifor Top (2.9 ha) (Map 4: site 11), and it can be found almost centred
upon the whole peninsula.
In the Brecon area, many sites are situated nearby to suggest clustering if they are
contemporary. Large enclosures like Slwch Tump (Map 4: site 12) at the bottom of the valley
138
Gerrard et al (2006).
42

near the Usk river may indicate a communal or central enclosure with connections to those on
the surrounding higher lands. With that in mind, the area may have been characterised by a
single or up to a few controlling parties over the entire valley.

Conclusion
Gwent has been the focus of most of the research on the Silures from Nash-Williams’
excavations to The Gwent County History; more so than Glamorgan and Brecon. 139 Because
of this, a better understanding, albeit still vague, of the Silurian settlement pattern is emerging
within the south-east corner of Wales. Yet, even though the archaeological record can
indicate a preference among modern scholars, it does not disprove any idea of Gwent being
culturally superior to the rest of South Wales.
Settlement patterns in the territory of the Silures probably varied not only in response
to the landscape but also as an expression of preferences and roles within local society.
Furthermore, a decentralised tribal structure is evident in South Wales through variation in
structure shapes and the relative scatter of larger enclosures suggesting their use locally with
the surrounding structures. This will be shown further with regard to the roles of the hillforts
within the Silurian society.

139
Nash-Williams, V.E. (1933). ‘An Early Iron Age Hill-fort at Llanmelin, near Caerwent,
Monmouthshire’, Archaeologia Cambrensis. 88, 237-315; Nash-Williams, V.E. (1939a). ‘An Early
Iron Age Coastal Camp at Sudbrook, near the Severn Tunnel, Monmouthshire’, Archaeologia
Cambrensis. 94, 42-79; Aldhouse-Green and Howell (2004); See above, above.
43

Chapter 4: The Role of the Settlements within the South


Wales Society
In Wessex, a recurrent discourse of settlement and social hierarchies has been the
topic of debate over many decades with many agreeing and disagreeing, while others offer
similar, alternative interpretations. 140 Cunliffe, who first proposed the hillfort-centred
systems, attributes his work at Danebury and the surrounding area to a system of tribes with
kings and clans similar to a Celtic society. 141 The discourse on South Wales presented herein
is not as direct nor assured as Cunliffe argued for Wessex, although parallel lines of thinking
will be found. As that suggested by Cunliffe, a hillfort-centred settlement pattern most likely
persisted in South Wales with local groupings possibly extending to a loose federation of
natives connected by a single, shared culture. With the work presented herein and further
research in South Wales, a similar discourse as that in Wessex may arise focused upon the
Silures.
Early hillforts in the Welsh Marches, north of the Silures, have often been explained
as reflecting control of resources, the resulting surplus and a dominant warrior aristocracy. 142
A similar situation most likely existed in South Wales which resulted in size dictating the role
the hillfort plays within the local community. 143 Large forts may be more permanent for
significant functions and families (of greater size and status) or even whole communities,
while small forts probably sustained only small families or groups. 144 Depending upon the
integration of smaller settlements with nearby larger hillforts, the larger sites may have also
become akin to regional centres of the Silurian decentralised tribal structure. Size has been a
consistent issue throughout this discussion and the importance of size as an indicator of social
structure is supported by the archaeological record.
In South Wales, evidence presented in the east has almost fully agreed with the
dominance of larger sites, yet nearly the opposite has arisen in the west. Most exotic goods
recovered within sites near the Wye and Usk rivers were found associated with a larger
settlement (i.e. Llanmelin), while those found along the western coast were often near smaller
sites (i.e. Merthyr Mawr Warren). A study of the beginnings of Iron Age South Wales will

140
A good overview of the discussion, although with a negative bias, can be found in Hill, J.D. (1995).
‘How Should We Understand Iron Age Societies and Hillforts? A Contextual Study from Southern
Britain’, in Hill, J.D. and Cumberpatch, C.G. (eds.) (1995). Different Iron Ages: Studies on the Iron
Age in Temperate Europe. Oxford: British Archaeological Reports International Series 602, 46-7.
141
Cunliffe, B. (1984). Danebury: An Iron Age Hillfort in Hampshire. 2 vols. London: Council for
British Archaeology.
142
Davies and Lynch (2000), 151.
143
Davies and Lynch (2000), 158; Children and Nash (2001), 129.
144
Davies and Lynch (2000), 148.
44

hopefully provide insight into the evolution of distinctly different regions within a similar
material culture.
This approach to understanding social aspects within settlements can be quite
difficult, as a modern view upon the socio-economic relationship of people with each other
and within ‘households’ can obscure and even bias our reconstructions of the past. 145 Yet, it
should not discourage research on the topic, as methodology towards any topic must start
somewhere; even modern social norms have natural origins in history. This thesis will not go
as far as looking into gender roles or even social roles within families, as the objective is the
interaction of the society as a whole with their settlements. Kinship ties will only be
discussed in reference to the roles they dictate and the extent to which natives decide to
interact with others.

Social Stratification
It is the viewpoint of this dissertation that the Iron Age hillforts presented in the
previous chapter represent the social stratification that developed within South Wales up to
and during the Iron Age. The catalyst for the evolution towards social stratification may have
been the increase of population within settlement groups where kinship ties are lost over time
as larger populations inevitably create social distances between groups of people. 146 Thus
social institutions, like kinship lost to overpopulation, may have been most easily replaced by
prestige and privilege concepts. Alternatively, the growing numbers of residents within a
population may have led to a greater administrative need, which then ultimately required a
hierarchy system in place. This last conjecture may coincide, or remain wholly independent
of, a viewpoint where a political leader, referred to as a chief among the smaller societies
usually in question, is seen as the compelling force leading a population towards a greater
society. This emerging leader sees the potential for the community to work together and
produce the surplus of resources and the material culture. If those last two ideas can be
merged, then in South Wales is found a growing population propelling a political system
which creates a leader who then drives the populace towards this better public economy. In
the case of the Silures, this political system most likely developed in many instances
throughout South Wales independently of one another, creating a decentralised political
structure of local tribal chiefs.
Thus, the Silures were affected by socio-economic stratification which later led to
those with greater wealth gaining power. This process most likely took place during the
Bronze Age or earlier, but with the adoption of hillforts in the Iron Age, this social
stratification became more prominent within the physical landscape, causing it to have a
greater effect upon the society. Many possible evolutionary trends may have led to the
145
Wigley (2007), 182-3.
146
Cancian, F. (1976). ‘Social Stratification’, Annual Review of Anthropology. 5, 227-48.
45

economic stratification found in South Wales, and this dissertation does not wish to advocate
a certain social stratification evolutionary trend over any other. In fact, it should be
mentioned that none of these ideas have wholly replaced or eliminated the plausibility of any
other. Instead, the purpose of including this discourse was to provide plausible origins of a
social and political system, which could then be discussed alongside settlement patterns, the
evolution of these settlements, and the Iron Age evidence found within the archaeological
record of South Wales and elsewhere. From this, ideas then may be surmised for the social
and political situation of the Silures before the Roman Invasion.

Social and Structural Evolution


The Silurian settlement pattern discussed in the previous chapter highlighted the
defining characteristics of the Iron Age hillforts with regard to location, enclosure shape,
defences, internal size and associated features. These aspects may indicate, individually or
collectively, varying social, political or economic systems in place during the Iron Age of
South Wales.
Kent Flannery has accumulated substantial work into the origin of villages about
Mesoamerica and the Near East. 147 Although this work is on societies outside of Britain, it
examines settlement size, enclosure shape (circular and rectangular) and archaeological
findings which may bring about a plausible evolution of the Silurian settlement pattern. As
with the abundance of circular and rectangular enclosures found in South Wales, a similar
pattern may have been at work in these areas as well.
In the study, Flannery’s main justification for the evolution from circular to
rectangular housing comes from the privatization of resources. 148 Circular enclosures often
were small and accompanied shared storage units. This leads to the supposition that groups
of circular enclosures could have incorporated a family that shared amongst themselves. A
village of multiple small enclosures could mean a large family of numerous brothers and
sisters with each sibling and family having their own ‘hut’, or possibly a culture in which
polygamy was practised allowing a different ‘hut’ for each wife or husband (depending upon
whether they were polygynous or polyandrous). Caesar mentioned large families as common
in Britain at the end of the Iron Age, a situation that possibly also persisted in South Wales. 149
Flannery found rectangular enclosures often included, and were possibly built for the
sole purpose of, private storage units within the structures. Obviously, this presumes a
change away from the sharing of resources into privatization; possibly along with this came a
view of the village less resembling an extended family, or even a familial association. Thus
147
Flannery, K.V. (1972). ‘The Origins of the Village as a Settlement Type in Mesoamerica and the
Near East: A Comparative Study’, in Ucko, P.J., Tringham, R. and Dimbleby, G.W. (eds.) (1972).
Man, Settlement and Urbanism. London: Duckworth, 23-53; Flannery (2002), 417-33.
148
Flannery (2002), 417-8.
149
Caesar. Gallic Wars. 5.14.
46

in the former villages, resources were acquired by and shared within the nuclear or extended
family, while the privatized enclosures kept resources for use only within the household.
A search for Flannery’s theory in South Wales must not go further west than the Vale
of Glamorgan. The abundance of rectangular enclosures in Gwent and eastern Glamorgan
likely coincide with the prevalent trade network connections pushing the local natives to look
for ways of privitizing their storage and distribution habits. A less concentrated trade market
in the west of Glamorgan suggests a reduced desire for those living in the area to restrict
access to their stored goods, and the predominantly circular settlement pattern indicates this.
Flannery later admits that although the connection between settlement type and size was
strong in his research, the geometric shape did not determine the purpose, but the society
within which the settlement was built.150 Polly Weissner, while studying hunter-gatherer
societies, proposes a similar theory along these same lines.151
Weissner contrasts societies through the level at which they accept the risk of
survival. The first consists of a society that chooses to attain resources as a group, therefore
the risk involved in the acquiring of such resources, or the inability to, affects the group as a
whole. Since whatever is gathered must be shared, there is less incentive to intensify their
mode of collection. This type of society traditionally incorporates an ‘open’ site plan to their
settlement patterns, and chooses to live either in large structures housing the entire group, or
split among smaller, closely built structures. Weissner’s second society personalises the risk
and reward of resource gathering by providing for and sharing only within the nuclear family.
Groups adhering to this strategy are expected to exhibit a more ‘closed’ settlement pattern,
with storage areas and household units being more privately structured. More incentive is
found in these families, as the rewards of their work is not shared, but benefits only
themselves and those close to them. It is amongst this latter society that distant familial or
kinship ties may be easier to break, as a focus upon only those within the immediate family
may lead to connections from generations past becoming insignificant or even forgotten. 152
Even though Weissner’s research focuses upon hunter-gatherer societies, as opposed
to the sedentary Iron Age lifestyle of the Silures, the logic behind the settlement pattern
remains relevant, and could prove more directly related when discussing the seasonal
pastoralist nature found in South Wales.
As is shown in the South Wales archaeological record, the prevalence of defended
enclosures indicates a later development which started with the change from a hunter-gatherer
to sedentary lifestyle.153 This change to a sedentary lifestyle incorporating agriculture has

150
Flannery (2002), 421.
151
Weissner, P. (1982). ‘Beyond Willow Smoke and Dog’s Tails: A Comment on Binford’s Analysis
of Hunter-gatherer Settlement Systems’, American Antiquity. 47, 173-4.
152
Flannery (2002), 421.
153
Cunliffe (1971), 61-2.
47

most recently been centred upon the Middle Bronze Age, although an indication of early
settlement in South Wales has often been through agricultural, and even lithic, evidence
which remains highly uncertain.154 An agricultural lifestyle, even if only for part of the
family’s resources, may have been easier for some than others. Weissner points out that those
having already privatized their resources would be able to adapt to farming much more easily
than those who share.155 With agriculture being an intensive subsistence, the willingness to
share with others who have less demanding responsibilities can be difficult to find. Thus,
those relying upon and providing for only themselves are in a much better position to adapt to
an agricultural lifestyle. With that, the Bronze Age ancestors of the Silures may have entered
sedentary and agricultural existences at different times, creating possibly a hostile
environment continuing into the Iron Age, and providing the necessity for social changes,
including isolation and security in the way of communities and/or defended enclosures.
It should be acknowledged that the work by Flannery and Weissner pertained to
housing structures and not defended enclosures. Yet, their work cannot be limited to
undefended structures. Security was important to Iron Age tribes, and therefore they chose to
provide defences around their structures. This desire for protection does not prohibit an
evolution towards a privatized lifestyle, or adherence of similar settlement patterns. In effect,
this could have even been another justification for the change: as populations grew, some
chose to remain in small groups within smaller enclosures, while others chose a ‘safety in
numbers’ approach.
Agro-pastoralist subsistence, as is present in South Wales, may even result in a
seasonal transhumance lifestyle where families cycle through locations depending upon not
only livestock migration, but plant abundance and seasonality as well. 156 Defended enclosures
amongst these settlement patterns may only serve as centralised gathering points, granaries, or
even workshops for the domestication and harvesting of the plants and animals. Some sites
may even have been visited by multiple families at various points in the seasonal cycle.
Large sites would almost seem unnecessary, therefore smaller settlements, characteristic of
western South Wales sites, with evidence of tools and plant and animal remains may be
logical starting points for further research. Unfortunately, such a settlement system, living
within semi-permanent or even temporary shelter, would obviously be difficult to detect in
the archaeological record as evidence would be easily lost over time.
Weissner believes that a ‘closed’ site plan may prevent jealousy and conflict that may
result due to the prosperity of one household over another. However, in a more complex
sedentary lifestyle, as seen in South Wales, the desire to display one’s wealth and prosperity
154
Hamilton, M.A. (2004). ‘The Bronze Age’, in Aldhouse-Green and Howell (2004), 95.
155
Weissner (1982), 175.
156
David, N. and Kramer, C. (2001). Ethnoarchaeology in Action. Cambridge: Cambridge University
Press, 247, 267-70.
48

was probably more common. Weissner also leaves her theory open to the settlement structure
type being of varying geometric shape, i.e. circular, rectangular, polygonal or irregular, all of
which are found in South Wales. Flannery points out that circular housing in Predynastic
Egypt remained circular while growing in size and privatising storage, even though those in
the Levant changed to rectangular housing. 157 This allows for the predominance of circular
housing found in the western Silurian territory as compared to Gwent and the Vale of
Glamorgan, yet the possibility of similar societal risk patterns. However, with the dominance
of archaeological trade goods being found in Gwent, there may have existed more incentive
for those in eastern South Wales to privatize storage and intensify production, most logically
for the desired effect of a raised social status.
In the central Welsh Marches, Wigley has proposed that from the Late Bronze Age, a
desire, either by communities or individuals, to claim land through recognisable boundary
systems may have affected social systems. 158 Claiming land that reached long distances
necessitated boundary markers and would have been a major task to complete. However, a
communal effort by several ‘neighbours’ with the initial family could make the undertaking
more obtainable. This then would create ‘networks of affiliation’, bringing together different
families and kin groups. Although this is a plausible outcome of land division, and in all
likelihood may have occurred in a few instances, a seemingly more believable outcome to
land claiming would have been the opposite. Competition over tracts of land inevitably leads
to conflict, as it still does in modern society, which then may have led to defensive
enclosures. On the other hand, Cunliffe sees strongly defended hillforts as just another stage
of continual social development due to increasing population and annexing of land leading to
conflict and the need for defences. 159 Boundary systems may even pre-date defended
enclosures,160 suggesting the rise in the desire or need to defend one’s property and family
may have come about through problems caused by or supplemented with boundary disputes.
Annexes found at Twyn-y-Gaer (Cwmyoy) and Pen Twyn in northern Gwent may
indicate social systems in place either similar to those found by Wigley in the central Welsh
Marshes, where land division was a communal affair, or to that of Cunliffe, signifying a
separation from neighbouring communities.
That larger building projects may have been completed through communal effort
should not be dismissed. However, most likely those living within South Wales were
frequently competing for land. This may not have necessarily resulted in physical
confrontation, but enclosures and boundary markers may have become necessities for
demonstrating and keeping possessions. As those with more wealth and power had greater
157
Flannery (2002), 421.
158
Wigley (2007), 179.
159
Cunliffe (1971), 61-2.
160
Wigley (2007), 179-80.
49

access to resources, their position within the local society was probably perpetuated through
their ability to acquire and maintain land.
Bowden and McOmish, when discussing late prehistoric settlements in Wessex, make
a distinction between multivallate and univallate enclosures. 161 They stress that the
construction of a multivallated defence is more towards a desire for increased prestige and
isolation, rather than greater security. This can be seen in that less defended enclosures often
allow their ditches to backfill, either through lack of maintenance or deliberate methods, such
as burials or refuse deposits. Since smaller enclosures exhibited such non-functional ditches,
suggesting security was not an issue, the larger sites which kept a level of upkeep on their
ramparts must have done so for other purposes (i.e. prestige and isolation). In South Wales,
this may be seen at the smaller enclosure ditches found at Church Farm, Caldicot near the
Severn Estuary.162 The defensive features there consist of low and wide banks and shallow
ditches, only 0.5 to 1 metre deep. Although the ‘defences’ may have been to keep low level
threats such as petty thieves, vagrants and marauding animals out, stock in and to provide
drainage, a symbolic use of status indication has also been suggested. If the latter, the small
‘defence’ sizes may have indicated a lower status, and higher social standing may have been
accompanied by deeper ditches. Furthermore, the four largest defended enclosures in South
Wales (Llanmelin Wood Camps (6.56 ha), Y Bwlwarcau (7.2 ha), Tredegar Hillfort (9.28 ha),
Lodge Wood (9.58 ha)) all exhibit multivallate defences, which stresses the correlation
between site size and importance.
Hingley proposes that kinship ties may still have existed amongst settlements that
maintained narrow boundaries and small distances between enclosures, creating a local
cooperative social group, whereas those choosing to settle further away from their neighbours
may have also chosen to dismiss any kinship connections. 163 These differing settlement
landscapes thus coexisted, usually inhabiting topographical locations that best suited their
chosen lifestyle. The uplands best allowed families to stay isolated and maintain control of
large tracts of land for themselves, while the lowlands often contained multiple settlements
that adopted closer integration with those around them, possibly choosing an open
settlement164 in which to live. Hingley’s work focused on the Oxford Clay Vale and the
Oxford Uplands of the Upper Thames Valley in central southern England; however some
aspects of this settlement situation may have existed within South Wales. This may be seen
in the majority of settlements in South Wales located in close proximity of others, but some

161
Bowden and McOmish (1987).
162
Insole, P. (2000). ‘The Archaeological Excavation of a Romano-British Farmstead at Church Farm,
Church Road, Caldicot, Monmouthshire’, Archaeology in Wales. 40, 29.
163
Hingley, R. (1984). ‘Towards Social Analysis in Archaeology: Celtic Society in the Iron Age of the
Upper Thames Valley (400-0 BC)’, in Cunliffe and Miles (1984), 72-88.
164
Hingley’s term ‘open settlement’ refers to that of the domestic area, not necessarily the entire
settlement; an enclosure may still exist within an open settlement.
50

sites remain in more isolated upland locations. Settlement landscapes which include a large
hillfort may very well have operated under a more cooperative relationship rather than
subservient. Size of enclosed settlements may also correlate with the scale of social groups:
i.e. large-scale enclosures incorporate large-scale social groups. This landscape would most
likely include the larger enclosures as communal areas, where goods were produced,
distributed and stored to the benefit of the entire community, regardless of a family living
within the ramparts. This idea will be stressed further below with regards to Llanmelin Wood
Camps, Lodge Wood, Sudbrook and the South Wales trade network. Hingley also proposes
that the social grouping of a settlement landscape may dictate the social relations of
production. Closed, isolated settlements may show more division of land per its function
while open settlement groups may exhibit the opposite, choosing to let their agriculture and
animal husbandry mirror their settlement pattern.
The Silurian preference towards an agro-pastoral lifestyle 165 may fit quite well with
the isolationist Oxford Uplands idea; however it is hard to believe the same families would
then adopt the Oxford Clay Vale open settlement kinship pattern during their lowland
occupation periods. These ideas should not be wholeheartedly dismissed as a few small
settlements, like those found in north-west Gwent and Merthyr Tydfil county, may have been
a part of a permanent isolated settlement pattern similar to the Oxford Uplands.
Hill proposes that hillforts in Wessex did not necessarily take a prominent role in
social hierarchy. Rather, those which were ‘not-farmstead’ settlements played a more central
role in the society as locations of wide-ranging communal activities. 166 Thus, a community
was established incorporating multiple settlements of varying size and shape, among the same
socio-economic class. Again, this stresses the larger enclosures of South Wales as having
more communal purposes while the smaller ‘farmstead’ sites remain more individual in
nature. Yet, even more isolated settlements may still in many ways have been connected to
the wider society through their use of community controlled land and exchange networks.
Wigley has a slightly different theory towards the social implications of small
enclosures.167 Throughout the central Welsh Marches, evidence suggests that enclosure
ditches were often reworked and at times even allowed to refill naturally. This latter point
does not deem the ditch non-functional, as Bowden and McOmish might believe, but instead
serves as testament to a reworking of the boundaries along a different line. It is in the
construction of the enclosures that Wigley finds evidence of social relationships. He sees the
evidence for gang working and the sheer laborious feat necessary to erect and rework
enclosures together as proof of an interdependent social system where several different family
groups work together for the sake of a single ‘household’. This mirrors his view on the
165
See above, page 11.
166
Hill (1995), 45-66.
167
Wigley (2007), 182-6.
51

establishment of boundary systems, expressed above. From there social authority may be
found within the ability to gather workers, either through kinship ties, mutual benefit
agreements, or payment of some kind; thus making the social status determined not by the
structure size, but by the workforce needed to complete the enclosure.
This social standing through workforce size can thus be a more fluid measurement of
status, as any of a number of reasons could dictate a family’s ability to recruit others for
labour. This may be seen in South Wales at Collfryn where a multivallate enclosure of fourth
century BC date is later reduced to a univallate rampart by the first century. This reliance
upon others to affirm a family’s social status may coincide with the community’s willingness
to allow a family the rights to land and resources. In other words, a household almost needs
to prove itself worthy of such a declaration before a workforce can be organised to make it
official. Furthermore, a connection to the landscape, possibly through kinship ties, could be
of equal importance as anything else within the social system. Genealogical associations to
others and the landscape could have strong bearing on the social evolution of a family, and
how the rest of the community views that connection solidifies its importance.

Conclusion
There has been much speculation as to the purposes for defended enclosures both
within this dissertation and across the myriad of published resources concerning the topic.
Hillforts may have developed along any number of evolutionary trends, from markers of
possession and security to communal importance within the society. As well, the importance
and evolution of a structure was not limited to the larger examples. Small enclosures could
have played an equally prominent role in the local society, or even as a privatization from it.
In some cases, however, the enclosure type could be less significant to the overall
function while that of the structures within the enclosures may more accurately show the local
settlement pattern. This has been previously mentioned above with the discussion of
roundhouses and rectangular buildings found within hillforts. Unfortunately, this idea must
be supported by more archaeological work done within enclosures, but it is this lack of work
done on defended enclosures in South Wales that prevents much more from being known
about the Silures.168
It has been demonstrated from the archaeological record that social stratification
existed within Iron Age South Wales. This differentiation may have come from any of a
number of origins: through a surplus of resources, the implementation of a material culture, or
a necessity through increasing populations. Either way, some groups of families or

168
This problem has been acknowledged recently: Gwilt, A., Bell, M., Cardy, B., Davis, M., Lodwick,
M., Makepeace, G., Northover, P., Olding, F., Sharples, N. and Yates, A. (2003). A Research
Framework for the Archaeology of Wales; South-east Wales – Later Prehistory.
http://www.cpat.org.uk/research/selpre.htm., Last Date of Access: 2/12/11.
52

individuals became more privileged than others, possibly creating a socio-economic


hierarchy.
From there, society in South Wales may have evolved along innumerable paths.
Smaller settlements choosing to form communities could have built larger communal sites for
certain purposes, possibly along pastoral boundaries where construction practices could have
created ‘networks of affiliation’.169 Examples of this could be seen where a single large
enclosure is surrounded by smaller sites (i.e. Cil Ifor Top in Gower). Elsewhere, settlement
groups of the same socio-economic standing possibly incorporated a social hierarchy where
the community decides each other’s social status. This status could then have been displayed
through a number of ways, including complex enclosures. Larger enclosures are not
necessary for smaller communities of this type, but a likely natural step for larger groups;
both of which may be seen in Neath Port Talbot area. These settlement systems, or countless
other forms, could have integrated kinship structures for stronger ties.
For any number of reasons, socio-economic differentiation in these settlement
systems may have further developed creating a wealthier elite class; possibly as a result of
warfare or other cultural influences. No matter the reason, those more fortunate may have
taken advantage of their position to establish further their higher standing through other
means, such as trade. Out of this an aristocracy may have formed where families living
within larger enclosures controlled their communities and flaunted their wealth to others
through personal displays.
All of the aforementioned settlement landscape situations are possible, and may have
existed in South Wales. However the archaeological record provides a wealth of evidence
pointing towards elite class goods perpetuating socio-economic segregation where the highest
class is also controlling and providing for those of lower social standing. This social
stratigraphy may have resulted in local chiefdoms, and from there possibly a larger Silurian
chiefdom hierarchy.

169
Wigley (2007), 179.
53

Chapter 5: The Trade Network of South Wales


The location of Gwent in the east of the territory of the Silures is instrumental in the
reagion’s relationship with other tribes. Substantial artefactual evidence in Gwent from the
neighbouring tribes, mainly those most nearby (i.e. the Dobunni and Belgae), brings about a
better understanding of the outside influence placed upon South Wales. Coins, brooches,
weapons and pottery from outside Gwent show how the Silures of the east were well
connected to their neighbours, primarily through trade networks. Conversely, the Silures
maintained their own production of pottery, and possibly axes, suggesting a stronger cultural
awareness of themselves. A look into the location of coin finds along with the distribution
patterns of local products will show the reciprocating cultural influence the Silures had upon
other tribes. The ‘exotic’ nature of goods acquired through trade most likely represented a
higher social status, and it would have been through these material goods, rather than those
imported or manufactured in South Wales, that the local ruling class would have been able to
assert their high status and provide for those within the local area, possibly in exchange for
services or monetary compensation. While locally produced goods have been stressed as
indication of a distinct South Wales material culture, it would have been the goods brought
from outside the region by the leaders that were prized and traded perpetuating such an
elevated social status.

Coins
Whilst coins scattered across South Wales are not numerous by comparison to other
British tribes, numismatic evidence can still provide insight into value-oriented trade systems
in the area. Gwilt recently updated the list of Iron Age coins in Wales, adding twelve new
finds.170 Of forty-two coins, twenty-two are found in Gwent alone. All Dobunnic coins found
in Wales are entirely within Silurian territory: twelve in Gwent, three in the Vale of
Glamorgan and one in the southern Powys/Brecon area. The concentration of Dobunnic coins
in Gwent is partly due to the close proximity with Dobunni territory; however the economic
influence must not be undervalued. Although the majority of Dobunnic coin finds being near
the Wye river may only be from short merchant trips across the river, the likelihood of a trade
route up the river, to sites such as Llanthony Abbey and Dingestow, should not be ruled out.
In fact, maritime trade seems prevalent in Gwent during the Iron Age as coins from South
England, Gaul and even the Mediterranean have been found in the area. It should be noted,
however, that the Severn-Cotswold region also contains non-local Iron Age coins, and much
of that which is found in Gwent may just be from across the Severn. 171 Yet, the fact remains

170
Gwilt (2007), 304-6; 320-1: Appendix 3.
171
Gwilt (2007), 306.
54

that Gwent was a part of a trade network that reached not only most of Britain, but cross-
channel and beyond, either directly or indirectly through the exchange with the Severn-
Cotswold region.
There is also cultural trade evidence between the Silurian and Dobunnic and
Durotrigian communities which supports the view of complex inter-tribal exchange networks
having been established in the area. 172 A strong connection with the neighbouring Dobunnic
trade network can be inferred by the prevalence of Dobunnic coins, and through this a heavy
reliance upon them for much of the Silures’ outside trade, especially with cultures on
mainland Europe. Nonetheless, the close proximity to and influence of the Dobunni does not
necessarily mean a dependence upon them for all of the tribe’s trading activities.
Furthermore, this amount of interaction with non-local and non-British items is not found in
South Wales outside Gwent.
With Gwent’s strong trade ties as evident through archaeological findings, the Silures
of Glamorgan most likely relied upon the eastern Silures for much of their trade. Yet, the
archaeological record indicates those outside Gwent may have had connections within Wales,
southern England and the continent independent of their neighbours. Gwilt’s updated list of
Iron Age coins in Wales provides insight into Glamorgan trade networks as well as Gwent. 173
Dobunnic gold at St. Nicholas and Penllyn and silver at Whitton, all in the Vale of
Glamorgan, show the surprising extent to which Dobunnic currency reached. With the
nearest Dobunnic coin around 45 km east at Magor, a trade route directly to the Vale of
Glamorgan by sea seems more likely than a land journey through southern Gwent; since if the
latter were the case, more coins would be expected to be found between the two sites.
Glamorgan trade connections do not end with the Dobunni. Further west, two Durotriges
staters, silver at Llyn-fach, Hirwaun and bronze at Pyle, and one Kentish potin at Merthyr
Mawr Warren suggest not only a trade network independent of any Gwent influence, but one
extending beyond the Severn Estuary. Although a land trade route for the coins is plausible, a
sea based merchant network seems most probable. Iron Age British tribes tended to keep
land based trade more local (e.g. Malvern and Glastonbury ware 174), therefore a coin from
Eastern England arriving on the shore of Southern Wales can be seen logically as the effect of
sea-borne trade. Furthermore, all coins found in Wales from an origin outside south-west
Britain can be found within five kilometres of a waterway capable of sustaining maritime
communications, albeit on a small scale.

172
Gwilt (2007), 308.
173
Gwilt (2007), 304-6; 320-1: Appendix 3.
174
See below, page 58.
55

Brooches
Archaeological evidence has shown a desire for brooches, most notably La Tène type,
flourished during the Iron Age in South Wales. 175 Forty-seven La Tène brooches of varying
types and twenty-two penannular brooches were found in Silurian settlements, most of which
within Gwent. La Tène brooches were present at Lodge Wood and Llanmelin Wood Camps,
two of the largest sites in Gwent, while also found at Sudbrook on the Gwent Levels where an
obvious connection to trade is found. Following that discussed previously, the roles of these
‘large’ sites may have been as trade centres of the Silures, where the trade goods were under
the control of the local elite. The most important find of brooches in Gwent is at Twyn-y-
Gaer (Cwmyoy) where nineteen iron examples or fragments are known. Dating of the
assemblage spans the Middle Iron Age, and has been compared to a similar find in
Herefordshire, showing a possible Ordovician connection. 176 Brooch finds in Gwent show a
consistent Middle to Late Iron Age trade, with the La Tène brooches showing connections to
Central Europe. Most likely, this was through the extensive trade network of the Dobunni,
however a direct connection to mainland Europe should not be dismissed. Iron brooches, not
only found in Twyn-y-Gaer (Cwmyoy) but also Lodge Wood, Sudbrook and Caldicot, show
the preference of the metal during the Iron Age with thirty-two of forty-seven brooches (68%)
from Iron Age Silurian sites, as compared to Romano-British locations. Hill has linked a
persistent existence of certain brooch styles in southern and eastern England to social
change.177 A similar situation may have been happening during the Iron Age in southern
Wales.
Brooch finds in Gwent best indicate an exotic goods trade network focused upon the
larger enclosures.178 In Glamorgan, the situation is vastly different, as brooches are found
among the smallest defended enclosures in the area: Whitton Lodge (0.3 ha), Mynydd Bychan
(0.3 ha) and Bishopston Valley (0.1 ha). While more brooches are present in Gwent 179, more
locations throughout Glamorgan contain La Tène and penannular types. With finds across the
county, in more locations, and in smaller settlements, the presence of brooches in Glamorgan
implies a different social system is in place involving a broader range of people.
Merthyr Mawr Warren, near modern Bridgend, shows the widest variety of brooches:
two La Tène, four Late Iron Age and four penannular. Adding to these foreign trade goods
are two Swan’s-neck/ring-headed pins. A large coastal sand dune system currently covers
much of the Iron Age settlements in the area. Cunliffe suggests that the area could have been

175
Gwilt (2007), 302-3; 318: Appendix 1.
176
Probert (1976), 115.
177
Hill, J.D. (1997). ‘‘The End of One Kind of Body and the Beginning of Another Kind of Body?’
Toilet Instruments and Romanization’, in Gwilt, A. and Haselgrove, C. (eds.) (1997). Reconstructing
Iron Age Societies. Oxford: Oxbow Monograph 71; Gwilt (2007), 303.
178
Gwilt (2007), 302-3; 318: Appendix 1.
179
This does not include brooches found at Romano–British sites.
56

a ‘port’ in a maritime trade network of SW England and Armorica in northern Gaul due to
pottery collections found in Merthyr Mawr Warren and Mount Batten, Plymouth (Devon). 180
If so, brooches from central Europe would have been available through the Gallic trade
network. These ceramic and cultural connections seem possible, but there is a lacking of
supporting numismatic evidence.181 An east Kentish potin found in the area follows the
connection through eastern England, but not necessarily Europe, since the coin could have
arrived via the trade networks through Gwent.

Metal Production
Metal working and production based artefacts show a society in Gwent creating its
own objects and not just relying upon trade. With Ariconium in the Forest of Dean, the
Silures obtained most of their smelting material from just over the Wye river. 182 Evidence has
been found of bronze working on Middle to Late Iron Age sites in Gwent such as Llanmelin
and Twyn-y-Gaer (Cwmyoy). Although of Late Bronze Age association, distinctive socketed
axes found within South Wales indicate not just a possible production centre, but also an
exporting enterprise, as the Silurian axe types have been found in England. 183 Evidence,
however, points to the productions of these axes ceasing by the Middle Iron Age. At Lodge
Wood, metal-working slag has been found associated with Middle Iron Age pottery, possibly
dating metal-goods productions there to that time. Additionally, the Llyn Fawr Hoard
contained an iron sickle closely imitating a native bronze style therefore establishing iron
manufacturing as early as the seventh century BC.184 A logical leap from bronze to iron was
expected, proceeded by possible productions centres in other parts of Gwent, especially near
the Wye river. As well, horse and chariot metal artefacts have been found indicating an elite
class that prized equestrian goods. 185 At Worm’s Head, on the south-western tip of Gower,
stone moulds for possible vessels were found with possible La Tène or Armorican
associations.186 Gwilt believes the moulds and accompanying discs are similar to other finds
in Wales and constitute evidence of sheet vessel manufacturing for cauldrons or other bowls.
Billets and jets of bronze found among the Seven Sisters Hoard in modern Neath Port Talbot
reiterate the idea of metal-goods production outside of Gwent (Illus. 1).
180
Cunliffe, B. (1997). ‘Armorica and Britain: The Ceramic Evidence’, in Cunliffe, B. and De Jersey,
P. (1997). Armorica and Britain: Cross-Channel Relationships in the Late First Millennium BC.
Studies in Celtic Coinage 3. Oxford: Oxford University Committee for Archaeology Monograph 45,
40; Willis, S. (2007). ‘Sea, Coast, Estuary, Land and Culture in Iron Age Britain’, in Haselgrove and
Moore (2007), 116.
181
De Jersey, P. (1997). ‘Armorica and Britain: The Numismatic Evidence’, in Cunliffe and De Jersey
(1997); Gwilt (2007), 306.
182
Walters and Walters (1989), 39.
183
Davies and Lynch (2000), 183; Aldhouse-Green (2004), 161; Gwilt, A. (2004). ‘Late Bronze Age
Societies (1150-600 BC): Tools and Weapons’, in Aldhouse-Green and Howell (2004), 111-39.
184
Howell and Pollard (2000), 82; Martin (2003), 13.
185
See below, below.
186
Gwilt (2007), 307-9.
57

Pottery - Lydney-Llanmelin Ware


Generally, it is believed that during most of the last millennium BC, society remained
aceramic in Wales, operating largely through wooden objects. Not until around the Late Iron
Age did pottery become prevalent, spread through trade. 187 However, Gwilt sees this ‘self-
fulfilling prophecy’ as just a conservative approach by academics to rely upon already
established fixed ceramic sequences, while ignoring the possibility of Bronze Age ceramic
traditions reaching the Early Iron Age or even Late Iron Age pottery styles having Middle
Iron Age origins.188 Around the fourth century BC, Malvern and Glastonbury ware appeared
and became widespread around the Severn Estuary. As mentioned above, Malvern ware was
not common in South Wales, but Glastonbury ware, although concentrated in the Mendips,
can be found at some coastal locations of South Wales as far west as Knave (Gower) in West
Glamorgan.189 For whatever reason, Malvern ware was not preferred by the Silures, and
although other styles persisted, a need for local ceramic production obviously developed.
Lydney-Llanmelin ware, a ceramic tradition characterised by chevron/’eyebrow’/zig-zag
decorations fulfilled this need.
These wares, first identified at each namesake site, constitute a probable South Wales
tradition.190 Outside South Wales, Hills Flats and Lydney contain remains of the style, yet all
other finds of Lydney-Llanmelin ware have been in South Wales 191: Caldicot, Coed-y-
Cymdda, Llanmelin, Magor Pill, Sudbrook, Thornwell Farm and Twyn-y-Gaer (Cwmyoy).
Dating the ceramic tradition has proven to be difficult; chevron-decorated sherds from
Sudbrook or Caldicot were found unstratified.192 However, a possible Middle Iron Age origin
may come from a find at Coed-y-Cymdda with a radiocarbon dating of 410-90 cal. BC, while
at Twyn-y-Gaer (Cwmyoy) finds indicate a late third to first century BC date, suggesting the
end of production just before, or a consequence of, the Roman Conquest. 193 The pottery
assemblage at Magor Pill, including two-thirds of a zig-zag incised pot, has been assumed to
be contemporaneous and of a first century BC date.194 Seven items found at Llanmelin were

187
Davies (1995), 688.
188
Gwilt (2007), 303.
189
Davies (1995), 691; Moore (2007), 81.
190
Spencer, B. (1983). ‘Limestone-Tempered Pottery from South Wales in the Late Iron Age and Early
Roman Period’, Bulletin of the Board of Celtic Studies. 30, 408.
191
Cunliffe in his 3rd edition of Iron Age Communities in Britain includes ‘broadly similar’ finds at
Salmonsbury (Gloucestershire) in the Lydney-Llanmelin ware zone, however Spencer contests this
inclusion owing to similarities with Severn Valley Ware and inconsistent decorations. Cunliffe does
not address this in his 4 th edition. See Spencer (1983), 407-8; Cunliffe, B. (1991). Iron Age
Communities in Britain: An Account of England, Scotland and Wales from the Seventh Century BC
until the Roman Conquest. 3rd Edition. London: Routledge, 81.
192
Spencer (1983), 407: footnote 2.
193
Savory, H.N. (1988). ‘Prehistoric Pottery (Figs. 17-18, Nos. 1-15)’, in Owen-John, H.S. (1988a). ‘A
hill-slope enclosure in Coed y Cymdda, near Wenvoe, South Glamorgan’, Archaeologia Cambrensis.
137, 93; Probert (1976), 116-8; Gwilt (2007), 303-4.
194
Webster, P. (1989) ‘A Natural Channel at Magor Pill’, in Whittle, A.W.R. (1989). ‘Two Later
Bronze Age Occupations and an Iron Age Channel on the Gwent Foreshore’, Bulletin of the Board of
58

labelled by Hawkes as of Iron Age B date, which Spencer agrees must be around the first
century BC.195 Thus logically the peak of Lydney-Llanmelin ware production must have been
during this time; however beginning and end dates remain undefined.
Allen would argue two more instances of Lydney-Llanmelin ware outside this trade
area. However further investigation proves otherwise. 196 Alcock describes a body sherd from
Cadbury Castle in Somerset as ‘apparently with crude grooved ‘eye-brow’ ornament’, which
Allen attributes to the Lydney-Llanmelin style. However, comparison between the sherd and
an ‘eyebrow’ sherd found at Llanmelin shows a stark contrast. At Llanmelin, the ‘eye-brow’
design forms a point between curves and is of coarse material, while the Cadbury Castle sherd
is less pronounced as well as ‘highly burnished’. 197 Also, Allen associates sherds displaying
‘unaccompanied chevrons’ identified at Glastonbury with Lydney-Llanmelin ware. Yet
Bulleid states that all chevron styles found at the site consist of nested, shaded, or side-
pointing chevrons along with zig-zag patterns, none of which are consistent with the simply
decorated ware of the South Wales style. 198 Furthermore, Bulleid attributes all designs to the
Bronze Age, long before Lydney-Llanmelin ware.
There are large carboniferous limestone outcroppings in the Bristol-Mendip area
which supported the Glastonbury tradition; therefore similar limestone rich areas in South
Wales could presumably support a Silurian assemblage. Lydney-Llanmelin ware, with its
concentration of finds on the Welsh coast of the Severn, seems most likely to be that
assemblage, with hillforts on or near a limestone cropping as possible centres of extraction
(i.e. Dinas Powys situated on a narrow ridge of carboniferous limestone 199). Limestone from
the Mendips is best recognized through its contents of calcite; although calcite-tempered ware
is present in South Wales, much of the Iron Age pottery remaining in the area of the Silures is
predominately calcite free limestone, making a local origin more likely. 200
Consideration of the entire Late Iron Age pottery assemblage is beyond the scope of
this study but the importance of the Lydney-Llanmelin ware and Silurian pottery production
within the local trade network must be stressed. Calcite-tempered ware, characteristic of
Bristol-Mendip produced pottery, is found almost exclusively along the edges of the Severn

Celtic Studies. 36, 221.


195
Nash-Williams (1933), 299-302: Nos. 39,41,42,45-7,53; Spencer (1983), 406-7.
196
Allen, J.R.L. (1998). ‘Late Iron Age and Earliest Roman Calcite-tempered Ware from Sites on the
Severn Estuary Levels: Character and Distribution’, Studia Celtica. 32, 37, 40.
197
See Nash-Williams (1933), 299: No. 39; 300: Fig.47: No. 39; c.f. Alcock, L. (1980). ‘The Cadbury
Castle Sequence in the First Millennium B.C.’, Bulletin of the Board of Celtic Studies. 28, 674: KX
024.3; 676: Fig. 9: KX 024.3.
198
Bulleid, A. (1917). ‘Pottery’, in Bulleid, A. and Gray, H.S.G. (1917). The Glastonbury Lake
Village: A Full Description of the Excavations and the Relics Discovered, 1982-1907. v. 2.
Glastonbury: Antiquarian Society, 507-8.
199
Alcock, L. (1955). ‘The Hill-Fort in Cwrt-yr-Ala Park, near Dinas Powis (Glam.): I. The Defences’,
Bulletin of the Board of Celtic Studies. 16, 242.
200
Gwilt (2007), 304.
59

Estuary.201 Probable local productions of Glastonbury ware flooded the area; trade, therefore,
remained highly local.202 Yet, this perception of ‘local’ may be ‘based purely on potentially
anachronistic definitions of ‘distance’’.203 Morris perceives a three-tiered system of pottery
distribution in western Britain.204 In this system, a small scale distribution can be up to thirty
kilometres away whereas regional distribution can be up to around seventy five kilometres
from the source. The latter can be seen in the Severn-Cotswold region, where pottery and
querns from the Malvern and Mendips areas were exchanged typically up to c. 80 km away. 205
To reach South Wales from the Mendips, a journey by land through Gloucester is
manageable; however the trip would be lengthy going far beyond 80 km. A more viable route
for any merchant of the period would thus be by water across the Severn Estuary most likely
landing somewhere along the Gwent Levels.

Conclusion
Silures living within the area were in a great economic position with the rivers Usk
and Wye providing access to the north, and a small trip over the Severn Estuary leading into
England both giving prime access to complex integrated trade networks. Although the only
distinctly Silurian trade-good, Lydney-Llanmelin ware has not been definitively found across
the Bristol Channel, the presence of Dobunnic coins suggests they still had goods to sell. In
fact, the multitude of numismatic and exotic goods evidence found within Gwent, more so
than Glamorgan, indicates a strong integration within the western England exchange network.
As well, the Gwent Levels have been archaeologically portrayed as an ideal sea-borne trade
centre with multiple ports.
There are many sites along the Gwent Levels that show Iron Age occupation which
were possibly suitable for a maritime merchant to land. Recent work in the Gwent Levels
suggests landing points across the area which were probably in consistent use, as well as
trackways leading inward, as possible points of exchange for the area. 206 Sudbrook is often
seen as playing an important role in the Severn Estuary. Constructed in the beginning of the
second century BC, the hillfort stood at 3.21 ha in size and included multivallate defences. 207
During excavation, timber huts were found on the outer edges of the enclosure, suggesting the
interior was left for stores or cattle corrals. 208 Being situated on the Gwent Levels where at
low tide the estuary was only .75 miles wide, this gave a great opportunity for trade through
201
Allen (1998), 39.
202
Howell (2009), 55.
203
Moore (2007), 83.
204
Morris, E.L. (1996). ‘Iron Age Artefact Production and Exchange’, in Champion, T.C. and Collis,
J.R. (eds.) (1996). The Iron Age in Britain and Ireland: Recent Trends. Sheffield: J.R. Collis
Publications, 44.
205
Moore (2007), 80-5.
206
Rippon (1996), 7; Gwilt (2007), 306.
207
Howell and Pollard (2004), 151; Howell (2009), 40; Wiggins (2006), 19: table 4.
208
Nash-Williams (1939a), 55.
60

Sudbrook at anytime of the day. As mentioned, Iron Age brooches and Lydney-Llanmelin
ware connect Sudbrook with Lodge Wood and Llanmelin Wood Camps, possibly showing an
elite trade network working through the estuary port. 209 As well as a trading post, Sudbrook
was used as a ferry-crossing in the Late Iron Age and into the Roman Period. 210 Being an
important access point between the Silures and the outside world, controlling the port was
essential. Magor Pill was also an important estuarine landing point during the Romano-
British, Early Medieval and Early Modern periods. 211 Thornwell Farm, as well, provides
evidence of long-term occupation, from the Late Bronze Age into the Romano-British
period.212 This, coupled with the location of this site overlooking the Levels, could indicate
Thornwell Farm was an important location possibly intertwined with the Gwent trade
network, although assemblages there tend towards a low status existence.
The view of an inter-continental communications network focused upon Merthyr
Mawr Warren cannot be wholly substantiated; however there is other evidence hinting at such
a possibility.213 Armorican axes and a stamped decorated sherd similar to La Tène ceramics
complement an Augustan dated Arretine sherd found on Steepholm in the Bristol Channel.
An exchange network with Gaul, or even Mediterranean cultures seems probable, yet limiting
highly lucrative long distance trade to a single port does not seem as likely. The Gwent
Levels exhibit a multitude of evidence towards multiple landing points for coastal exchange,
as shown above. In the end, Merthyr Mawr Warren may have just been a larger or more
important port amongst many along the western coast.
Numismatic evidence is relatively scarce in Glamorgan, yet brooch finds, Twyn-y-
Gaer (Cwmyoy) notwithstanding, are almost equal in number between Glamorgan and
Gwent. Ceramics and possible sheet vessel manufacturing hint at trade and exchange
networks with the continent benefiting both sides. All evidence suggests Glamorgan had a
sustained economic system in place which most likely centred upon several points along the
coast, Merthyr Mawr Warren included. This economic system likely intertwined with the
exchange network in Gwent, but on a small scale as much of what is prominent in the Gwent
archaeological record is not as prevalent in Glamorgan (i.e. Dobunnic and Mediterranean
coins). Thus, the tribes of the Silures seemed to evolve independently but along parallel lines.
Overall, the county’s archaeological record is lacking when compared to Gwent,
nevertheless there is still strong evidence of a healthy economic system in place. Numismatic
evidence may be slim, but brooch findings are almost equal; there are even possible
connections to long-distance trade independent of Gwent. Once again, however, size betrays

209
Nash-Williams (1939a), 57-9; Gwilt (2007), 302-3, 318: Appendix 1.
210
Manning (2004), 188.
211
Allen (1998), 30.
212
Hughes (1996), 89-91; Howell (2009), 47.
213
Davies and Lynch (2000), 207; Gwilt (2007), 306-7.
61

the county, as the artefacts found in Glamorgan do not indicate a population double that of
Gwent. Thus, although an exchange system is in place providing the area with exotic goods
and demand for their local metal-goods, Gwent appears to be economically, and culturally
stronger.
Even though they may be culturally inferior to Gwent, the presence of exotic goods,
local production and even long-distance trading indicates a substantial system in place in
Glamorgan independent of their eastern neighbours. A ruling elite class almost certainly
existed, and indeed it seems quite likely as a wealthy class is almost necessary for the area to
get such economic success away from the western England trade network. More plausible is
an organization of multiple aristocratic leaders along the coast in different groupings. How
prominent of a part the western Silures played in the overall Silurian society, as well as the
likely social structure within, will be made apparent through the examination of the elite, the
possible chiefdom system and the political affiliation of the entire Silurian population.
62

Chapter 6: Silurian Social Politics


Settlement and artefactual evidence depict a society where a strong, wealthy ruling
class provides for those around them through their hillforts and ties to local and long-distance
trade. A structure may even exist with multiple aristocratic rulers within Gwent each
controlling their own region. Justification for this social system comes through the discussion
below of elite markers within the archaeological record and the plausible indication of a
chiefdom system present during the Iron Age, followed by a look into signs of a Silurian
decentralised federation.
Chiefdoms are often viewed as ‘providing an evolutionary bridge between acephalous
societies and bureaucratic states.’214 Although originally seen as part of a unilinear sequence
of stages towards modern society, this evolutionist idea has been largely criticized as
outdated, often to the detriment of chiefdoms as a whole. 215 However, chiefdoms need only to
be shown as a varying step unique for each society’s culture and situation. In all societies,
chiefdoms provide similar services: management of large-scale subsistence, storage of wealth
and food, controlling trade and providing security (usually through maintaining alliances). 216
It is then each society’s structure and environmental situation that dictates which services are
more important. Thus, recent studies have focused upon dividing chiefdoms into varying
schemes.217 Due to the worldwide variance upon chiefdoms, many of these schemes focus
upon issues unrelated outside Europe, Britain and South Wales. For that reason only those
relevant to this dissertation will be discussed.
Chiefdom origins in Europe are present from the Neolithic Age, which can be seen in
Britain from the megalithic monuments of ‘henge’ cultures still standing. The religious
identification of the monuments coupled with the high centralization of domestic labour
needed to attempt such a construction hints at a ruling class. However the lack of wealth and
social differentiation present at the monuments point towards a ‘group-oriented chiefdom’
where leaders serve group interests, not personal. 218 The later Copper and Bronze Ages show
a change to ‘individualized chiefdoms’ where wealth is portrayed in burials and hoards
through exotic objects acquired from extensive trade-networks. 219 Whether or not a chiefdom
political system was present in South Wales during these times in not positively known, yet

214
Earle, T.K. (1987). ‘Chiefdoms in Archaeological and Ethnohistorical Perspective’, Annual Review
of Anthropology. 16, 279.
215
Earle (1987), 280.
216
Johnson, A.W. and Earle, T.K. (2000). The Evolution of Human Societies: From Foraging Groups
to Agrarian State. Stanford: Stanford University Press, 267.
217
Earle (1987), 280.
218
Earle, T.K. (2002). Bronze Age Economics: The Beginnings of Political Economies. Boulder:
Westview Press, 50.
219
Earle (1987), 286.
63

any political structure in place before the drastic changes brought about by the climatic
deterioration at the end of the Bronze Age may not have subsisted into the Middle to Late
Iron Age.
When looking for the existence of a chiefdom several factors must be considered.
Earle perceives a chiefdom’s scale of integration, centrality of decision-making and act of
stratification as the defining characteristics. 220 These characteristics, however, need not be
either present or absent, but are instead interrelated variables. How or even if these variables
can be seen in the archaeological record is not always clear. Yet there are still telling signs of
societal change and reflection that may be seen archaeologically that could hint at the
presence of, or at least lay the foundation for, a chiefdom system to exist. 221 Burials are
common places to find social stratification: an elite class may show the signs of their high
status, or possibly the presence of a grave itself may allude to a privileged class. Housing
remains better portray the everyday aspect of living in any period, and if there was a strong
desire for personal differentiation, then it can be seen most readily in the record of
settlements. Evidence of this can be seen through any exotic items found that may have
required high levels of sophistication, craftsmanship, or even long-distance trading. Trade
networks and related manufacture centres often create circumstances where only those with
enough wealth to start or maintain production of trade goods can benefit. Thus, the
appearance of localized production centres can suggest an elite interaction. In societies where
conflict is ever-present, and warfare is a constant necessity, a warrior class is more than likely
to exist. These warrior elites are most often in the best situation to seize control of a society
either by force, or through the lower classes’ constant need for security. Loosely related to
this is the role of ruling classes as wilfully accepting the risk involved in gathering and
maintaining resources. A benefit gained by the lower classes for living within a governed
society is the necessity for those in power to ensure resources for those under their rule. With
the most important necessity of each day managed by another, it often leaves the lower class
with much more time for other tasks (and it is in this that complex cultures emerge). Lastly,
religion is a constant aspect of past societies, but not always the most recognizable. If a
culture’s religion dictates ritualistic activities, most notably ceremonies, then a chief often has
a role to play. Again, this can be hard to find in the archaeological record, and furthermore is
not the focus of this thesis, but nevertheless must be mentioned to further any investigation
into the presence, or absence, of a chiefdom. All of these aspects must, and will be, discussed
to see if there is any implication of a chiefdom political system in South Wales before the
Roman Conquest.

220
Earle (2002), 53.
221
Earle (2002), 56-65.
64

Evidence in the Archaeological Record


During the Iron Age, there is an evident simplification of society due to a lack of
wealth shown in the archaeological record as compared to the Bronze Age. 222 This could be
the result of the many changes in the social, economic, political and cultural structures that
probably occurred as a result of the increased flooding in the Gwent Levels and throughout
Britain during the Early and Middle Iron Ages. This does not, however, signify a society, or
societies, across South Wales without a political leader. Burials, being the predominant
method of display for wealth in the Bronze Age, were generally the best indicators of wealth
as the energy invested in the grave correlated closely with social and economic standing.
Outside factors, however, such as culture and economy could impact the social perception of
public displays of wealth in burials.223 Natural resources through agriculture and pastoral
farming may also shed light on how powerful aristocrats used these to gain and maintain
control over the regional population. Trade goods, as well, show the outside influence of
trade networks upon the area providing ever increasing wealth markers, allowing the local
elite to further publically demonstrate their status.

Burials
Although less common culturally to Iron Age tribes than during the Bronze Age,
burials still existed in South Wales. Llanmelin, the fourth largest hillfort (6.56 ha) in South
Wales, contained incomplete skeletons of a male and female in the annexe of the site. The
burials are similar to others found in England dating from the fourth century BC to the Roman
conquest, while the annexe is believed to have been constructed sometime in the first century
BC/AD, narrowing the burial’s probable age. 224 Incomplete skeletons are also found at RAF
St. Athan in the Vale of Glamorgan: two are buried in the internal bank of an enclosure,
whereas only juvenile leg bones were found in a pit nearby. 225 These three are given a Late
Iron Age radiocarbon dating of 170 BC – AD 60, while 70 metres west of the enclosure, two
probable female inhumation burials are found 5 m apart dated 400-200 BC.226 All four
incomplete skeletons discussed probably date to the first century BC/AD; however the
complete inhumations date towards the Middle Iron Age. This may show a pattern of less
reverence towards burials as the Iron Age progresses. A Late Bronze Age burial (1110-840
BC) also found at St. Athan contains no grave goods besides a large wild boar canine tooth
deliberately placed over the pelvis.227 This is unique from the typical Bronze Age burial as it

222
Earle (2002), 51.
223
Earle (1987), 286; Earle (2002), 56-7.
224
Howell and Pollard (2004), 150.
225
Barber, A., Cox, S. and Hancocks, A. (2007). ‘A Late Iron Age and Roman Farmstead at RAF St.
Athan, Vale of Glamorgan. Evaluation and Excavation 2002-03’, Archaeologia Cambrensis. 155, 64.
226
Barber et al. (2007), 64,66.
227
Barber et al. (2007), 53.
65

is void of grave goods, and is the only burial of its period in Wales without an obvious
association to a funerary monument.228 A break from Bronze Age burial traditions is evident
in South Wales, as Late Bronze Age sites are not used in the Iron Age, and the Bronze Age
practice of buried grave goods all but ceases, albeit temporarily, before resuming in the first
century BC.229 Iron Age inhumation burials were found at Mynydd Bychan, deposited almost
haphazardly, and Merthyr Mawr Warren.230 A cremation burial radiocarbon dated to 550-200
BC at Trostrey in Gwent, although considered more common among Iron Age burials, is rare
amongst the inhumation burial tradition of the period in South Wales. 231 Skeletal deformities
are also common amongst Iron Age Burials in Britain, but not in South Wales; where only
one such burial has been found at Castle Ditches (Llancarfan) showing the second and third
cervical vertebrae were fused.232
Overall, the archaeological record of burials changing over time shows the shifting
social, and possibly even religious, perception towards the act and by association the dead
themselves. For example, the burial at Castle Ditches (Llancarfan) revealed the arms and
head had been damaged, possibly from the force of placing the skeleton in a small hollow. 233
This hints at a complete disregard for the deceased’s remains. Yet, it could imply a shift from
the physical remains to other, almost spiritual, aspects of the deceased. This latter point may
be a consequence of warfare, which will be shown as culturally important to the Silures.
Because of a necessity for a quick burial, the remains may show signs of less care; however a
more spiritual reverence may have come about due to the shortened association with the
physical remains. From this, an overall religious shift may have come about to a more
spiritual reverence.
During the Late Iron Age of South Wales, incomplete skeletons almost became
standard, possibly due to a ritual use of some bones of the deceased and burial of the rest. On
the Gwent Levels, two skulls, one Middle Bronze Age and one Early Iron Age in date, were
found less than two metres apart in an area also containing animal skulls. 234 This and a
Bronze Age femur found at Chapeltump suggest a continuation of ritual practices in the
Levels partway through the Iron Age, but no further. It must be noted that the Iron Age
228
Barber et al. (2007), 105.
229
Davies and Lynch (2000), 212-4; Gwilt (2007), 316.
230
Savory, H.N. (1954a). ‘The Excavation of an Early Iron Age Fortified Settlement on Mynydd
Bychan, Llysworney (Glam.), 1949-50: Part I’, Archaeologia Cambrensis. 103, 98-9; Savory, H.N.
(1954b). ‘Early Iron Age Discoveries on Merthyr Mawr Warren (Glam.)’, Bulletin of the Board of
Celtic Studies. 16(1), 54; Gwilt (2007), 315.
231
Gwilt (2007), 315.
232
Brothwell, D. and Bourke, J.B. (1995). ‘The Human Remains from Lindow Moss 1987-8’, in
Turner, R.C. and Scaife, R.G. (eds.) (1995). Bog Babies: New Discoveries and New Perspectives.
London: British Museum Press, 56-7; Martin (2003), 353.
233
Martin, C. (2003). Iron Age Artefacts in Wales: An Investigation into the Material Culture of South-
east Wales during the Pre-Roman Iron Age. Oxford: British Archaeological Reports British, Series
353.
234
Aldhouse-Green (2004), 164; Gwilt (2007), 316.
66

skull’s dating overlaps the second half of climatic deterioration in the area, which may have
subsequently played a large part in ending the ritual practice in the area. What must be
stressed is not the perceived religious aspects surrounding this part of Silurian society, but
how different life has become for those living in South Wales at this time. This change
should not be viewed as for better or worse, but rather for the variation itself proving a shift
from Bronze Age traditions.
Burials are frequently used as measurements of social and economic differentiation
within a society. In fact, complex variability has been documented marking burials of
differing socio-economic status around the world beginning in the Middle Paleolithic Era,
sometimes even along rigid levels of social involvement and energy expenditure. 235 However,
the exclusive use of burials as status markers can be misleading as economic or religious
relationships amongst societies can alter the importance placed upon inhumations and
funerary objects.236 During times of economic surplus, burials and funerary offerings often
became ways of differentiating social class, yet during economic contraction a more outward
and personal display may have been reserved for the living leaving less available to inter with
the deceased. Burials amongst similar social classes could even differ enough to imply
contradicting social status.237 Furthermore, burials often show only a single ritual event,
whereas repeated offerings at other religious sites may have been the medium of public
displays of wealth within societies. These displays then will rarely be found within the
archaeological record.
In South Wales, the continued use of burials indicates a social connection between
the population living in the area and their deceased. Burials associated with the prominent
sites in the area (i.e. Llanmelin and Merthyr Mawr Warren) support the viewpoint stressed
herein of a plausible political connection to the inhabitants of these sites. As stated, these
sites may have been local centres for members of a ruling class plausibily of a higher social
and economic status. These political leaders may have used these burials not only for
religious purposes, but to stress further their position within the local culture and society.

235
Binford, L.R. (1971). ‘Mortuary Practices: Their Study and Their Potential’, Memoirs of the Society
for American Archaeology, No. 25: Approaches to the Social Dimensions of Mortuary Practices, 22-3;
Tainter, J.A. and Cordy, R.H. (1977). ‘An Archaeological Analysis of Social Ranking and Residence
Groups in Prehistoric Hawaii’, World Archaeology. 9(1): Island Archaeology, 95-112; Tainter, J.A.
(1980). ‘Behavior and Status in a Middle Woodland Mortuary Population from the Illinois Valley’,
American Antiquity. 45(2), 309-10; Hayden, B. (2001). ‘Richman, Poorman, Beggarman, Chief: The
Dynamics of Social Inequality’, in Price, T.D. and Feinman, G.M. (2001). Archaeology at the
Millennium: A Sourcebook. New York: Springer, 234-45.
236
Earle (2002), 57.
237
Braun, D. P. (1981). ‘A Critique of Some Recent North American Mortuary Studies’, American
Antiquity. 46(2), 411.
67

Settlements
Viewing social distinction through residential housing and then comparing with
burials may provide better overall understanding, possibly eliminating any doubts arising
through the study of burials alone. ‘In essence, housing involves a daily use and display
function much more likely to represent economic and political relationships than burials.’ 238
Llanmelin is one of the largest hillforts in South Wales at 6.56 ha, and Castle Ditches
(Llancarfan) is similarly large at 4.2 ha. Contrasting this are Mynydd Bychan and St Athan,
both 0.3 ha in size. Hillforts of great size are logical places for centralizing societies and thus
can contain connections to elite and ruling classes. Therefore, a burial at a larger hillfort may
be a display of a higher class. A burial at a smaller settlement site may imply a ruling class
placing less importance upon outward signs of wealth, a disassociation from the ruling class
altogether, a society where traditions associated with death are extended to all social classes,
or even a varying social system settled in smaller enclosures. These small defended
enclosures less than 0.5 ha in size can contain multiple roundhouses suggesting Glamorgan
could have exhibited such an elite class living within smaller settlements. 239 As well, Castle
Ditches (Llancarfan), Mynydd Bychan and St Athan are all in the Vale of Glamorgan, and
following Flannery’s theory on village evolution, the three may even be connected through
extended family ties.240
Pastoral farming has been demonstrated as the predominant method of resource for
the Silures and may also provide clues to local social structures. 241 This pastoralist nature
may have been more socially integrated than just a simple means for food. Over half of
known Gwent hillforts are within 2 km of good pasture land, and the vast array of cattle hoof
prints found in the Gwent Levels could be linked to seasonal cattle byres. With the Levels as
possible trading centres, it is possible that livestock could have been exchanged as well, even
if only on a local level. Intensive livestock farming, especially of a transhumance nature,
easily leads to a reluctance to share amongst other outside family groups. Closed settlement
patterns probably appeared quickly after the pastoral lifestyle was adopted, while seasonal
houses being built on private uplands were best as unenclosed settlements. Signs of
pastoralism in north-west Wales come from only single huts found above 200 m. 242 A similar
system may have existed in South Wales, but signs of single huts, likely unenclosed, may
have been lost over time.
Cunliffe sees possible connections between livestock ranch boundaries and hillfort
placements, where cattle may have become another manifestation of wealth, although whether

238
Earle (2002), 57.
239
Davies and Lynch (2000), 146; Howell (2009), 48.
240
Flannery (2002), 417-8.
241
See above, above.
242
Davies and Lynch (2000), 162.
68

it was a cause or an effect of local social change is not known. 243 Quarley Hill in Hampshire
was built over a system of livestock boundaries, while Rams Hill in Berkshire and
Thundersbarrow Hill in Sussex both show earlier pastoral enclosures later incorporated into
defended enclosures. These may demonstrate emerging powers building sites to protect their
property, or further substantiate their claim. Conversely, Sidbury in Wiltshire and Danebury,
Hampshire, both show later boundary enclosures built around and connecting with the
hillfort. There, a growing significance of livestock as wealth displays may have led the
inhabitants to construct enclosed ranches. Either way, there is a desire for raising and
maintaining large herds in nearby enclosed fields. Thus, Silurian pastoralism may have been
equally important as a display of wealth in South Wales.
Similar annexes are found within Silurian territory. Most notably, in Glamorgan, Cil
Ifor Top, the largest (2.9 ha) and most central site in Gower, has an associated lobate annexe,
while Caer Dynnaf (3.8 ha), also centrally located, has two: one detached and a cross-ridge
field similar to Twyn-y-Gaer (Cwmyoy) which has been connected to animal husbandry. 244 Y
Bwlwarcau, the largest hillfort in Glamorgan (7.2 ha), also has an associated concentric
annexe. In Gwent, seven sites over 2 ha in size are linked with annexes: Great House Camp
(2.37), Twyn-y-Gaer (Cwmyoy) (2.98), Gaer Fawr (3.5), Gaer (3.5), Pen Twyn (4.17),
Llanmelin (6.56) and Tredegar Hillfort (9.28). Twyn-y-Gaer (Cwmyoy) and Pen Twyn are
prominent sites guarding the river Honddu valley, while the sheer size of Tredegar suggests
the site’s significance within the Gwent social system. Llanmelin’s importance has been
reiterated as archaeological evidence mentioned throughout this thesis has repeatedly been
associated with the site. In addition, multivallate enclosures have even been attributed to a
reflection of greater commitment to animal husbandry, with univallate enclosures focusing
more upon agriculture.245 In South Wales, the defences are known at twenty-five of twenty-
nine sites with associated annexes. Of these, only nine are univallate, while the rest choose
stronger defences, quite possibly for better protection for the livestock, reinforcing the idea of
animal husbandry as a sign of wealth. Whether or not annexes can be certain markers for
elite occupation, all of these Silurian sites show logical centres for local aristocratic rule. It
seems fitting enough, however, that a controlling class would use an important aspect of
Silurian life to further substantiate their position.

Artefacts
The best material evidence of Iron Age South Wales comes from artefact deposition.
Items found may have been left for later retrieval, which obviously did not transpire, or, most

243
Cunliffe (1971), 55, 62-3; Cunliffe (2005), 419.
244
Probert (1976), 109.
245
Olding (2000), 73.
69

likely, were deliberate acts of abandonment, sacrifice, or propitiation. 246 A thorough review
on artefactual evidence in South Wales was completed recently by Martin, during which
items were classified by cultural function, while Gwilt brings the collection up to date in his
more recent publication.247 This brings about a better overall understanding of which aspects
of life were either more revered or were given higher cultural importance by the Silures
during the Iron Age. Artefacts found in South Wales will be viewed more closely in relation
to each function that they served. Then, all of these artefacts will be discussed with regard to
the societal structure that possibly existed during the end of the Iron Age, and how they may
imply the presence of a chief or elite ruling class.
Even though there is a ‘paucity of survey, excavation, and research’ 248 on South Iron
Age Wales, there is suitable artefactual evidence for supposing an individualized elite Silurian
class which most probably ruled the area collectively. While larger housing and burials are
strong indications, residential housing associated with an elite status is often defined by exotic
craft goods found within.249 Artefacts showing signs of sophistication, production, or long-
distance trade hint at a high class population focused upon exotic goods as displays of wealth.
Coin finds in South Wales have been demonstrated as mostly showing the exchange network
with other British tribes, yet they may also exhibit the desire of the Silures for the gold coins
as displays of wealth, gifts for exchange, or even mediums of personal or spiritual
expression.250 Although most coins are of British origin, Mediterranean and Gaulish coins
have been found in Gwent.251 A Central Gaulish tête diabolique potin found at Llanmelin
may be of Iron Age stratigraphic date. Yet, most coins of origin outside Britain are found at
locations with Roman connotations. This highlights the possibility of a Roman period
context, but does not prevent a pre-Roman date for their usage.
La Tène brooch remnants are abundantly found within Iron Age contexts. Since
brooches of these types travelled long distances through trade networks with either eastern
England or by sea around Cornwall, the importance placed upon the items must have been
great, and the high demand indicates the existence of an elite class. This is supported by the
locations at which brooches have been found in the archaeological record. Prominent sites of
larger size and complexity exhibit brooch findings in eastern South Wales (Lodge Wood
Camp, Llanmelin, Sudbrook and Twyn-y-Gaer (Cwmyoy)), while brooches found in the west
246
Hingley, R. (2006). ‘The Deposition of Iron Objects during the Later Prehistoric and Roman
Periods: Contextual Analysis and the Significance of Iron’, Britannia. 37, 214-6
247
Martin (2003), 33; Gwilt (2007), 297-328.
248
Gwilt (2007), 297.
249
Plog, F. and Upham, S. (1983). ‘The Analysis of Prehistoric Political Organization’, in Tooker, E.
(ed.) (1983). The Development of Political Organization in Native North America. Seattle: University
of Washington Press, 206; Costin, C.L. and Earle, T. (1989). ‘Status Distinction and Legitimation of
Power as Reflected in Changing Patterns of Consumption in Late Prehispanic Peru’, American
Antiquity. 54(4), 691-714.
250
Aldhouse-Green (2004), 168.
251
Gwilt (2007), 304-6; 321: Appendix 3.
70

tend towards sites of less physically displayed importance (Merthyr Mawr Warren, Whitton).
Thus, a change in settlement preference is exhibited where those of the eastern Silures who
could afford exotic goods chose to live within larger enclosures, and their western
counterparts decided upon more humble dwellings.
An Iron Age metalworking tradition is apparent in artefacts found from the western
tip of Gower spanning South Wales into Gwent. 252 Davies and Lynch have proposed high-
status metalwork found outside the area at Moel Hiraddug and Pen-coed-Foel as evidence of
elites in power.253 The evidence indicates that the same existed in South Wales. The Worm’s
Head stone moulds have been mentioned as possible evidence of sheet vessel manufacture
(e.g. cauldrons and bowls), while crucibles for bronze production have been found at
Llanmelin in Gwent and Candlestone Castle, Castle Ditches (Llancarfan) and Merthyr Mawr
Warren in Glamorgan.254 Excavations at Llanmelin also uncovered two ribbed bronze
bracelets, possibly manufactured at the site. 255 Production based artefacts, such as ingots and
billets, have also been found throughout South Wales, including Coed-y-Cymdda, Merthyr
Mawr Warren, Mynydd Bychan, Sudbrook and Twyn-y-Gaer (Cwmyoy). 256 The Seven
Sisters Hoard (Illus. 1), found near Neath in 1875, and the Lesser Garth-Pentyrch Hoard, from
Pentyrch near Cardiff, are both dated to the Late Iron Age and contain production artefacts;
the significance of which is obvious when compared to the Llyn Fawr Hoard and the Cardiff
Hoard which both date from the Late Bronze Age/Early Iron Age period and contain no
production artefacts.257 The existence of these hoards could have been through abandonment
or even left for later retrieval, yet a religious act of sacrifice or even propitiation seems most
plausible. Either way, the presence of production specific artefacts suggests an increase in
local production of metal material culture goods which had then become a more significant
part of Silurian life. Additionally, a strong warrior elite material culture is present in South
Wales which relies upon bronze and iron objects for status displays.
Ceramic production in South Wales is present in most excavated sites. Gwilt has
stressed the lack of a ‘regional ceramic synthesis’ of Iron Age South Wales, as well as the
persistence of recent ceramic work to focus upon already defined periods. 258 As mentioned
above, Lydney-Llanmelin ware may comprise the closest example of a Silurian pottery
tradition peaking in production around the end of the first millennium BC. Finds of this type
within Silurian territory are mostly confined to Gwent, with Coed-y-Cymdda being the only
Glamorgan example. This site is in the eastern Vale of Glamorgan, near Cardiff; thus the

252
Howell (2009), 58; Gwilt (2007), 306-9; 315.
253
Davies and Lynch (2000), 162.
254
Morris (1996), 56: table 5.6.
255
Aldhouse-Green (2004), 166.
256
Martin (2003).
257
Martin (2003).
258
Gwilt (2007), 303.
71

close proximity to Gwent can account for this presence. Since there is currently no evidence
of Lydney-Llanmelin ware reaching the western half of South Wales, it must be assumed that
pottery in that area was from outside the cultural influence, yet it does not suggest local
pottery was wholly excluded. Therefore, local pottery production seemingly persisted
exclusively in the east. An elite class in Gwent may have controlled or financed the
production and trade of Lydney-Llanmelin ware with possible production centres at
Thornwell Farm and/or Llanmelin as excavations at both sites yielded ceramic assemblages
among the largest in South Wales. 259 With Malvern ware finds scarce in the area of the
Silures, a demand for pottery may have existed, and although Glastonbury type is found at
Silurian sites, a locally produced pottery would have proved to be cheaper to trade than that
which was brought in from across the Bristol Channel. Sudbrook also contained a large
amount of sherds but giving the location on the Gwent Levels coast, these were probably due
more so to trade than production.
Exotic goods have been found throughout Silurian territory. These markers of an
elite class not only indicate a widespread influence, but when found repeatedly at the same
sites, further stresses the importance of those sites within the Silurian material culture. All of
the above mentioned objects which indicate an elite presence within the area (e.g. burials,
coins, brooches and metal and ceramic production) have been found at Llanmelin in Gwent.
The fourth largest site in South Wales at 6.56 ha, Llanmelin was of complex shape with
multivallate defences and lobate and detached annexe enclosures. Size alone hints at an
importance of at least a regional scale, yet the evidence for elite occupation and burial
suggests far more. Sudbrook also contains elite markers in brooch remnants and evidence for
metal and ceramic production. However with the coastal site within 7 km of Llanmelin,
Sudbrook may have had an important connection with the larger hillfort as a port for the local
trade network. As mentioned, Twyn-y-Gaer (Cwmyoy) may have been a site of cultural
assimilation and with similar findings of heavy brooch influence and metal and ceramics, the
location may have served more important purposes in the northern Silurian territory. In
Glamorgan, Merthyr Mawr Warren exhibits metalworking and brooch remnants, the latter of
which has led to possible intercontinental trade connections, as well as burial remains. More
work needs to be done on the western Bridgend site, but the possible importance within the
western Silurian territory has already been documented.
Cultural material can be found throughout Britain as a medium for social display.
Exotic goods did not begin reaching tribes in Scotland until the end of the first millennium
BC, which then initial contact was in the form of La Tène brooches and Irish-style ring-
headed pins.260 The interest in exotic displays of wealth continued into the first century AD,
259
Woodward (1996), 36-45; Gwilt (2007), 303
260
Hunter, F. (2007). ‘Artefacts, Regions and Identities in the Northern British Iron Age’, in
Haselgrove and Moore (2007), 293.
72

and included beads and torcs as well as the tradition of hoard depositions. Hoarding is also
evident in Norfolk, and coupled with other archaeological findings, the area displays similar
preference towards torcs as well as a Silurian-esque interest in horse equipment. 261 A social
change is documented during the end of the first century BC, which can be seen in the
relationship between hoarding practices and landscape, including sites away from isolated
locations. A further examination into the Silurian hoards within the surrounding landscape
may be able to give insight into the local relationship. A broad assessment of Iron Age
artefacts found throughout Britain may show similar cultural preferences existed outside of
South Wales. There were 3,226 artefacts of certain or possible Iron Age date found through
the Portable Antiquities Scheme from September 1997 to April 2003. 262 Coins were found the
most (1565) followed by brooches (651), which indicate a comparable interest in personal
display as demonstrated in Wales. Horse and vehicle equipment artefacts were also
prominent among the findings (215), with large caches in Norfolk (57), Suffolk (37) and
Lincolnshire (24). Obviously, anything found from the Portable Antiquities Scheme is
suspect at best when placed within an archaeological context, yet they cannot be all together
ignored, as the cultural material still gives a small insight into the culture of the period.
Overall, Silurian preference towards highly valued objects, as well as an overt fondness for
warrior and equestrian related items, was parallel to that found in other cultures throughout
Britain.

Trade
As mentioned, Merthyr Mawr Warren has been linked with Armorica in a possible
intercontinental trade network, while Sudbrook as well has been shown to be an invaluable
port for trade and transportation across the estuary. Throughout the Gwent Levels there may
have been multiple landing places for all trade and communication networks. The Gower
peninsula, geographically most suitable for sea borne trade, may have exchanged metal-
worked goods through their local productions, apparent in the stone moulds found at Worm’s
Head, although evidence of such sea borne trade is scarce. It is also evident that trade
networks were essential to the Silures for many of their resources, revolving mostly around
exotic goods prized by the higher classes. Trade networks may also have served the
aristocracy as a form of control over the lower classes.
In South Wales, the local controlling elite may have arisen through, or found
themselves in, an economic power system through controlling local production and trade.
Silurian metal and ceramic production provides wealth and trade goods to be exchanged with

261
Hutcheson, N. (2007). ‘An Archaeological Investigation of Later Iron Age Norfolk: Analysing
Hoarding Patterns across the Landscape’, in Haselgrove and Moore (2007), 368-9.
262
Worrell, S. (2007). ‘Detecting the Later Iron Age: A View from the Portable Antiquities Scheme’, in
Haselgrove and Moore (2007), 371-88.
73

other British tribes through the Dobunnic network or over sea to the continent and Armorica.
From this exchange, brooches and other luxury goods arrive which are either kept for
personal display or sold to facilitate the original production costs. Hillforts positioned near
resources (i.e. Dinas Powys situated on a narrow ridge of carboniferous limestone 263) could
have been used to control the local resources allowing only elite owned production. This
system could have supplemented a substantial elite class, including possibly a higher status
for the specialists hired for production264. Unfortunately not much artefactual evidence of
certified Silurian origin has been found far from South Wales; yet, this does not exclude the
possibility. Silurian products may not have had definitive characteristics which prohibits any
connection with the area (Lydney-Llanmelin ware excluded). Alternatively, the Silures may
have been late getting into the long distance trade network, which then never progressed due
to the Roman Conquest. Lydney-Llanmelin ware, definitively Silurian, has been radio-carbon
dated to the last quarter of the first millennium BC. This Late Iron Age origin coupled with
an already saturated ceramic market (e.g. Malvern ware and Glastonbury ware) may have
limited the local pottery from establishing long-distance networks.
Davies and Lynch place the Silures within one of two socio-economic systems in
Wales: a redistributive economy.265 In this system, chiefs show their dominance through the
goods they possess and the size of the hillforts in which they live, and as well demand
contributions of surplus goods from those nearby in exchange for services, most likely
protection or other goods.266 This surplus of goods could then be used to facilitate
associations with surrounding groups to maintain exchange relationships for the benefit of all
involved. These benefits could range from supplying locals with long-distance goods, helping
during bad agricultural periods, or even keeping alliances against others. Controlling the
local trade network would allow further regulation of goods, both domestic and exotic, and
provide added power for the Silurian chiefs.
Trade may not have been the only service supplied by the ruling aristocracy. Cunliffe
suggests two subsystems may have co-existed linked by obligation and clientage: where the
elite maintained control of animal herds (e.g. cattle) and the low status freeholders managed
the agriculture.267 Extending beyond this, since a pastoral nature seems more suited to the
Silurian lifestyle, the lower classed natives could have even been tasked to care for the local
aristocrat’s herd through obligation or as payment for the elite leader’s protection and
resources.

263
Alcock (1955), 242.
264
Aldhouse-Green (2004), 166.
265
Davies and Lynch (2000), 219.
266
Cunliffe, B. (2003). Danebury Hillfort. Stroud: Tempus, 142-4. Cunliffe (2005), 590-4.
267
Cunliffe (2004), 594.
74

Warrior Elite
An ever growing picture is emerging where a social system centred upon an elite
aristocracy provides for the population within their settlement group. This provision most
likely included necessities such as resources and security. This latter point may have been
achieved through negotiating and balancing alliances with neighbouring groups or through
drastic measures such as warfare. In South Wales, artefactual remains have been found that
indicate an elite society, but there are also developing ideas, with origins in Iron Age Britain,
of upper classes participating in warrior-based cultures. In these cultures the higher classes
often wore elaborate decorative attire while riding horseback or in a chariot with their mount
equally adorned in only the most expensive displays. Not much evidence supports such a
luxurious existence for the wealthy Silures, but some objects found in the archaeological
record do suggest a desire existed. In addition to the bracelets, brooches and other elite
adornments mentioned above, a Silurian upper class lifestyle has also been substantially
linked with battle, horses and chariots.268 At Old Castle Down in the Vale of Glamorgan iron
weapons and gold and silver decorated bronze helmets were found dating to the end of the
first millennium BC, while the Seven Sisters Hoard also contains a helmet crest knob and
hooks likely used on a warrior’s cloak. Twyn-y-Gaer (Cwmyoy) yielded a sword and
spearhead, both of iron and of third or second century BC date, reinforcing the connection of
the hillfort’s cultural revolution to that of a military nature. Horse-related equipment is also
quite prominent in the South Wales archaeological record, as terrets, harness mounts, bridle
bits, linch pins and various other cavalry-based artefacts have been discovered with elaborate
decoration and enamelling (Illus. 2 & 3). An exaggerated bull’s head, thought to be a
harness- or chariot-fitting, illustrates the use of animal-motifs as elite metalwork decoration.
These objects date throughout the Iron Age with increasing preference in the Late Iron Age,
showing a growing cultural predilection towards horses and battle decoration. Similar
chariot-based warfare systems were in place elsewhere in Britain. Native Britons used
269
chariots against Caesar during the first century BC. Chariots heavily strengthened
Boudicca’s rebellion of AD 60/1, as well as Calgacus’ clash against Agricola a quarter-
century later.270 Thus a predilection towards, and even reverence for, chariots was an
established cultural aspect throughout Iron Age Britain before the Roman period.
On occasion for celebration, perhaps after a victory in battle, or other events requiring
dress attire, items of personal adornment have been found in the area showcasing the use of
non-warfare elite goods.271 The bronze bracelets of Llanmelin along with electrum and gold

268
Davies and Lynch (2000), 191; Martin (2003); Aldhouse-Green (2004), 166-7; Howell (2009), 21-5,
60-4; Gwilt (2007), 309-312.
269
Caesar. Gallic Wars. 4.24; 4.33; 5.16-17
270
Tacitus. Annals.14.35; Sealey (2004), 51.
271
Martin (2003); Howell (2009), 61; Gwilt (2007), 313-4.
75

bracelets found at Merthyr Mawr Warren and near Llantwit Major respectively, as well as
rings found near Cardiff and Mynydd Bychan are all examples of this. Feasting and
consumption practices can also be deduced, albeit through a scarcity of evidence, likely
corresponding to these elite events. Tankards appear in the Late Iron Age as an easily
apparent vessel of most likely communal drinking habits. That they continue in use into the
Romano-British period is not a surprise; however it should be considered when dating any
tankard remains. Tankard handles (Illus. 4) have been found throughout Silurian territory
including many Roman era sites, but those found in the Seven Sisters Hoard (Illus. 1) can
probably be best identified with the Silures, as the act of depositing hoards was not a typical
Roman practice. Furthermore, one important note should be made that the abundance of
tankard examples coupled with the relative lack of early Roman amphorae indicates a Silurian
preference for beer over wine; possibly passed around a table or group as opposed to the more
individualised Roman style of drinking. Other communal feasting evidence can come from
cauldrons and the associated goods, although this may be an Early Iron Age biased viewpoint.
Fragments of cauldrons and bowls have been found at Llyn Fawr and Llanmaes, both of the
Early Iron Age in date, but cauldron remains in the first century AD Seven Sisters and Lesser
Garth-Pentyrch hoards show a continual use into this period. Thus, the archaeological record
may show an Iron Age communal feasting practice that persisted from the early period into
the Roman era. These feasts likely included all Silures at certain important, presumably ritual
events; however in all probability the elite also hosted many small gatherings where they
could further perpetuate their status among those of the same class.

Conclusion
Searching for a political system within the South Wales archaeological record has
revealed much evidence of a hierarchal society in place led by an elite upper class. This
aristocracy may have placed themselves in powerful, self-perpetuating positions where they
provide for others any required necessities in exchange for goods that can then be used
towards supplying themselves with exotic items for personal display. As well, artefacts of the
Iron Age Silures have pointed towards a predilection for combat related objects, suggesting
this possible warrior elite may have gained and maintained their prestige through warfare.
Now that an outline has been delineated along which the evolution of Silurian social
and political systems may have progressed, the plausible end result needs to be discussed.
Although powerful leaders are presumably in place in South Wales, it is widely believed that
all of Silurian territory is not under the control of a single person or family. In fact, most
likely a decentralised federation is in place which connects multiple settlement groups,
possibly even under different forms of local government, within a similar material and
religious culture.
76
77

Chapter 7: Defining a Silurian Political System


The natural landscape of South Wales, with the Bristol Channel on the south and the
scarcely populated uplands to the north, confines much of the settlement along the coast.
With the over 100 km separating the eastern and western most points of the Silurian territory,
that distance as well as the natural topography lends to a detached lifestyle from those on
either end. Using those in the centre as a liaison between the two ends is possible, but seems
quite unlikely. A centrally located site with access to all parts of the territory would be best
for any single Silurian state; however there is no evidence to suggest such a site.
A centrally located site within each region would most likely be sufficient for a ruling
aristocracy within a developed hillfort to provide all the necessary functions to support a
population (e.g. social, political, religious, agricultural/pastoral, or as a source of refuge
during tough times272). In the previous chapter, it was shown archaeologically that a
chiefdom system was most likely in place within South Wales, rather than a communal
organisation. The very effort involved in constructing such oversized enclosures hints at a
structure or authority capable of mobilising and coordinating the necessary resources and
labour.273
As well, the evidence has shown that although there are similar findings in all aspects
(e.g. settlement; production and trade goods) throughout South Wales, different lifestyles
exist with diverging needs in regards to settlement size and location, goods production and
trade intensification. Yet, there should be little doubt that all of the Silures shared a material
culture system where the same items were prized throughout (e.g. brooches and horse
equipment). Therefore, a decentralised coalition seems most plausible for the separate yet
shared lifestyles present in the area. One such avenue for connecting these independent clans
together may come from a local to sub-regional intervisibility system interlinking hillforts
across long distances.

Possible Clan Groupings


As mentioned above, early hillforts in the Welsh Marches north and east of the
Silurian territory have often been explained as reflecting control of resources, the resulting
surplus and a dominant warrior aristocracy. 274 The Silures have been shown to favour a
pastoralist society supplemented with agriculture and surpluses controlled by the ruling class
divided up as necessary. With the trackways and cattle hoof prints present at the Gwent
Levels, the possibility of a surplus cattle trade cannot be disregarded. Along with animals,
trade networks exchanged necessary and exotic goods, which could have been controlled by
272
Davies and Lynch (2000), 162.
273
Children and Nash (2001), 138.
274
Davies and Lynch (2000), 151.
78

the elite as well through a system perpetuating the status of the wealthy class. A wealthy
warrior ruling class could have arisen through the need for security, the importance of which
is apparent through the dominance of defended enclosures, or a situational leadership
developed around an intrinsic warfare aspect of Silurian culture shown at Twyn-y-Gaer
(Cwmyoy).
A separate situation developed in western South Wales, evident through the differing
archaeological record and prevalence of smaller hillfort sizes associated with elite status
markers. Glamorgan varies from Gwent in the natural landscape rising to uplands throughout
central South Wales. Gwent remains at a low elevation throughout most of the county, only
rising in the north and north-west. This allowed Gwent to keep a broader settlement pattern
across the county while the majority of the Glamorgan-based Silures remained close to the
shore. There the settlement pattern becomes closer and more personal, yet large settlements
may still play central roles in Glamorgan, only at a more local level.
Second to first century BC settlements in the Sussex Downs show an emerging
pattern of sites which dominate territories of 25-40 square miles, usually bordered by river
valleys; a similar situation occurred in Wiltshire and the Chilterns. 275 These patterns were
found by deducing the timeline of usage for each site. The same could be accomplished in
South Wales with further investigating and reliable dating sources. Until then, just looking at
the settlement pattern and the locations of the larger hillforts of South Wales can give some
initial ideas (Map 5).
In Gwent, the Usk river served a central role as many settlements were situated
nearby. Tredegar Hillfort, the second largest in South Wales, was near the mouth of the Usk,
while further upstream sat the pairing of Priory Wood and Lodge Wood east and west of the
river, respectively. Even further north, Coed-y-Bwnydd and Llancayo are situated closely,
both on the Usk’s east bank. These could have easily been connected through transport or
communication networks along the river valley. 276 Llanmelin and Sudbrook have been
repeatedly shown as an important Silurian cluster, likely serving as the economic and cultural
centres of Gwent, if not all of South Wales. Furthermore, Twyn-y-Gaer (Cwmyoy) and Pen
Twyn, situated respectively west and east of the river Honddu, guard not only the river but the
South corner of the Black Mountains. Local ruling classes could have existed in each of these
pairings, plus the other larger sites found throughout Gwent.
Glamorgan best demonstrates a loose association of local rulers due to its larger area
and coastal settlement pattern. The Vale of Glamorgan contains six large hillforts over 2 ha
in size possibly suggesting up to six separate logical groupings. In the east of the area, a
group near the Taff river may have existed with Caerau Camp, Ely as the central enclosure.

275
Cunliffe (1971), 59.
276
See below, page 81.
79

Map 5: Possible Iron Age Clan Groupings


©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved. Settlement locations taken from Wiggins (2006),
Gerrard et al (2006), RCAHMW (1986) and Silvester (2007). © Annotations Copyright GGAT/Cadw.

This area includes parts of Cardiff, where, as explained, the archaeological record has been
lost to modern development. West of Cardiff is a larger grouping near the Castle Ditches
(Llancarfan) and Porthkerry Bulwarks sites, which were most likely clustered with the latter
coastal site supporting the former inland location. Settlement patterns grouped around each
of the Caer Dynnaf, Castle Ditches (Llantwit Major) and Dunraven hillforts suggest separate
communities in those locations as well.
A preference in Glamorgan towards local connections away from the Gwent
communities seems most probable as well due to the relative lack of settlements between
central Gwent and the main settlements of the Vale of Glamorgan area. It must be reiterated
that modern development has built on top of past societies, creating gaps in the archaeological
record. A similar separation may exist immediately east of the Gower community, and,
although less likely, on the modern Neath Port Talbot/Bridgend county borough border. East
of the Ogmore river a group around the Coedymwstwr hillfort seems probable, although there
may be some community overlap with those around the Caer Dynnaf site. West of the
Ogmore, a lack of a larger hillfort may be due to a different settlement system, or a larger site
lost to the development of the city of Bridgend. Although, with Y Bwlwarcau and Mynydd-
y-Castell in the north and the Dunraven hillfort in the south, these sites may have been
connected to one or both communities. As Y Bwlwarcau (7.2 ha) and Dunraven (6.5 ha) are
the two largest sites in Glamorgan, possibly built to such a size as to incorporate sites further
80

away, this latter possibility should not be dismissed. Settlements around the rivers (from east
to west) Afan, Nedd and Tawe are also lacking a central larger settlement which may be due
to similar circumstances as those near Bridgend. The pair of Y Bwlwarcau and Mynydd-y-
Castell may have also had influence on those to the north, possibly extending past the rivers,
or a lost hillfort within the areas of Neath and Swansea may also have affected the settlement
pattern in the area. In the west, Cil Ifor Top is situated in a perfect central location to preside
over the Gower Peninsula, while in the uplands, Craig-y-Dinas sits on the lower edge of the
Brecon Beacons near the confluence of tributaries into the Afon Nedd. In Brecon, Slwch
Tump and Pen-y-crug are situated within the valley with views overlooking the surrounding
uplands giving the sites great influence over the area. Whether or not the two sites are
contemporary is unknown, but a pairing of the two may be similar to Llanmelin/Sudbrook and
Y Bwlwarcau/Mynydd-y-Castell.
These aristocratic powers of South Wales could have worked independently or
cooperated with each other through shared cultural, economic, religious or even familial ties.
It seems unlikely, but not impossible, for a single ruling group or individual to have had
jurisdiction over all of Gwent from a single site. Yet if the ruling system was more complex
than believed, a hierarchy may even have existed within the aristocratic class. A system like
this would be hard to define from the archaeological record, yet has been seen to exist
elsewhere.277 During Caesar’s second invasion in 54 BC, Cassivellaunus, a British king, ‘sent
envoys to Kent ordering the four kings of that region, Cingetorix, Carvilius, Taximagulus and
Segovax, to collect all their troops and make a surprise attack on the naval camp.’ 278 This
suggests a perceived, if not defined, hierarchy of local kings within England at that time. The
Brigantes, in northern England, have most recently been viewed as a hierarchical system in
the first century AD led by Queen Cartimandua after the Roman conquest of the area. 279
Besides the Brigantes, the Iceni are also believed to have consisted of three or more sub-
groups of similar material culture.280 During the revolt of Boudicca in AD 60/61, the allies of
the Iceni were no doubt led by their warrior leaders, but with Boudicca in command, this
hierarchical system was maintained. Thus, a similar hierarchical system, where some rulers
have more power over other rulers, may have been in place in South Wales at the end of the
Iron Age.
These instances of settlement groups apparently situated around or centred upon
larger hillforts are only theoretical possibilities. Much more work would need to be done to
verify any likelihood of definite regional settlement groups, although it still remains that some

277
Malinowski, B. (1922) Argonauts of the Western Pacific. New York: Dutton.
278
Caesar. Gallic Wars. 5.22.
279
Higham (1987), 1-19.
280
Warner (1996), 29-31; Sealey (2004), 10.
81

evidence has been irrevocably destroyed by the subsequent generations that have settled
within South Wales over the last two millennium.

Long-distance Communication
Communication networks are a key aspect of maintaining group connections among
locations at long distances apart. Natural routes along the mountain ridges of Glamorgan
have been associated with Bronze Age communication and travel. 281 Often, these routes are
near prehistoric stone markers, suggesting an older origin to the networks. There is little
doubt that these routes were continually in use up to the Roman Period and beyond as the
modern A48 overlays a Roman road which itself followed a trackway designated by standing
stones of prehistoric date. As well as routes for physical traffic, long-distance networks could
have reached across and along entire valleys through simple line-of-sight communication.
These systems enabled connections to extend farther distances and take less time to relay
messages, allowing for a controlling party to maintain power over a larger area.
Intervisibility between hillforts along the Usk and Wye river valleys in Gwent show
such a circumstance from which a single power could maintain control over a long distance,
or multiple rulers in a hierarchy or alliance could remain well connected. 282 Work on
intervisibility has not been attempted among Glamorgan sites, but some visual connections
between hillforts should logically be assumed, and visual communications could have then
developed there as well. Most likely, any network in Glamorgan would appear centred upon
hillforts in a valley, as is found in Gwent. The uplands of South Wales could also have been
utilised for long-distance communication since the high elevations allow for longer lines of
sight. A similar situation on the Gower Peninsula may have existed focused upon the
centrally located Cil Ifor Top hillfort. Howell has found pairings and clusters of hillforts,
particularly at river or valley entrances, which imply sophisticated planning to utilise
intervisibility.283 In Gwent, Llanmelin and Sudbrook are an obvious pair, while Hatterall Hill,
Pen Twyn, Twyn-y-Gaer (Cwmyoy) and Ysgyryd Gawr form a cluster at the edge of the
Black Mountains around the river Honddu. Near the mouth of the Ogmore, in modern
Bridgend, two clusters of three sites sit, including Merthyr Mawr Warren, Flemings Down
and Old Castle Down. Given the archaeological record found at Merthyr Mawr Warren and
Old Castle Down, this area may be the seat of a sub-regional, clan-based tribal federation; as
well as one at Llanmelin, which has been repeatedly suggested throughout this dissertation.
These communication networks only go as far as suggesting connections between
locations covering long distances. This implies that some chiefdom systems or regional

281
RCAHMW. (1976a). An Inventory of the Ancient Monuments in Glamorgan, Vol. I: Pre-Norman.
Pt. I: The Stone and Bronze Ages. Cardiff: HMSO, 8-10.
282
Thomas (2000).
283
Howell (2009), 49.
82

hierarchies could have existed including more groups of settlements. Through this South
Wales could have been organised into a smaller number of political groupings using
communication networks to provide for those further away, possibly incorporating multiple
larger hillforts within the same political structure.

Conclusion
Although in all likelihood the exact tribal boundaries of the Silures will never be
positively known, substantial work has been expressed towards the limits being placed at the
Wye river in the east, Loughor river in the west and the Usk river valley in the north.
Artefactual evidence and settlement patterns provide solid evidence for such a territorial
area.284
Possible evolutionary trends of the South Wales social, political and economic
systems have been established; the origins of which has arisen within the archaeological
record. While it is obvious that a social differentiation existed, the avenues from which they
evolved are varying, and in this development may be found the reasoning behind the differing
settlement record found throughout South Wales. Subsequently, these diverse settlement
patterns may have contributed to divergent social situations from which emerged contrasting
lifestyles. In other words, Iron Age South Wales is invariably contrasting within itself, and
although a similar material culture system has been found, the same cannot be said for the
social patterns.
Gwent is mostly lowlands with a single river flowing through the centre and an
economically dominant society nearby. The Vale of Glamorgan through Neath Port Talbot is
mostly coastal lowland settlement area, while the Gower peninsula is a prime location for an
isolated community with a sea-based lifestyle. The Brecon area is within a valley occupied
by multiple river convergences from the surrounding uplands. Populations living within these
starkly different environments were to inevitably lead contrasting lives; yet a shared material
culture is prominent throughout, providing a seemingly unifying tradition.
This culture may have been brought about through hierarchal systems that evolved
from the varying social patterns. Brooches, pottery, metal workings and equestrian-related
goods are just a few of the material artefacts that could be found across South Wales, and
although groups living in the area may have perceived a separate existence, these pieces
proved to unify the population under a single culture. It may have been through the
aristocratic leaders perpetuating the material culture that the area became a collective entity.
Yet, even under the same culture, the settlement landscape persisted, most likely
through a desire to maintain the social status quo. From this is most likely where a federation
of individual groups (or possibly ‘clans’) emerged; all in agreement that any threat to another

284
See Chapter 2 above, pages 13-20.
83

‘clan’ ultimately threatened themselves and deserved a unified retribution. In this can best be
seen the beginnings of the Silurian warrior force that repelled Roman occupation for almost
three decades.
It has been shown that even through all the varying origins of Silurian social, political
and economic systems, an aristocracy emerged becoming wealthy through their prestige and
associated power, and trading for exotic goods to flaunt such a status. Furthermore, the
breadth of South Wales did in fact exhibit contrasting lifestyles, which is most evident in the
defended enclosure patterns (or lack thereof) found by GGAT and CPAT. Gwent most likely
became economically complex due to the close proximity of the neighbouring tribes, which
numismatic evidence suggests were more highly developed (at least economically), and thus
naturally influenced the eastern Silures towards this same tribal standing. Trade networks
through the Dobunni probably gave the aristocratic families more power which only proved to
multiply over time. The high ratio of small defended enclosures to large suggests a more
socially dominant position of these latter sites; also hinted at by the large number of elite
artefacts. Glamorgan, conversely, portrays a contrasting communal society through the
prominence of smaller sites among fewer large enclosures. The relative isolation from
eastern British influence (not lacking but less dominant) correlated with distance from the
tribes, implies that the local aristocracy probably did not then become as socio-economically
divergent from the rest of the populace, keeping closer social (maybe even kinship)
relationships.
From this, it is also the view herein that most likely a hierarchical system of
aristocrats (or even a communal oligarchic group) emerged from the collective Silurian
leaders; which would best explain the decentralised system recognized in the South Wales
archaeological record. Whether or not social stratification within the larger community of
elites existed is unclear; although one would presume the eastern Silures would have at least
tried to assert such an order.
In part two of this dissertation, the Roman Invasion and subsequent Romano-British
period will be shown to have drastically altered the economy, society and politics of Iron Age
South Wales. The notoriety of the Silures is due to their resistence to the Roman invasion,
most notably through the decentralised political system being more difficult to conquer. Yet,
once defeated, the Roman influence upon the area was, for all intents and purposes,
immediate, and the first century of Roman occupation looked quite different from the
previous Iron Age period as the Silures were forced to change their social and cultural ways
to suit the new Roman lifestyle.
84

Part 2: Roman South Wales

Chapter 8: From AD 43 to Southern Wales


In AD 43, Emperor Claudius launched an invasion of Britain. Initial positive results
would end less than 400 years later with the island natives being left to fend for themselves.
The consequences arising from the Roman conquest and subsequent occupation of Britain,
and most notably South Wales, would compel the pacified natives to adapt to a new way of
life involving changes to the local politics, economy and society.
The Romans knew the benefits of fighting independent tribes for ‘rarely will two or
three states confer to repulse a common danger; accordingly they fight individually and are
collectively conquered.’285 The truth of this was never more so than in Britain. Military
tactics against the tribes of Britain differed with each. Some tribes became Pro-Roman and
thus were given independence, albeit on a restricted level. One such example of this is found
in Tiberius Claudius Togidubnus: client king of Rome and a full Roman citizen, adopting his
name from the Emperor, who kept much of his power through willingness to help the
Romans.286 Tribes less willing to give in easily were persuaded to concede to Roman rule
either through force or capitulations.287 Through this method, tribes were probably given less
independent rule and forced to give tribute, more than likely hostages, as was a common
practice of Caesar’s during his campaigns in Britain. 288 Lastly, however, the tribes that
decided to wage war with Rome and were beaten in the end were initially treated worse than
their neighbours, but only to a point, as Rome still needed some cooperation among the tribal
chiefs after pacification.289 The Silures were one of the tribes that opposed Roman occupation
for almost thirty years. In the case of the Silures, Rome’s strategy of conquest had to be
adapted to the decentralised political structure of the tribe, as discussed in the first part of this
study; this strategy implemented by Rome can best be seen from the military sites in South
Wales.
After the initial conquest of AD 43, fighting continued in southern Britain. There it
would be concentrated until around AD 47 when the second governor of Britain, Publius

285
Tacitus. Agricola. 12.
286
First recorded in Tacitus Agricola 14 and an inscription at Chichester. See Collingwood and Wright
(1965), 25-6: no. 91. In the inscription found at Chichester, the beginning of Togidubnus’ name is no
longer legible. Therefore the first reading of the inscription coincided with the translation of Agricola
14 which mentions a native king by the name of Cogidumnus. Modern interpretation have changed the
name to Cogidubnus, then Togidubnus. See Cunliffe, B. (1998). Fishbourne Roman Palace. Stroud:
Tempus; Mattingly (2006), 99.
287
Dio. 60.20, 21.
288
Caesar. Gallic Wars. 4.21; 4.27; 4.36; 5.20; 5.22; Tacitus. Annals. 11.19.
289
Millet (1990), 44.
85

Ostorius Scapula, arrived.290 After first putting down the revolt in the east, Scapula turned his
attention towards the Deceangli in north-east Wales. 291 This campaign was justified by the
defensive actions of protecting and securing the Roman socii (allies) in the West Midlands;
security which could not be complete without the pacification of Wales. 292 Legio XX (Valeria
Victrix after AD 61), which was initially stationed at Colchester, left the garrison to be
occupied by a force of retired soldiers, and moved to Kingsholm near Gloucester in AD 49 to
help pacify the area including the Silures. 293 They may have been stationed partly in
Kingsholm and partly in Clifford.294 In the following years, advances into Silurian territory
would result in many confrontations, most of which involved guerrilla warfare tactics by the
Silures. Caratacus, after fleeing in defeat from the battles of AD 43, fled west and inspired the
support of the Silures, along with the Ordovices, through memories of their ancestors’
victories over Caesar.295 In the ninth year of the Roman invasion, AD 51, Caratacus was
firmly defeated, but once more managed to escape, and he fled to northern Wales then into
Brigantian territory. He subsequently appealed to Queen Cartimandua of the Brigantes, but
because of her loyalty to Rome, he was captured and turned over to Claudius. 296 The leader of
the British resistance was finally subdued and taken to Rome to be paraded through the streets
as a monument to the determination and power of the Roman military.297
The location of the final defeat of Caratacus may have taken place near Hay-on-Wye
making the placement of part of Legio XX in Clifford important.298 Yet, a more northern site
is also suggested near the Llanymynech hillfort where the large area encompassed within the
hillfort (around 54ha (133 acres)) and an interior copper mine suggests a hilltop village,
possibly a powerful cultural and social centre. 299 Nearby camps at Abertanat and Clawdd
Coch could also be connected to the battle. 300 Caratacus may have fed off of a proud cultural
connection the Silures possessed which brought about the strong desire to defend South
Wales from the Roman attack. The defeat of Caratacus did not end the hostilities from the
Silures. In fact, in the years following South Wales would continue to be the site of constant
conflict.

290
Tacitus. Annals. 12.31.
291
Tacitus. Annals. 12.32. Tacitus states the tribe name as Decangi, but most likely they called
themselves the Deceangli.
292
Arnold and Davies (2000), 4.
293
Tacitus. Annals. 12.32.
294
Jarrett, M.G. (1994). Early Roman Campaigns in Wales. The Seventh Annual Caerleon Lecture.
Cardiff: National Museum of Wales, 16.
295
Tacitus. Annals. 12.33-35.
296
Tacitus. Annals. 12.35-37; Histories. 3.45.
297
Tacitus. Annals. 12.36; Webster, G. (2003). Rome Against Caratacus: The Roman Campaigns in
Britain AD 48-58. London: Routledge, 36.
298
Jarrett (1994), 17; Howell, (2009), 66.
299
Davies and Lynch (2000), 155.
300
Arnold and Davies (2000), 9; de la Bédoyère, G. (2003a). Defying Rome: The Rebels of Roman
Britain. Stroud: Tempus, 39.
86

After pacifiying the Deceangli in north eastern Wales, Scapula looked to South Wales
and the eventual final conflict which resulted in Caratacus’s capture at the hands of Queen
Cartimandua of the Brigantes.301 Rome presumably thought this would end any trouble from
the Silures, but soon they would be proven wrong as the Britons went on the offensive. After
a few successful raids on Roman detachments led to many Roman casualties including senior
officers, a change of strategy was implemented to a more aggressive approach, likely borne
out of frustration at the Silures’ unwillingness to concede. 302 Scapula opened this new tactic
by intimidating the Silures insisting that they ‘must be utterly exterminated,’ 303 and in all
probability accompanying the threat with harsher ground attacks. 304 In AD 51-2, Silurian
guerrilla warfare would surprise the Roman forces, giving the natives the upper hand on many
occasions. Sometimes this would be through chance encounters while other time they were
planned attacks on legionary cohorts building forts in Silurian territory. 305 Scapula would
soon succumb to the pressure of warfare and died suddenly in AD 52.
Aulus Didius Gallus was despatched immediately after Scapula’s death, but not
before the Silures took advantage of the absence of a Roman governor by plundering far and
wide including defeating a Roman force commanded by Manlius Valens. 306 Gallus is not
looked upon favourably by Tacitus. In Annals, he is said to have ‘merely held his own’ 307
while in Agricola he is given more credit as having ‘pushed forward a few forts into remote
districts in order to gain credit for enlarging the province.’ 308 Yet, this may not be the case.
The first establishments of legionary fortresses in and near the Welsh border are found in Usk
and Wroxeter, respectively. Archaeological excavations of the sites place their construction
around AD 55-60. With Gallus in power through AD 56, the expansion of the Roman frontier
into the areas of the fortresses and their subsequent construction, along with forts at
Abergavenny and Monmouth, can be accredited to the governor. 309 His efforts would be the
starting point of later governors for their campaigns against the Silures.
Usk and Wroxeter were built with an obvious strategy in mind. A network of roads
connecting the early pre-Flavian forts was designed to consolidate and maintain the gains of
the previous campaigns. Legio XIV Gemina first held Wroxeter c. AD 55, and it is presumed
that Legio XX built and then garrisoned Usk around the same time, if not soon thereafter.
Small forts are found on or near the road between the fortresses. The northern end of the
301
Tacitus. Annals. 12.36.
302
Mattingly (2006), 103.
303
Tacitus. Annals. 12.39.
304
Mattingly (2006), 104.
305
Tacitus. Annals. 12.38-39; Manning, W.H. (2004), 180. One site where such a planned attack on the
Romans is believed to have happened is in the area around Hay-on-Wye. This is the same area
believed by Ray Howell to be the location of Caratacus’ final defeat.
306
Tacitus. Annals. 12.40.
307
Tacitus. Annals. 14.29.
308
Tacitus. Agricola. 14.
309
Manning (2004), 180.
87

Roman road is unconfirmed but may include Whitchurch, while the southern road may end at
Chepstow, which seems like a logical control base for the Severn. However, there is no
strong evidence to support the idea. Just south of Usk at Coed-y-Caerau (Langstone) (Pen
Toppen Ash Camp310) lies a small enclosure with a view of the Severn estuary, a strategic
location for a fort to support the legionary fortress. The layout of the enclosure shows the site
more likely is of Iron Age origin311; however this does not negate the possibility of Roman
occupation, as many instances are found in Britain of Romans using existing hillforts,
including the ferry crossing port at Sudbrook on the south-east coast of Wales. 312 Forts such
as Abergavenny,313 Cardiff,314 Monmouth315 and Pen y Gaer316 were also constructed around
the same time as Usk and were connected to the fortress through a network of roads spanning
south-east Wales. Abergavenny, Cardiff, Monmouth and Pen y Gaer are not near any Silurian
political centres but instead are strategically placed to oversee territory under Roman control.
This is similar to the system Vespasian maintained against the Durotriges and Dumnonii
suggesting militarily that the Silures had a decentralised society as well. The idea of the
Silures as a clan-based confederation has already been established through Iron Age evidence,
and, with such a similar campaign strategy as that implemented against the tribes of South-
east Britain, the idea is reinforced. Yet this will be strengthened when the Flavian conquest
extends the defensive network already in place continuing with the similar strategy. This
network of forts and roads would work well to assert control until matters in other parts of
Britain would force the Roman army to consolidate, losing power over the Silures
In AD 57, Quintus Veranius, with a reputation in mountain warfare, was appointed
governor of Britain. He gained this skill from his five years as governor in Lycia, and with
this appointment as governor of Britain, Nero must have had an assult on Wales in mind. 317
Tacitus records Veranius as having ‘only conducted minor raids against the Silures when
death terminated his operations’ within the year. Yet, Veranius probably still managed to
further the Roman occupation in Wales as forts were probably under construction around the
time of his passing.318

310
Coed-y-Caerau (Langstone) and Pen Toppen Ash Camp are often mentioned as a single site.
Wiggins (2006), however, refers to them individually, and as such they are listed separately in
Appendix 1.
311
Manning (2004), 201: footnote 4.
312
Arnold and Davies (2000), 10.
313
Ponsford, M.W. (2006). ‘Archaeological Excavations at Castle Street Car Park (1999) and Ewers’
Garden (1968-72), Abergavenny, with a Reconsideration of the Early History of the Roman Fort’,
Archaeology in Wales. 46, 63-4.
314
Webster, P.V. (1984). ‘The Roman Period’, in Savory (1984), 280-2.
315
Clarke, S., Jackson, R. and Jackson, P. (1992). ‘Archaeological evidence for Monmouth’s Roman
and early Medieval defences’, Archaeology in Wales. 32, 1-2.
316
Symons, S. (2009). Fortresses and Treasures of Roman Wales. Derby: Breedon Books, 53.
317
Mattingly (2006), 178. Burnham and Davies (2010), 37.
318
Webster (2003), 41.
88

Pre-Flavian control of Wales ended with a defensive system of forts stemming from
legionary fortresses at Wroxeter and Usk through the road network to forward bases reaching
possibly as far west as Cardiff. 319 Romans had experience in mountainous terrain due to the
natural landscape of central Italy. The series of roads and colonies in the area was an
effective way of enacting control. Instead of colonies, forts were commonly found in
southern Wales in 15-20 km intervals, which became the common military strategy for
Flavian governors, Frontinus and Agricola, in the AD 70s.320 This system was also partly
modelled after the first emperor of the Flavian dynasty, Vespasian, who used forts to
consolidate and hold military gains. The fact that Claudius and Nero placed governors with
mountain warfare experience such as Veranius and Paullinus in power further stresses the
focus of the Roman military on Wales. Warfare in Britain, however, remained distinct with
the implementation of auxilia and the close placement of forts inside mountainous zones not
common outside of Britain.321 Yet, this pre-Flavian system appeared incomplete, with the
status quo being maintained from the time of Paullinus until the eventual conqueror of
southern Wales, Sextus Julius Frontinus, arrived in AD 74.322
With the death of Nero in AD 68 and the chaotic Year of the Four Emperors that
followed, Britain was under the governorship of Marcus Vettius Bolanus, who is credited for
achieving a small amount of success, yet most likely did little but quell unrest amongst the
idle armies.323 In AD 71, Quintus Petillius Cerialis was appointed governor, and there is
nothing that suggests he entered Wales during this tenure save for timber being cut down near
Caerleon in the winter of AD 73/4.324 This may have been the beginning of construction on
Caerleon suggesting that preparations for what would be the final conquest of Wales had
begun during this time. The following year, AD 74, Sextus Julius Frontinus known for his
campaign experience in the Rhineland began his time as governor by completing the
legionary fortress at Caerleon in southern Wales and along with another fortress at Chester in
the north to be garrisoned by Legio II Augusta and Legio II Adiutrix respectively. Legio XX
remained at Wroxeter in central Wales creating a strong line of legionary garrisons along the
eastern border of Wales.325 At this point it must have become obvious to the natives what
Rome’s intentions were. In AD 76 the longstanding conflict with the Silures came to an end
through the swift conquest of the area that included the Demetae west of the Silures. 326
Frontinus continued his campaigns north eventually conquering the Deceangli and Ordovices
319
Arnold and Davies (2000), 11.
320
Jarrett (1994), 29.
321
Syme, R. (1988). ‘The Subjugation of Mountain Zones’. in Birley, A.R. (ed.) (1988). Roman Papers
V. Oxford: Clarendon Press, 653; 656.
322
Jarrett (1994), 20; Arnold and Davies (2000), 11-2.
323
Tacitus. Agricola. 16.
324
Arnold and Davies (2000), 13; Burnham and Davies (2010), 43.
325
Arnold and Davies (2000), 13.
326
Tacitus. Agricola. 17.
89

as well, bringing all of Wales under Roman influence. The only exception to the complete
pacification of Wales by Frontinus came through the later rebellion of the Ordovices,
however this was quelled in a short campaign season by Gnaeus Julius Agricola immediately
after he gained the governorship.327 Thus in AD 77, thirty years after Scapula first brought
Roman arms against the Deceangli, Wales was under total Roman military control
With the three legionary fortresses in place in the mid-70s AD, the conquest of Wales
under Frontinus was enacted quickly. Tacitus fails to mention much of Frontinus’ success,
possibly due to his efforts in highlighting Agricola’s achievements. The conquest, however,
can be retraced partly through the archaeological evidence that remains. 328 Temporary camps
are found throughout Wales, but more so in the central and southern areas. Many temporary
camps are found near Flavian forts including one at Coelbren on the southern edge of the
Brecon Beacons and the largest camp in Wales at Blaen-cwm-bach overlooking the lower
Nedd, showing a desire to quickly consolidate and maintain conquered land. Some Flavian
forts may have been built during campaigns including those in southern Wales at Neath and
Llandovery,329 the former near the camp at Blaen-cwm-bach, while other forts were just
reused pre-Flavian garrisons such as Abergavenny,330 Cardiff,331 and Pen y Gaer.332 This
shows the majority of Flavian forts were built using the already established strategy of
consolidating territory gained through campaigns, the most efficient strategy against
decentralised societies. Forts at Caernarfon and Pen Llystyn and a fortlet at Brithdir, on the
other hand, were placed over settlements of late prehistoric origin; although whether or not
these settlements were in use before the placement of the military establishments is
unknown.333 One characteristic of Flavian garrisons often found is the placement of bases on
valley bottoms which helped control the native populations while also easing the burden of
supplying the army. This is also key to the sites of Chester and Caerleon, where the latter
replaced Usk due to the pre-Flavian fortress being subject to flooding and hard to supply by
sea.334
Another important aspect to the Roman military strategy was the internal organization
of the tribe with whom they were fighting. A centralised tribe which controlled a large area
from a single capital centre made the pacification of the entire area complete through the
defeat of only that said capital. Such an example is found in the successful taking of
Camulodunum (Colchester) which gave the Romans control over the entire
327
Mattingly (2006), 116.
328
Arnold and Davies (2000), 13-4.
329
Davies, J.L. (2004). ‘Soldier and Civilian in Wales’. in Todd, M. (ed.) (2004). A Companion to
Roman Britain. Oxford: Blackwell, 94
330
Ponsford (2006), 49.
331
Webster (1984), 280-2.
332
Symons (2009), 53-4.
333
Arnold and Davies (2000), 15.
334
Arnold and Davies (2000), 15.
90

Catuvellaunian/Trinovantian kingdom through a single Roman legionary fortress. 335


However, a complete contrast to this is found in the areas where the social structure was more
similar to an association of smaller tribal units. 336 This meant that a conquest over the entire
tribal region would not be completed through a victory at one site, but rather subsequent
battles would need to be fought individually against each branch of the tribe. During the
initial conquest in AD 43, Legio II Augusta, at the time under command of future Emperor
Vespasian, was given the task of subduing the Durotriges, and only a conquest of twenty
oppida (hillforts) would be sufficient in pacifying the area; which he completed. 337 This
proved to be an arduous task, as each tribal centre had to be pacified and then a small garrison
left to keep control. The archaeological evidence shows that on many occasions the Iron Age
hillfort seized by the Romans would become their garrison, and would thus need more
soldiers to man each one. Therefore, a campaign against a decentralised tribal society like the
Durotriges proved much more difficult than a centralized tribe similar to the Catuvellauni.
After subduing the Durotriges, Vespasian moved into the southwest of Britain and proceeded
to campaign against the Dumnonii. This tribe proved difficult to pacify as well, and evidence
points to a similarly decentralised tribal structure as their neighbours to the east. After
pacifying the Dumnonii, a legionary fortress was established at Exeter by Legio II Augusta in
c. AD 55. While known Roman forts in the area are for the most part not affiliated with tribal
centres, this is important as this is the first known instance in Britain of the Romans setting up
forts to consolidate territorial gains: a similar strategy which would be used during the
pacification of South Wales. However, the placement of the fort at Exeter may have been
only due to the absence of evidence for tribal political centres, or to the forts being used to
oversee mining in the area instead of just native land. 338 Strategic planning by Vespasian
against the Durotriges and Dumnonii would prove successful given the political organization
of the tribes. The same military strategy would be implemented by the Romans during the
conquest of South Wales. This is evident through the archaeological record left by the
Roman military while building a network of military sites as they spread westward.
Finally, there were numerous direct and indirect impacts which the Roman military
had upon the natives both during and after conquest. 339 During the invasion, much of that
found upon the Silurian landscape by the Roman was most likely destroyed (i.e. villages,
agriculture and resources), burnt by the soldiers in efforts to terrorise, intimidate and
eliminate the native resistance.340 Moreover, massacres, rape, population displacement and

335
Millet (1990), 48.
336
Millet (1990), 49-50.
337
Suetonius. Div. Vesp. 4.
338
Millet (1990), 50-1.
339
Millet (1990), 57-9; Mattingly (2006), 91-4.
340
Tacitus. Annals. 12.39.
91

enslavement were just a few direct consequences the invasion of the Roman army had upon
the native populations, most notably the Silures.341
Indirectly, the administration of the conquered population, furthermore, led to many
Romans accumulating local property, taking positions within the tribal political council and
presumably marrying native women. Notwithstanding, the Romans also indirectly disrupted
the native social order through the reduction of native warriors in acts of war or later
recruitment, as well as the influx of luxury goods such as wine or precious metals that may
have been used by the locals as elite high-status goods. This flood of elite trade goods then
became available to more of the native population thereby eliminating its power as a high-
status item.
The actions by the Roman military in the first century AD altered Silurian South
Wales and its native population forever. There is no doubt that the change brought about by
the Roman occupation had an effect upon the native population of South Wales. This change
is readily seen in the archaeological record, and as will be shown many of those living in the
Silurian territory embraced the new dominating culture and society while others chose to
maintain their previous existence within the new Roman lifestyle. Most notably, this change
will be evident in the first centuries of the Roman occupation, which is the focus of this
second part of the dissertation.

341
Tacitus. Annals. 14.30.
92

Chapter 9: The Roman Military Network in South Wales

Map 6: Major Roman sites in South Wales.


©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved.

As seen in the conquest of Wales, initial attempts to pacify the Silures led the
Romans only to the Usk river valley, where they set up the first legionary fortress. The route
the Romans took to the Usk from the fortress at Kingsholm is believed to have been north
through the Forest of Dean, crossing the Wye near Ross-on-Wye, then through Monmouth to
the Olwy valley and to the Usk river valley. 342 This route is attested by the pre-Flavian fort at
Monmouth, and possible fort at Weston-under-Penyard (Ariconium) east of Ross-on-Wye, the
latter of which possibly built during the initial campaign to the area. At that time, smaller
forts were also established at Clyro on the northern Wye, Abergavenny and Pen-y-Gaer on the
northern Usk and Cardiff on the Taff. Abergavenny is located at the edge of the modern
Brecon Beacons, while Pen-y-Gaer is further north along the river, halfway between the
former site and modern Brecon. After the final conquest over the Silures in the mid-70s AD,
many more forts were placed at key sites throughout South Wales to cover the lowland
coastal and upland valley topographies and limit future attacks. A new legionary fortress at
Caerleon (Isca Silurum), down the Usk river, south of the previous site, was complemented
by new military installations at Brecon, Caerphilly, Coelbren, Gelligaer, Loughor, Neath and

342
Manning (2004), 182.
93

Pen-y-Darren, along with another site known in the Antonine Itinerary as Bomium.343 In
essence, the campaign over the Silures led to the Roman occupation, while the network of
forts initiated and maintained control throughout (Map 6).344

Pre-Flavian Forts
Although Publius Ostorius Scapula’s defeat of Caratacus in Wales ushered in new
Roman military initiatives against the Silures, it is likely that Aulus Didius Gallus was, as
governor, responsible for the first concerted effort to pacify South Wales. It was during this
time that the initial fort construction in eastern South Wales, possibly begun by Scapula, was
completed (Map 6). The military strategy against the Silures mirrored that of Vespasian’s
exploits against the Durotriges and Dumnonii as evidenced through the initial placement of
forts in Monmouth and Clyro along the banks of the Wye river, and at Pen-y-Gaer,
Abergavenny and the Legionary Fortress at Usk, all along the Usk river.
Usk (Burrium) was constructed around AD 55, and was symbolic of the initial
movement into Silurian territory and the extent to which the first campaigns by Legio XX
reached.345 Abergavenny is believed to have been built c. AD 57, only two years after the
fortress at Usk, to protect the northern Usk river at the foot of the modern Brecon Beacons. 346
Cardiff’s fort on the southern coast, about 4 km from the Bristol Channel, allowed not only a
Roman garrison on the channel, but a safe moorage for Roman ships on the way to the
fortress at Usk.347 Facilities on the northern coast of Devon have also been interpreted as
lookout posts over the land further west of Cardiff, and if so some communication link over
the Bristol Channel would be expected. Not much is known about the pre-Flavian forts at
Monmouth and Pen-y-Gaer beyond pottery and coins, yet construction dates contemporary
with Usk and Abergavenny could be determined.348 The location of the latter fort along the
northern reaches of the Usk Valley within the Brecon Beacons confirms its place among the
military network as such a sitting would no doubt be to protect the northern access through
the valley.
A Roman period settlement and possible fort is believed to have been situated at a
ferry crossing over the Wye river at modern Chepstow. 349 Finds in the area, through
excavations and by chance, show material from the first century and the late third to fourth
century AD. This dearth of second century finds cannot at this time be explained, but is

343
See below, page 109.
344
For maps of the occupation histories of the Roman forts of South Wales see Appendix 2.
345
Manning (2004), 180-91.
346
Manning (2004), 188; Symons (2009), 16-8.
347
Simpson, G. (1963). ‘Caerleon and the Roman Forts in Wales in the Second Century A.D.’,
Archaeologia Cambrensis. 112, 69-72; RCAHMW (1976b), 90-4: no. 735; Webster (1984), 280-2.
348
RCAHMW (1986), 146-9: RF 3; Manning (2004), 188; 190; Symons (2009), 53-4.
349
Shoesmith, R. (1991). Excavations at Chepstow 1973-1974. Bangor: Cambrian Archaeological
Association, 156-60.
94

interpreted as only a lapse in occupation. A fort would be plausible here to guard the crossing
that later would be replaced by a bridge as part of a Roman road, and would be a continuation
of Roman fortifications along the Wye river as found at Monmouth and Clyro further north.
At Sudbrook, on the coast of the Severn Estuary, early Roman period occupation is found
starting in the mid-first century AD.350 Iron Age occupation at Sudbrook has been addressed
above, and the probable cross-channel trading links with the site are obvious. The site has
even be conjectured as key to the Roman road Iter XIV running from Caerleon to Silchester. 351
However, the validity of this is in doubt with the evidence provided of short Roman
occupation at Sudbrook, and that suggesting a Roman crossing at Chepstow.352
With the movement of legions around Britain, Legio XX left Usk c. AD 66/7 for
Wroxeter and Legio II Augusta moved to Gloucester at about the same time, it appears that
the Roman strategy into South Wales was placed on hold. Archaeological evidence indicates
the systematic dismantling of the Usk fortress and reuse of its timbers and stone in the
building of Caerleon in the mid-70s. This may suggest that the Legionary Fortress at Usk was
used only sparingly, or possibly never completely garrisoned. A smaller Flavian fort would
later be built within the north-western corner of the fortress foundations, possibly as a
metalworking depot or to control the northern boundary of Caerleon’s territorium, but
ultimately Usk had lost its dominance within South Wales.
At Abergavenny, archaeological evidence suggests a period of abandonment in AD
60s, then reconstruction of the fort in the later first, mid-second and mid-third centuries by
Legio II Augusta with periods of neglect in between and soon after. A date of abandonment
at Monmouth is not known for certain, but nothing suggests use after the early Flavian
period.353 The construction of the auxiliary fort at Brecon c. AD 75 would logically
have made Pen-y-Gaer obsolete, yet evidence does not suggest this. Samian ware dating from
the mid-second century suggests a continued garrison, although some signs suggest the fort
was abandoned before then during Hadrian’s construction of the northern defences. Early
abandonment at Clyro, c. AD 60, leads to the fort quickly becoming unnecessary, possibly by
the fort at Pen-y-Gaer providing enough military interaction for the area. 354
In opposition to the fates of the other Pre-Flavian forts, the history of the Roman fort
at Cardiff was long, as coastal garrisons would continue to be necessary in the later Roman
period.355 However, the occupation would undergo alterations during that time.
Archaeological evidence suggests the fort was rebuilt on three occasions to accommodate

350
Nash-Williams (1939), 54.
351
Rivet and Smith (1979), 177.
352
Shoesmith (1991), 156-7.
353
Clarke, S. (1993). ‘Monmouth, The Gloucestershire House, St. Mary’s St. (SO 505 126)’,
Archaeology in Wales. 33, 61; Brewer (2004), 215.
354
Symons (2009), 64.
355
Jarrett (1969), 70.
95

changes in the garrison over time, with numismatic evidence indicating occupation remaining
into the last quarter of the fourth century. Overall, the initial Roman military strategy of a
network of forts in eastern South Wales did not persist long after it began. With the exception
of the site at Cardiff, and possibly Pen-y-Gaer, the military establishments of the AD mid-50s
quickly fell into disuse in favour of military actions elsewhere in Britain and eventually to a
new strategy inacted by Julius Frontinus in AD 74.

The Flavian Forts


In AD 74, Julius Frontinus initiated his campaigns into Wales, and at that time
decided that a new legionary fortress was needed. These campaigns mirrored closely the
same strategy stressed above with a network of forts spread across the region (Map 5).
Reasoning behind the move to Caerleon may be found in the geographical location of Usk, as
the area may have been too liable to flood from the Usk river and Olwy brook, it was
constricted by the surrounding waterways and hills, and may have been placed too far up the
Usk river to be reached by larger ships. 356 Furthermore, the position of the fortress in the east
of South Wales may have been due to strategic reasons, as fortresses were consistently placed
well behind the front line, and a legion may have been needed within a short distance of
south-west England for security purposes. The territorium of Caerleon, discussed below,
most likely included that which was taken by Usk two decades earlier. Construction included
the reuse of timber and stone from the Usk fortress and is generally believed to have been
completed c. AD 75, yet fortresses may have taken longer especially if detachments were
away for a period of time, thus a later date around 76 or 77 may also be possible. Stone walls,
towers and gates began replacing the early timber fortifications sometime after AD 86, further
stressing the importance of the fortress within the strategy of the conquest of South Wales. 357
Caerphilly is about 11 km north of the fort at Cardiff, and was a strategic location
between the Roman road leading from Cardiff up the Rhymney Valley to the forts at
Gelligaer and Pen-y-Darren to the north, and the Roman road west from Caerleon. 358 North of
Caerphilly along the road to Pen-y-Darren and into the modern Brecon Beacons sits the fort at
Gelligaer.359 The first fort was built of earth and timber during the 70s; a second smaller
stone fort was built to the south-east in the first decade of the second century. Pen-y-
Darren fort, further north on the Roman road from Cardiff, is, as expected, also of origin after
the conquest of Frontinus.360 Two periods of construction are noted, the first of timber and

356
Boon, G.C. (1972). Isca: The Roman Legionary Fortress at Caerleon, Mon. Cardiff: National
Museum of Wales; Manning (2004), 189-201.
357
Zienkiewicz, J.D. (1993). ‘Excavations in the Scamnum Tribunorum at Caerleon: the Legionary
Museum site 1983-5’, Britannia. 24, 85.
358
RCAHMW (1976b), 94-5: no. 736; Symons (2009), 43.
359
RCAHMW (1976b), 95-8: no. 737; Brewer, R.J. (1996). Gelligaer Roman Fort. Caerphilly:
Rhymney Valley District Council; Symons (2009), 44-50.
360
Simpson (1963), 45-9; RCAHMW (1976b), 84-6: no. 732.
96

earth presumably c. AD 75, and the second in stone most likely in the early second century.
The Roman fort at Neath (Nidum) has been excavated on many occasions since 1950. 361 An
early Flavian fort of c. AD 75-80, soon after the pacification of the Silures, was then
constructed, although possibly not finished. On the Roman road north up the Neath valley
sits a fort at Coelbren on the southern edge of the Brecon Beacons. 362 It is believed that
military installations in areas such as Pen-y-Darren, Coelbren and Gelligaer were more for the
continual pacification of the locals within the upland areas that were far removed from the
southern, more heavily occupied coastal lowlands. This mostly becomes necessary during the
annual movements of cattle and the transhumance lifestyle of those associated. 363 Therefore,
the establishment of forts at these sites further substantiate not only the pastoral subsistence
lifestyle of the Silures in the Iron Age, but the continual practice of pastoralism by the
Romano-British natives.
Eighteen kilometres west of Neath, the furthest Roman fort in Silurian territory was
the site situated at Loughor (Leucarum) on the banks of the river Loughor. 364 This location
allowed for sea-borne access and resupply similar to the installation at Cardiff. Initial
occupation is assumed to be similar to the Flavian forts of South Wales, but given the western
location, a later date may be more probable.
North of the Brecon Beacons, the fort at Brecon (Cicutium) sat next to the confluence
of the Ysgir river into the Usk river. 365 With the possible exception of the fort at Clyro,
Brecon was the most northern Roman fort used to secure the Silurian territory. Built of clay
and timber around AD 75, the auxiliary fort housed the Ala Hispanorum Vettonum, a Spanish
cavalry unit of 500 horsemen.
In addition, Caergwanaf, a site in modern Rhondda Cynon Taff 14.5 km north-west
of Cardiff, was recently discovered as having characteristics of a Roman fort. 366 Timber and
earth foundations, along with archaeological evidence, suggest that the site was built in the

361
Nash-Williams, V.E. (1950a). ‘The Roman Stations at Neath (Glamorgan) and Caer Gai (Mer.)’,
Bulletin of the Board of Celtic Studies. 13, 239-45; Nash-Williams, V.E. (1952). ‘The Roman Station at
Neath, Further Discoveries’, Bulletin of the Board of Celtic Studies. 14, 76-9; RCAHMW (1976b), 88-
90: no. 734; Marvell, A.G. and Heywood, B. (1992). ‘Excavations at Neath’, Bulletin of the Board of
Celtic Studies. 171-298. In the present dissertation, the 1958, 1984-85 and 1989 excavations are
presented. In 2010, further excavations took place by the GGAT, before the scheduled construction of
a school building, at which the author was honoured to have been allowed to volunteer.
362
RCAHMW (1976b), 83-4: no. 731; Symons (2009), 69-72.
363
Arnold and Davies (2000), 76-7.
364
RCAHMW (1976b), 86-8: no. 733; Jones, C. (2000). The Roman Auxiliary Fort of Leucarum.
Swansea: University of Wales Swansea; Symons (2009), 78-83.
365
Simpson (1963), 16-37; RCAHMW (1986), 135-46: RF 2; Symons (2009), 55-63.
366
Young, T. (2003). ‘Caergwanaf, Pont-y-Clun, Rhondda Cynon Taff (ST 04 80)’, Morgannwg. 47,
67-9; Burnham, B.C. (2004). ‘Roman Britain in 2003: 1. Wales’, Britannia. 35, 263-4; Young, T. and
Guest, P. (2004). ‘Caergwanaf, Pont-y-Clun, Rhondda Cynon Taff (ST 04 80)’, Morgannwg. 48, 73-4;
Burnham, B.C. (2005). ‘Roman Britain in 2004: 1. Wales’, Britannia. 36, 390-2. Excavation of the site
was partly conducted by Cardiff University: see
http://www.cardiff.ac.uk/share/research/projectreports/miskin/index.html.
97

last quarter of the first century AD, which coincides with the other Flavian installations.
Caergwanaf is the likely location of Bomium, a site listed in the Antonine Itinerary between
Caerleon and Neath on the Roman road Iter XII.367 Until recently Cowbridge was identified
as Bomium due to its Roman material evidence and location in close proximity to where
Bomium is said to be.368 However, textual and linguistic problems are well documented for
texts such as the Antonine Itinerary and discrepancies are often found in the distances
between places listed.369 Furthermore, the lack of any structural evidence of a fort at
Cowbridge coupled with the close proximity of the location with Caergwanaf, leaves doubt in
the two being simultaneously garrisoned, let alone even existing at the former location.
Thus, the second military conquest of South Wales by Julius Frontinus expanded
further than Aulus Didius Gallus’ previous efforts. Yet, similar to that enacted by Gallus,
Frontinus placed forts in key locations to keep a military presence throughout South Wales, as
well as to guard strategic access to routes by land or sea which allow contact to resources and
other military installations. Furthermore, archaeological evidence indicating reinforcement in
stone of the fortress at Caerleon and forts at Neath, Loughor, Coelbren and Brecon attest that
this strategy was maintained and reinforced into the second century AD.

The territorium
At the end of the first century AD, the Roman military had a strong hold over South
Wales. The territory (territorium370) under direct control of the Legio II Augusta of Caerleon
most likely incorporated that which was attached to the previous fortress at Usk, as well as
extending south to the coast at the Gwent Levels, and west to the mining operations at
Draethen and Risca. Advantages from the assimilation of Usk’s territorium may have been
twofold: the reincorporation of existing Roman military features, and less land would need to
be sequestered from the Silures which may have been cause for provocation. This vast
amount of land may have added up to over 375 km 2 and might also have created a boundary
separating the Silures of east and west (see Map 6). 371 Even with the establishment of Venta
Silurum within fifty years of the conquest of the area, much of South Wales may have
remained at least partly under Roman military control. 372 As can be seen through the military
367
Rivet and Smith (1979), 173-5. The location is also referred to as Bovium. See: Rivet and Smith
(1979), 273.
368
Robinson, D. M. (1980). Cowbridge: The Archaeology and Topography of a Small Market Town in
the Vale of Glamorgan. Swansea: The Glamorgan-Gwent Archaeological Trust Limited Town Survey
No. 1, 23-4.
369
Rivet, A.L.F. (1970). ‘The British Section of the Antonine Itinerary’, Britannia. 1, 37-9; Rivet and
Smith (1979), 29-36.
370
Epigraphic evidence shows the land under a legion’s control was initially termed a prata, yet by the
second century AD became more commonly known as a territorium: see Mason, D.J.P. (1988). “Prata
Legionis’ in Britain’, Britannia. 19, 165.
371
Mason (1988), 184; Mattingly (2006), 414-5.
372
Rivet, A.L.F. (1969). The Roman Villa in Britain. London: Routledge and Kegan Paul, 161: fig. 9;
Mattingly (2006), 262: Fig. 10.
98

network, the Romans placed installations to police the upland and valley areas which
stretched away from the coastal lowland towards the Black Mountains. The presence of
watchtowers (or fortlets) along the routes between the forts attests to this continual policing of
South Wales. Hirfynydd and Rheola Forest watchtowers are believed to have been examples
of this between Neath and Coelbren forts, while at least five more are alleged to have been
necessary to effectively patrol and control the same route. 373 With the decades of dissidence
from the Silures, and the Roman apprehension towards the Britons still lingering from the
Boudiccan revolt, this continual military presence is well founded. Yet, most of the modern
Vale of Glamorgan and Monmouthshire lowlands remained free of military intervention,
likely due to the establishment of the territorium.
Wetland reclamation at the Gwent Levels is well attested archaeologically, and the
Goldcliff stone indicates the role of Legio II Augusta in the undertaking.374 In 1878, the
Goldcliff Stone (Illus. 5) was found; dating to the late second or third century AD, it refers to
work by legionaries in the area, and has been argued as delimiting the territorium of Legio II
Augusta at Caerleon.375 Although it dates at least 100 years after the initial garrison of the
area, the stone must have been erected on the newly reclaimed land because it was of
significant value to the Romans, enough so to maintain possession. The reasoning behind the
projects most likely stems from the necessity for more agricultural and pastoral land to
maintain the increase in population of the area after the Roman conquest. An additional
benefit may have come in relieved tension between the Romans and Silures as the increase in
agro-pastoral land would have been a welcome addition for the native population. 376 There
should be little doubt that much of the reclaimed land was subsequently included in the
territorium, however some may have been returned to the local population for this reason,
albeit under military supervision. The greater number of documented sites on the Caldicot
Level as compared to the Wentlooge Level may suggest that the former area was given to the
Silures, while the latter remained military property.377
Mining operations in Draethen and Risca are almost assuredly militaristic in nature, at
least in the early stages.378 Roman structures are found in the area, including a settlement at
Lower Machen, a bath-house at Risca and early Roman coins and pottery in the mine at Coed
Cefn Pwll-Du, Draethen, suggesting occupation from the late first to late second century AD.
373
Chouls, W.H.M. (1993). ‘Rheda Forest, Hirfynydd (SN 8124 0401)’, Archaeology in Wales. 33, 59;
Evans, E. (2010a). ‘Hirfynydd (NGR: SN 8284 0663; NPRN 93165)’, in Burnham and Davies (2010),
296; Evans, E. (2010b). ‘Rheola Forest (NGR: ST 8127 0404; NPRN 301353)’, in Burnham and
Davies (2010), 298.
374
Allen, J.R.L. and Fulford, M.G. (1986). ‘The Wentlooge Level: A Romano-British Saltmarsh
Reclamation in Southeast Wales’, Britannia. 17, 91-117; Rippon (1996), 25-35.
375
Collingwood and Wright (1965), 132: no. 395.
376
Mason (1988), 183; Brewer (2004), 233.
377
See below, page 127.
378
Nash-Williams, V.E. (1939b). ‘A New Roman Mining Site at Lower Machen, Monmouthshire’,
Archaeologica Cambrensis. 94, 108-10; Mason (1988), 184; Arnold and Davies (2000), 98.
99

Supporting evidence of stamped lead ingots does not indicate a widespread distribution of the
mined objects; however this could be solely due to a lack of surviving material.

Communication links
One of the most recognizable Roman features is the construction of roads and
communication networks within conquered territories. 379 Roads built along valleys and rivers
and across the lowlands centre upon Caerleon and maintain a link through the uplands with
other forts. Main roads within South Wales have been documented in ancient sources such as
the third century AD Antonine Itinerary and the seventh century AD Cosmography of
Ravenna. There are three roads mentioned by the Itinerary as including Caerleon: one
moving west and ending at Moridunum Demetarum (Carmarthen): Iter XII through Bomium,
Cowbridge, Neath and Loughor; two moving east and ending at Calleva Atrebatum
(Silchester): Iter XIII through Colonia Nervia Glevensium (Gloucester) and Corinium
Dobunnorum (Cirencester), and Iter XIV through Caerwent and Bath (presumably Gloucester
as well)380 In addition to those mentioned in the Itinerary and Cosmography, more roads have
been documented and excavated in the area connecting the military installations to a single,
all-encompassing network.381 Possibly the most important of these is the road extending north
from the legionary fortress through the former fortress at Usk, north to Abergavenny and into
the modern Brecon Beacons passing through Pen-y-Gaer and Brecon (Cicutium). If this road
is followed west out of Brecon, it is believed to pass through the Roman forts at Alabum
(Llandovery) and Llandelio, also ending at Carmarthen, circling the northern boundary of
South Wales. The Roman road from Neath through Coelbren, known in modern times as
Sarn Helen, meets this northern road at Brecon, as, it is believed, does the route from Cardiff,
through Caerphilly, Gelligaer and Pen-y-Darren. Therefore, the southern route of Iter XII and
the northern road through Brecon create outer ring roads into which the internal paths through
the inner forts end, creating an efficient network by which the military can adequately police
the Silures.
It is also on these routes that Romano-British roadside settlements and urban centres
grow. Cowbridge, in the Vale of Glamorgan, is sited on Iter XII and provides the area with a
marketplace which is not available in between the Cardiff and Neath vici. Bulmore will be
shown as a probable legionary establishment for retired veterans which was established two
kilometres north-east of Caerleon on the road north through Usk. After the decrease in
Roman military garrisons of South Wales, vici and other rural settlements remained and
became self-sufficient, owing much to their position on the Roman roads.

379
Davies (2004), 98-100; Silvester, R.J. and Toller, H. (2010). ‘Roman Roads in Wales and the
Marches’, in Burnham and Davies (2010), 91-7.
380
Rivet (1970), 57-60; Rivet and Smith (1979), 173-8; Mattingly (2006), 258: table 8; 259.
381
Sherman, A. and Evans E. (2004). Roman Roads in Southeast Wales. GGAT Report No. 2004/073.
100

Communication networks as well are important in the control of South Wales; as was
stressed by the possible Bronze Age and Iron Age long-distance contact points on hilltops and
in defended enclosures.382 Possible watchtowers or signal stations between Neath and
Coelbren have been mentioned at Hirfynydd and Rheola Forest, and mansiones at Cold Knap,
near Barry in the Vale of Glamorgan and Cwrt Herbert outside Neath. 383 These are all part of
the Roman system of sites keeping communication and contact points nearby for use during
travel and in emergency.

The Roman Military and the Silurian Settlement Pattern


In previous chapters, the development of South Wales through the Iron Age has been
presented as indication of a decentralised society focused upon local wealthy aristocrats in
‘large’ defended enclosures who provided for those around them in return for payment and
prestige. This association of sub-tribal groupings then maintained a single cultural connection
through artefacts that can be characterised in the archaeological record as a shared Silurian
social awareness. Further to the decentralised settlement pattern, the Roman military network
can be seen as reflecting the Iron Age presence at the time throughout South Wales. The
necessity for the number and locations of the military installations after the initial invasion
and subsequent Flavian conquest bring about a renewed viewpoint stressing the settlement
pattern of the Silures and need to secure the area in such a way.
It has been shown that the mid-first century military activity in South Wales ended
with the legionary fortress built at Usk, on the river, and auxiliary forts constructed along the
northern reaches of the same river at Abergavenny and Pen-y-Gaer, as well as the coastal
location at Cardiff on the Taff river. The placement of Abergavenny at the foot of the Brecon
Beacons indicates the strategic objective of policing the Usk valley from hostile natives from
the north. Then, the only logical reasoning behind the installation of Pen-y-Gaer further north
must be from the view that the Silures in the Brecon area needed further control. A fort near
the confluence of rivers beneath Mynydd Epynt may not have been secure enough, thus
showing a strong presence of natives in the Brecon area supported by the prevalence of the
Iron Age settlements in the Brecon area, as indicated above.
Interpretations of the strategy behind the fort at Cardiff, besides the advantages of a
coastal and riverine location, may be lost to the development of the city of Cardiff over time.
However, the question remains why the fort was placed on the Taff river, and not the Usk,
guarding the entrance to the legionary fortress. Geographically, the Cardiff-Newport area is a
thin, lowland coastal zone with the uplands beginning south of Caerphilly. The placement of

382
Arnold and Davies (2000), 35-9; See above, page 81.
383
A Research Framework for the Archaeology of Wales: Southeast Wales – Roman, key sites: see
http://www.archaeoleg.org.uk/pdf/roman/KEY SITES SE WALES ROMAN.pdf., Last Date of Access:
4/17/11.
101

a military installation within the lowland ‘bottleneck’ controls the easy access from east to
west South Wales. The decision of Cardiff over Newport must have derived from closer
proximity to the larger number of native settlements that are spread throughout the Vale of
Glamorgan, as the area of modern Newport does not contain many known defended
enclosures.
After the Flavian pacification of the area, installations along the river valleys at
Caerphilly, Coelbren, Gelligaer, Neath and Pen-y-Darren are obvious strategic defences
against attacks from upland natives, stressed above as practising pastoral transhumance
subsistence. Loughor reflects another coastal, riverine location that then polices the western
boundary of the Silures, and must presumably allow easier access to the Gower peninsula by
sea than by land. Brecon as a strategic location within the centre of the Usk valley north of
the Brecon Beacons shows the Roman pacification of the area enough to need only a single
fort to maintain control, although Pen-y-Gaer did initially remain garrisoned. Leaving
Monmouth and Usk relatively unmanned at the end of the first century AD also shows the
extent to which the Silures had been controlled in the east, while the same may be said in the
west as Loughor and Neath presumably served as strong enough garrison to maintain the
Gower peninsula. Furthermore, the relative dearth of a military presence in the Vale of
Glamorgan must also indicate a total conquest of the southern lowland coastal region as
Caergwanaf is the only known fort in the area, and its short occupation period and placement
off the Roman road can only suggest the site quickly became unnecessary.

The Decline of the Military in Roman South Wales


Shortly after the consolidation of South Wales, military operations elsewhere in
Britain and the Empire took detachments of Legio II Augusta away from their garrisons.
Many it would seem never to return. 384 Agricola’s campaigns in the north would almost
instantly take troops from South Wales, while further encounters by Trajan in the beginning
of the second century, the outbreak of war in northern Britain and the subsequent Hadrianic
fortification building regime all continued to pull forces from the Welsh fortifications.
Resulting from the consistent interference of military actions elsewhere was the eventual
abandonment of most garrisons in South Wales. Caerleon remained the headquarters of
Legio II Augusta through the second century, but detachments are noted at different locales
throughout the Empire from the late first century and into the third century.
From the third century to late fourth century AD, the Roman army and Roman Britain
underwent drastic changes.385 Reforms by Diocletian and Constantine on the military led to
384
Davies (2004), 100-4; Burnham and Davies (2010), 48-62; Casey, P. J. (2010). ‘The end of the
Roman army in Wales and the Marches’, in Burnham and Davies (2010), 62-6.
385
Owens, E.J. (1997). ‘Zosimus, the Roman Empire, and the End of Roman Britain’, in Deroux, C.
(ed.) (1997). Studies in Latin Literature and Roman History VIII. Bruxelles: Latomus Revue d'Études
Latines, Collection Latomus 239, 478-504; Southern, P. (2004). ‘The Army in Late Roman Britain’, in
102

much smaller units separated by their newly formed roles within the empire, while the two
already divided provinces of Britannia became partitioned further into four. Further
complicating matters, the desire for the Imperial throne led many to use the Roman military
for personal gain. The Gallic Empire from AD 260 to 274 separated Britain from the rest of
the Roman Empire and led to more military changes in garrisons and administration.
Aurelian managed to reunify the Empire shortly before Carausius rebelled in 286 gaining
control of Britain and Northern Gaul. Carausius and his assassin Allectus remained in power
for ten years, during which a defensive building regime was implemented on the southern
shores involving the Legio II Augusta and Legio XX Valeria Victrix. Much of the fourth
century was spent fighting internal wars amongst would-be emperors or focused on the
Roman frontier in the north of England. A last effort for securing Britain would come in the
latter half of the century under the construction of towers along the east coast.
The last half of the fourth century AD was also characterised by political and
economic turmoil coupled with the reduction in military garrisons throughout Britain that
eventually led to the end of Roman control. A rebellion by Magnentius from AD 350 to 353,
during which a large number of troops from Britain withdrew to the continent, continued the
habitual usage of British troops for personal gain. The Roman Empire’s reactions to
consistent struggle over the power of emperor and the inflation in the cost of living plus the
deflation of the currency created an unstable environment. In the first decade of the fifth
century, the military in Britain consistently elected leaders wishing to become emperor, only
to kill that person for another. The military force within Britain lessened as the usurpers took
legions to the continent to wage war, leaving Britain without support against consistent raids
from the Irish in the west and Saxons in the east. In AD 409, the Diocese revolted against
Constantine III, the latest usurper, essentially ending Britain’s connection with the Roman
Empire.
In South Wales, the military garrisons were not immune to the events happening
around them.386 Much of Caerleon is believed to have been abandoned by detachments of
Legio II Augusta during the construction of Hadrian’s Wall and the Antonine Wall in the
second quarter of the second century, yet the continual use of features such as the fortress
baths show a large part of the legion remained. At the end of the second century AD,
refurbishment began on the fortress initiated by Septimius Severus, and continued scattered
throughout the area during the rest of the third century. However, a different understanding
of the third century garrison can be seen in some empty barracks and the closure of the
fortress baths c. AD 230-40, plus systematic demolition of important buildings within the
fortress and the disuse of the amphitheatre at the end of the third century lead to a view
Todd (2004), 393-408; Mattingly (2006), 518-20; 529-36; Allason-Jones, L. (2008). Daily Life in
Roman Britain. Oxford: Greenwood World, 130-2.
386
Davies (2004), 106-8.
103

towards the end or reduction of the garrison by this time. 387 More recent excavations have
shown fourth century occupation evidence and even construction, yet whether this activity is
civilian or militaristic is not known. 388 Theories behind the later activity at Caerleon exist
centred upon the political events of Carausius at the end of the third century, and the army
reforms of Diocletian and Constantine around the same time and in the fourth century,
respectively. Carausius usurping power over part of the western Roman Empire separated
Britain from Rome and a movement of Legio II Augusta to the southern coast of England
could have been part of his defensive strategy. Additionally, the reformation of the army led
to smaller units of a fifth of the size of the former legions, which would make a vast area of
the fort obsolete.
Later political events within the Roman Empire would take a toll on the garrisons in
South Wales through the changing uses of the military in the area. Initially, the forts were
built and designed strategically for internal policing of the area. Changes in the late third
century shows evidence of a new purpose for the coastal locations of Cardiff, Loughor and
Neath. By this time, most military establishments in South Wales had been abandoned,
including Neath and Loughor, supporting a strategic coastal regarrison initiative. Also during
this time, evidence suggests that at least a part of Legio II Augusta was stationed at
Richborough.389 Early fourth century abandonment is attested to at garrisons throughout
South Wales, yet a desire for security was still felt in the area, as seen in a probable garrison
at Caerwent.390 The last withdrawal of forces in the end of the fourth century would have
made an obvious statement to the British that from that point they were on their own.
Soon after the pacification of the Silurian territory, the Roman military strategy was
apparent: control of the natives through military establishments in strategic locations
throughout South Wales. During the subsequent centuries, garrisons in the area would
decrease due to continual actions of the Roman military outside Wales and Britain, but the
continual pacification and cultural assimilation of the natives into the Roman lifestyle most
likely also brought about a more peaceful coexistence and thus less need for a strong military
presence. Eventually, the decline of the Roman Empire in Britain would result in the
abandonment of military establishments, leaving the native populations to continue through
the urban centres.

387
Wilson, D.R. (1970). ‘Roman Britain in 1969: I. Sites Explored’, Britannia. 1, 272; Zienkiewicz
(1993), 85.
388
Boon (1972), 67-9; Evans and Metcalf (1992), 74-5.
389
Frere, S. (1987). Britannia. 3rd ed. London: Routledge and Kegan Paul, 175.
390
Knight, J.K. (1996). ‘Late Roman and Post-Roman Caerwent: Some Evidence from Metalwork’,
Archaeologia Cambrensis. 145, 46; Arnold and Davies (2000), 57.
104

Chapter 10: Urban Life


Urban centres were a key import of the Romans in the first century AD. In the effort
of the Romans to govern, control, tax and ultimately ‘civilize’ their newly conquered lands
with Roman culture, urban lifestyles were created centred upon coloniae, municipia and
civitates.391 Each civitas was designed as a tribal centre which provided provincial
organisation, administration, business, amenity, entertainment and any other aspects of
Roman urban life. This establishment of a civitas centre may have also reflected
‘trustworthiness, political maturity and a willingness on the part of tribal elites to copy
Romanised institutions,’ all the while implying that the locals had the financial basis required
to invest in the newly formed politics of the area.392
In the territory of the Silures, a civitas-capital was established called Venta Silurum
(Caerwent) in modern Monmouthshire. This tribal centre is only mentioned in two
documentary sources: the third century AD Antonine Itinerary as Venta Silurum, and the
seventh century AD Cosmography of Ravenna as Ventaslurum.393 The location of the town,
later to be named Caerwent, is nine Roman miles (8 English miles) east of the legionary
fortress Isca (Caerleon) on one of the main Roman roads across South Wales. Of the two
roads mentioned by the Itinerary as starting from Caerleon, on Iter XIV sits Caerwent before
leading to Bath and ending at Silchester.394

The Origin of Venta Silurum


Caerwent’s beginnings are still currently unknown, although much recent
archaeological work has been completed on the town. Pottery evidence points towards some
Roman context in the second half of the first century AD, however the earliest structures
found so far are of timber dating c. AD 100.395 Some sort of settlement is expected on the site
by the late first century AD as the Roman occupation of South Wales included the area east of
the Usk river since the mid-50s AD. Yet, this may have just been a small settlement with a
lack of depth centred upon the Roman road, as suggested by second century AD cremation
burials. Other excavations in the town have shown the existence of buildings only after the
foundation of the civitas. Construction of Caerwent was along perpendicular roads creating
twenty insulae: I through XX (Map 7). Much of the gridwork structure has been maintained
over time, and modern excavations are sited within these numbered insulae.

391
These classifications vary in distinction, mostly falling upon the status of the inhabitants within the
Roman Empire: see Wacher (1995), 15-6.
392
Arnold and Davies (2000), 45.
393
Rivet and Smith (1979), 493.
394
Rivet (1970), 58-60; Rivet and Smith (1979), 176-8; Mattingly (2006), 258: table 8; 259.
395
Arnold and Davies (2000), 49; Brewer (2004), 219.
105

Map 7: Venta Silurum (Caerwent)


© Cadw, Welsh Assembly Government (Crown Copyright)

Archaeology suggests urbanization in Roman Britain peaked early from AD 90-


120.396 The existence of a forum-basilica is generally believed to be the prime indicator of a
Roman Britain civitas-capital. Nash-Williams first places the existence of the forum-basilica
(Insula VIII) at Caerwent by the end of the first century due to mint condition coins of the
Nerva reign (AD 96-98) found during excavation.397 Lately, a Hadrianic foundation for the
civitas has been suggested due to recent excavations at the site where coins, architecture and
ceramic evidence of second century AD date were found.398 This latter view coincides well
with the influx of municipal building in Britain by Hadrian after his visit in AD 121 or 122.399
Withdrawal of some garrisons in South Wales that were to be used building the wall may
imply that the Silures were pacified enough to require less military control. Thus the
founding of Venta Silurum at this time may have been to supplement Legio II Augusta at
Caerleon with tribal administration until an eventual release of control to the provincial tribal
council.

396
Brewer (2004), 207.
397
Nash-Williams, V.E. (1954b). ‘The Forum-and-basilica and Public Baths of the Roman Town of
Venta Silurum at Caerwent in Monmouthshire’, Bulletin of the Board of Celtic Studies. 15, 163.
398
Brewer, R.J. (1993b). ‘Venta Silurum: A Civitas-Capital’, in Greep, S.J. (ed.) (1993). Roman
Towns: The Wheeler Inheritance. York: Council for British Archaeology Research Report 93, 61; 65;
Arnold and Davies (2000), 49; Brewer (2004), 221.
399
Wacher (1995), 378.
106

Alternatively, an indication of a Roman civitas has also been linked with certain
public amenities such as a bath-house, adequate running water and drainage and town
houses.400 Nash-Williams found earlier Flavian pottery during his excavations of the first
bath-house (Insula XIII) and nearby domestic buildings at Caerwent, however Wacher
dismisses this as remnants of the vicus at Caerleon.401 After the conquest over the Silures, the
area may have grown enough to merit a bath-house, although this does not assume the
founding of the civitas. By the middle of the second century AD, Venta Silurum presumably
existed as a civitas-capital of the Silures, hitherto only through the presence of a forum-
basilica and a few amenities. The site may have taken longer to become established within
South Wales, as may be seen in the regular street grid, also synonymous with civitas town
status, not reaching its final form until the late second to early third centuries AD.402
With dating evidence established, the reasoning behind the founding of Caerwent in
its location still remains highly questionable. Canabae and vici developed around most forts
of Roman Britain, unless the military occupation was short lived. From these settlements, a
town could later be established by the Romans for provincial administration into a civitas, and
in the majority of the Roman towns of Britain, this was the case. Until recently, it was widely
assumed that the origin of Venta Silurum came from the existence of a fort predating the
civitas.403 This would give a military background to the urban settlement and the surrounding
area. However analysis of the archaeological evidence suggests this is highly unlikely, as
typical military evidence, such as barracks and administrative buildings, has not been
discovered.404 Furthermore, the placement of the legionary fortress, Caerleon, nearby
sufficiently controlled the area in which Venta Silurum was placed, thereby making a fort at
that location unneccessary. Thus, the reasoning behind the placement of the civitas at
Caerwent remains unknown.
At Corinium Dobunnorum (Cirencester) and Durnovaria (Dorchester) forts were
placed near tribal centres. Those tribal centres were eventually abandoned by the natives who
took advantage of the monetary draw of the nearby vicus. From these vici, civitas-capitals
were later established.405 This same process most likely did not happen in South Wales as
there was not a significant tribal centre due to the decentralised political structure of the
Silures, as stressed in the first part of this thesis. The only nearby tribal centre would have

400
Wacher (1995), 20.
401
Nash-Williams, V.E. (1930). ‘Further Excavations at Caerwent, Monmouthshire, 1923-5’,
Archaeologia. 80, 230-2; Nash-Williams (1954b), 164; Nash-Williams (1956), 108; Wacher (1995),
379.
402
Wacher (1995), 379; Brewer (2004), 219.
403
Millet (1990), 102-3: table 4.4; Brewer (2004), 238: footnote 22.
404
Webster, P. (2003). ‘An Early Fort at Caerwent? A Review of the Evidence’, in Wilson, P. (ed.)
(2003). The Archaeology of Roman Towns: Studies in Honour of John S. Wacher. Oxford: Oxbow,
214-20; Brewer (2004), 217-9.
405
Wacher (1995), 29; 324 (respectively).
107

been Llanmelin two kilometres to the north-west, with the connection of the hillfort with
Sudbrook, which was no doubt in use by the Romans, although possibly abandoned by this
time.406 Nash-Williams viewed the area as a centre of Silurian habitation, with the small
settlements in the area centred upon Llanmelin as evidence of such. 407 Yet, as is shown in the
settlement pattern of the whole of South Wales, this area was by no means the tribal centre of
the entire Silurian territory.408
The Dumnonii and Cornovii also exhibited decentralised political structures. The
Roman response to this at Isca Dumnoniorum (Exeter) and Virvonium Cornoviorum
(Wroxeter), respectively, was to establish the civitas-capitals at abandoned fortresses which,
probably including the vici, were the largest concentration areas of each tribe. 409 With this in
mind, the logical belief would be to establish the civitas centre at the former legionary fortress
at Usk upon its abandonment by AD 70. Yet Manning believes the new territorium of
Caerleon included much if not all of that belonging to Usk, and therefore leaving the fortress
to the Silures would have left them without much land on which to build a settlement. 410
However, the Silures were not wholly pacified by AD 70. Only the eastern territory from the
Usk to the Wye can definitively be seen under Roman control, thus an establishment of a
civitas centre at that location and at that time may not have been a valid option. Likewise, a
possible establishment of Venta Silurum at Caerleon after its abandonment by Legio II
Augusta would have been reasonable, yet obviously this is not the situation. 411 It was
common Roman practice to govern and control decentralised tribes from nearby military
posts; this is seen most readily in the placement of the forts of South Wales explained
herein.412 Since this initial military presence served as not only control over the local
inhabitants but also for protection, the continued garrison at Caerleon during the founding of
Caerwent presumably indicates a desire or need for this presence near the new civitas for
either or both reasons. Use of the legionary fortress for early protection is suggested in the
late addition of defences around Caerwent, as is the case with most developing settlements of
Roman Britain.413 Therefore, with the continued presence, albeit minimal, of Legio II

406
Nash-Williams (1939), 54.
407
Nash-Williams (1956), 104.
408
See above, page 22
409
Wacher (1995), 335-7; 362-3 (respectively).
410
Manning (2004), 190.
411
A recent geophysical survey by Cardiff University outside Caerleon has found the existence of
multiple buildings, previously unknown. They may be a part of a harbour facility, as they are near the
Usk River, however one interpretation suggests they may be the beginning of an effort to urbanise the
area for an eventual release to the Silures. See:
http://www.cardiff.ac.uk/share/research/projectreports/miskin/index.html. Last Date of Access:
7/22/11.
412
See above, page 91; Appendix 2, page 156.
413
Wacher, J. S. (1978). Roman Britain. London: Dent, 70; 160.
108

Augusta at Caerleon, the civitas could not be formed from the fortress and its vicus, and
another site was needed.
Wacher puts forward the idea, albeit in passing, that, similar to the establishment of
Caesaromagus (Chelmsford) for the Trinovantes, Venta Silurum was a gift to the Silures from
Rome.414 This is also highly unlikely, which presumably Wacher also believes due to his lack
of argument for the theory, as the Trinovantes were considered a friendly ally of Rome at the
time, and one cannot believe a similar sentiment would have existed in South Wales less than
a century after the conquest of the area. Conversely, after the defeat of the Silures by Julius
Frontinus, the tribe was certainly viewed by the Romans as dediticii (surrendered people)
without rights. This could have resulted in a treatment of the Silures similar to the Roman
General Corbulo had to the Frisii, whereby the general delimited a reservation where he
forced the Frisian clan to settle with his own appointed senate, magistracy and laws. 415 From
then, possibly the location of Venta Silurum was the closest to a tribal centre at the time of
founding.
Ultimately, the development of the civitas centre may have been restricted to the level
of resources available, and the motivation of certain groups, including retired soldiers from
Legio II Augusta and the local free-born, land-owning males (appointed to an ordo416), to
provide the town with particular essentials. 417 This may give the late first century AD date
some credibility as a foundation at this time could have been enacted, yet the markers of a
civitas as discussed above came about later through varying reasons.
In the first half of the second century AD, some forces were withdrawn from South
Wales to help with the construction of Hadrian’s Wall. An establishment of the civitas-
capital near to this time would suggest that Rome wished to compensate for the smaller
military presence by giving some provincial administrative power to the Silures. This transfer
of some power to the natives may have been in part a reward for continued peace over the
decades, recognition by the Romans to a new generation of Romanized Silurian leaders with
high intellect or political consciousness 418, or as a necessary resort to compensate for the
smaller military force. It must be noted that this was not a complete turnover of power to the
Silures: retired veterans and wealthy Roman civilians still held political power throughout
Britain, including South Wales. Either way, Rome was not prepared to abandon Caerleon at
that time; therefore the civitas needed an alternative location. Possibly the area on the Roman
road that later became Venta Silurum had grown in importance, although maybe not in size,
due to its central location among Caerleon and the mining operations at Draethen and Risca to

414
Wacher (1995), 208; 379.
415
Tacitus. Annals. 11.19.
416
See below, page 120.
417
Jones, M.J. (2004), ‘Cities and Urban Life’, in Todd (2004), 169.
418
Davies (1995), 695; Arnold and Davies (2000), 45-6.
109

the west, the Roman ferry crossing at Sudbrook to the south, and Gloucester and the majority
of Roman Britain to the east. Furthermore, keeping the civitas centre on the eastern side of
the Caerleon territorium may have been for the purpose of better policing the Silures in case
the new found freedom and responsibility caused problems. No matter the case, by the end of
the second century AD, the Silures were self-governing themselves in some manner with the
tribal centre located at Caerwent.

Other Urban Centres


Other urban centres existed connected to military installations: canabae adjacent to
legionary fortresses and vici next to forts. These would have been the result of the civilian
camp-followers that accompanied the military on campaigns. 419 These followers would have
comprised servants, slaves, entertainers of all kinds, possibly wifes of the soldiers and most
importantly the merchants always looking for financial opportunities. Most likely, native
populations would not have been involved in the initial development of the urban centres
attached to military installations, yet it would not have been long before the opportunities for
wealth attracted the locals. Such a pattern is seen as the foundations of Corinium
Dobunnorum (Cirencester) and Durnovaria (Dorchester).
The canabae around Caerleon existed on at least three sides of the fort encompassing
an area of no less than 31,900m2.420 Settlement outside the fortress was partly military in
origin early, but then later handed over to civilian expansion. By the mid-second century AD,
the development of the canabae may have been inhibited by the growth of Venta Silurum
only 13 km south-east, yet this did not prevent the extra-mural settlement from achieving its
currently known large size as well as the small roadside settlement at Bulmore. 421 Although
the garrison in the fortress declined over the years, military activity has been dated to c. AD
380, and extra-mural activity in the canabae also persisted to at least this date.422
Outside the fortress, two baths, a parade ground and an amphitheatre are known. 423
The fortress baths were located inside the walls, so a primarily civilian usage of the outer two
can be presumed, while the amphitheatre and parade ground were probably more for the
provision of the legion, again a civilian usage cannot be dismissed. Thus entertainment must
have also been a big draw to the fortress and allowed for the canabae to receive the economic
rewards during such events.
Two kilometres east of Caerleon sat a roadside settlement, Bulmore, on the road from
the fortress to Usk.424 Obvious connections to the fortress, and possibly the latter site, may be
419
Brewer (2004), 209.
420
Evans (2000), 459; Brewer (2004), 209-13.
421
Boon (1972), 61-2; 69-70; Evans (2000), 6.
422
Evans and Metcalf (1992), 207.
423
Manning (2004), 197-8.
424
Zienkiewicz, D. (1983). ‘Great Bulmore (ST 360 915)’, Archaeology in Wales. 23, 47-9;
Zienkiewicz, D. (1984). ‘Great Bulmore (ST 360 915)’, Archaeology in Wales. 24, 57-8; Arnold and
110

found through the gravestone of veterans of Legio II Augusta and their families. It is believed
that up to 200 veterans left the legion each year, and upon retirement, were granted land or a
large sum of money as payment.425 Thus it must be expected that some if not most remained
within South Wales, and it is apparent that a few selected to live near their final garrison. The
settlement was founded in the last decade of the first century, soon after the pacification of the
Silures and completion of the legionary fortress, and retains a civilian settlement nature even
with the legionary connection. The history of Bulmore fluctuated, with decades of intense
activity followed by virtual abandonment. Four such cycles of occupation then neglect
culminated in the end of the settlement habitation at the end of the third century AD. These
cycles are believed to have been related to the garrison or absence of Legio II Augusta at
Caerleon, as well as the proximity of Caerwent possibly preventing the roadside settlement
from developing further.
Occupation outside the fort at Cardiff from late first century to mid-third century AD
was focused on modern High Street running north into the fort, and consisted of some tile
production and metalworking.426 Some extra-mural features have been revealed outside the
pre-Flavian Usk fortress that may be remnants of a vicus.427 Pre-Flavian auxiliary forts at
Abergavenny and Monmouth have yet to reveal evidence of vici, while a few remains of the
expected vicus at the Flavian fort in Neath have recently been unearthed. 428 A bath-house,
parade ground and a tile and pottery kiln were found outside the fort at Gelligaer. 429 These do
not guarantee the presence of a vicus but civilian and native populations would have been
more likely to have settled with such amenities present; possible evidence of such may lie
under the stone fort.430 Further finds, including bath-houses, outside Brecon, Coelbren, Pen-
y-Darren and Loughor suggest vici of unknown sizes, but not much more can be
substantiated.431
From the nineteenth century, Roman finds in Cowbridge hinted at a substantial
settlement in the area, which was confirmed in the late 1970’s. 432 A settlement in this locale
seems probable as the area between Cardiff and Neath is scarce of urban Roman occupation,
and such a site on the Roman Road Iter XII between the two forts would have proved a

Davies (2000), 59; Brewer (2004), 213-5.


425
Brewer (2004), 208.
426
Webster (1984), 297-8.
427
Marvell, A.G. and Maynard, D.J. (1998). ‘Excavations South of the Legionary Fortress at Usk,
Gwent, 1994’, Britannia. 29, 258.
428
Maynard, D.J. and Marvell A.G. (1993). ‘Cwrt Herbert, Neath, Glamorgan (SS 745 979)’,
Archaeology in Wales. 33, 60; Lawler, M. and Marvell, A. G. (1994). ‘Neath, Dwr-y-Felin Road’,
Archaeology in Wales. 34, 54; Brewer (2004), 214.
429
Brewer (1996), 16-20.
430
Manning (2004), 238: footnote 15.
431
Brewer (2004), 215.
432
Robinson (1980), 21-3; Parkhouse, J. and Evans, E. (eds.) (1996). Excavations in Cowbridge, South
Glamorgan, 1977-88. Oxford: British Archaeological Reports, British Series 245.
111

convenient location for the existing rural settlement in the area. Numismatic evidence
suggests continual activity from Flavian period until the third quarter of the fourth century
AD.433 A possible official origin may be surmised in the early second century bath-house,
which was later remodelled and extended. Such municipal structures may be evidence of
Roman influence, as mentioned at Caerwent.434 However it was quickly out of use by the
Hadrianic period. Coinciding with this, a military foundation was once proposed due to
legionary tiles found within the bath-house, yet there are no features in the area that suggest a
fort; instead possibly the settlement started as a mansio or mansiones and then grew. Another
origin may exist in the nearby hillfort Caer Dynnaf, where the Iron Age population may have
over time moved to the Roman road for opportunistic purposes similar to the native vici at
Corinium Dobunnorum (Cirencester) and Durnovaria (Dorchester).435
The site of Bomium (Bovium) has been debated from at least the sixteenth century
through archaeological and etymological means, and interpretations of the Antonine Itinerary
have been added but lead to no conclusive location. 436 With the discovery of the site at
Caergwanaf, much debate centres on that possible location for Bomium as well.437 The
importance of the settlement is also unknown and nothing is present to suggest Cowbridge as
a prominent urban location.
Sites with little to no urban settlement during military occupation sometimes
developed post-garrison roadside settlements.438 At Usk, there is evidence of iron working
from the mid-second to mid-fourth century AD; possible military interest at the site may be
because the area is still within the territorium of Caerleon. Metalworking at Monmouth also
dates from the same period and suggests a settlement based on the local iron deposits.
Abergavenny is interpreted as Gobannium, which is listed on Roman road Iter XII in the
Antonine Itinerary where a possible mansio existed, and Roman finds and burials support
such a claim.439 Similar finds also suggest a roadside settlement at Chepstow, an important
point of crossing over the Wye river.

Settlement Patterns
In the inital years after the founding of Venta Silurum, residences within the town
would have mirrored those found outside the town limits. However, with the expanding
responsibilities of the civitas-capital as an economic, political and even Romano-British
cultural centre, a more urban lifestyle would have emerged overtime. This would have

433
Lloyd-Fern, S., Boon, G.C. and Sell, S.H. (1996). ‘The Coins’, in Parkhouse and Evans (1996), 180-
2.
434
Wacher (1995), 20; see above, page 104.
435
Robinson (1980), 26: n. 61; Arnold and Davies (2000), 88.
436
Robinson (1980), 23-4.
437
See above, page 97.
438
Arnold and Davies (2000), 62-4; Brewer (2004), 215.
439
Rivet and Smith (1979), 173-5; 369.
112

pushed the natives into adapting to a way of life vastly different from the Iron Age
convention. The forum-basilica and bathhouse would have both been drastic changes to
traditional Silurian religious and cultural lifestyles. As the need for more commercial shops
grew, the common practice was for the keepers to connect their residential quarters behind or
above the workshop. Accommodation for wealthier merchants, military veterans from
Caerleon, immigrants and travellers and even any council members embracing a Roman
lifestyle would have demanded larger houses most likely within the town. 440 These buildings
often mimicked ‘villa’ style constructions as found in the rural landscape with corridors and
multiple rooms situated in rectangular blocks, possibly surrounding a central courtyard. In
Caerwent, and most of Britain, the style of house differed from the Mediterranean precedence
and is often attributed to the military architecture. 441 Depending on the placement within the
town and availability of space at the time, the building could have been highly compact or of
an expansive design. Furthering the Roman culture of these structures, earlier construction
would be of timber-framed wattle and daub, then later of limestone and sandstone or stone
foundation with higher timber support. Within, painted plaster, glass windows and tile
roofing would have surrounded and enclosed stone, wooden or even mosaic flooring. All of
these features would have been an impressive site to the Iron Age Silures, as the by the fourth
century AD, the internal area of walled Roman Caerwent would have looked nothing like that
found within the larger Iron Age hillforts.
Much of the most substantial urban building in Caerwent is from the third century AD
or later.442 As mentioned, the regular street grid was not completed by the early third century.
In Insula I, the earliest building was constructed in the late second century or early third
century AD, which was later demolished and rebuilt within the century. A larger, more
substantial residence was built in the early fourth century AD and housed many rooms
surrounding two courtyards. Further survey within the same insula suggests this latter house
was connected to other nearby buildings of agro-pastoral nature. The Pound Lane shops,
explained further below, have a late first century origin, but subsequent additions became
most prominent in the early third century. Stone walls were constructed in front of pre-
existing earthwork defences during the early to middle of the fourth century, as well as
external towers on the north and south ends. Next to the Romano-Celtic temple, a long strip-
building, which served as a workshop and opened onto the main Roman road, remained as
such well into the third century. Although of a religious nature, the Romano-Celtic temple’s

440
Wacher, J.S. (1989). ‘Cities from the Second to Fourth Centuries’, in Todd, M. (ed.) (1989a).
Research on Roman Britain: 1960-89. London: Britannia Monograph Series 11, 108-11; Brewer
(1993a), 20-1; Arnold and Davies (2000), 55.
441
Allason-Jones (2008), 59.
442
Brewer (1993a), 33-7; Brewer (1993b), 58-9; Arnold and Davies (2000), 53-5; Brewer (2004), 228-
32.
113

construction c. AD 330 further indicates the substantial building regime of the later Roman
town.
The characteristics of Roman town building and planning were foreign to the Silures
in the first century AD. As the civitas-capital came into its own as a centre of the tribal
territory, substantial construction was necessary to accommodate the changing economic and
political environments. As is evident from the archaeology, over time the city gained more
prominence and importance within the community and continued to grow larger to support
this role well into the fourth century.

Urban Economic Influences


From the beginning, the military was a strong drain on local resources, but in turn
became a large source for economic wealth in the area. 443 The military strained the limits of
the South Wales agricultural and ceramic resources, while exploiting the natural mineral
deposits. All of this would have created a steady flow of wealth throughout the area and into
the pockets of military, civilian and native alike.
Vici, canabae and the civitas centre at Caerwent would have been the local suppliers
of the needs for those within the surrounding country as well as the military, and it should be
assumed that they all provided much of the same resources: pottery, metal goods and leather
goods, for example.444 With the location of Venta Silurum on the Roman road Iter XIV
between Caerleon and the east, as well as the logical connections with Sudbrook to the south-
east and the mining at Lower Machen in the west, economic prosperity through these
crossroads must have been the key to the development of the civitas centre. Although this
may not have been the initial reasoning behind the placement of Caerwent, steps toward
utilising the location must have taken place not long after its founding.
Initially upon founding of the town, it is believed that natives built and funded the
construction, albeit with help.445 As these settlements developed Roman character,
architecture and building practices must have been introduced through architects from either
neighbouring provinces or the local military. Similarities between towns around Britain stress
this consistent Roman influence upon the natives during building, while the differences also
reflect the local impact. Economic stimulus towards these building practices were provided
by, amongst others, governors such as Julius Frontinus and Agricola who provided grants,
and Emperor Hadrian who remitted taxes.446
Along with this, the continual supplying of the garrison at Caerleon, which was often
to the detriment of locals, at times may have been a great economic opportunity. Merchants

443
Davies (2004), 108-9.
444
de la Bédoyère, G. (2003b). Roman Towns in Britain. Stroud: Tempus, 129.
445
Wacher (1978), 70; de la Bédoyère (2003b), 76-81.
446
Tacitus Agricola 21; Historia Augusta, Hadrian. 7.6; Dio. 69.8.1.
114

settled within vici attached to forts may have seen the founding of Caerwent as a chance to
centralise their business. Local resources were preferred for obvious reasons, but at time long
distance trading and transporting of goods was necessary, and provided civilians with
economic stimulus through military contracts.447 Thus contracts issued through Legio II
Augusta must have often benefited those nearby at Venta Silurum. With this in mind, the
economic influences of the military are closely tied with the surrounding urban areas, and as
such both will be discussed herein.
The forum-basilica at Caerwent represents the common economic, administrative
and, with a Romano-British temple nearby, possibly religious centre of the Romano-British
civitas, although on a relatively small scale when seen alongside other towns of Britain.
Venta Silurum was designed as a market centre for the area, and the construction of the forum
with surrounding market stalls indicates this best. 448 There, shops could be rented and
temporary stalls set up on certain market days. Unfortunately not many goods remain within
the archaeological record, but some indications like furnaces or tools can suggest blacksmiths
or other related professions.449 One room of the forum contained oyster and mussel shells,
and was interpreted as a fishmonger’s shop or oyster bar. 450 On the far end of the block west
of the forum-basilica sit the Pound Lane Shops. 451 These shops have a late first century AD
origin with additions and changes on three occasions up through the mid fourth century, and
consist of residential areas behind commercial rooms which front onto the main Roman road
(Iter XIV) running east to west. The owner of the western of the two shops worked as a
blacksmith which can be deduced through a hearth and iron slag found in excavation; the
other shop trade remains unknown. The presence of these and other shops along the main
Roman road suggest the forum was not large enough to accommodate the growing
commercial centre Caerwent became over time. This is consistent with the smaller, more
densely packed layout of the entire town when compare to others around Roman Britain. 452
Possibly similar to the Lydney-Llanmelin Iron Age ware of South Wales, grey-ware
has been found associated with kilns within Silurian territory, suggesting a Romano-British
ceramic tradition possibly from ‘immigrant’ potters that arrived with the military. 453 Pottery
works producing this type have been found at Caldicot, 454 Llanedeyrn455 and Llwyn-y-grant,

447
Mattingly (2006), 511-3.
448
Nash-Williams (1954b), 159-63; Brewer (2004), 227.
449
Wacher (1995), 66-8.
450
Nash-Williams (1954b), 160.
451
Brewer (1993a), 33-7.
452
Wacher (1995), 388.
453
Arnold and Davies (2000), 109.
454
Barnett, C., Stanley, P., Trett, R. and Webster, P.V. (1990). ‘Romano-British Pottery Kilns at
Caldicot, Gwent’, Archaeological Journal. 147, 120-4.
455
Vyner, B.E. and Evans, G.C. (1978). ‘The Excavation of a Roman Pottery Kiln at Llanedeyrn,
Cardiff’, in Boon, G.C. (1978). Monographs and Collections: Relating to Excavations Financed by
H.M. Department of the Environment in Wales. I. Roman Sites. Cardiff: Cambrian Archaeological
115

Penylan; the latter two are near Cardiff and all are of second and third century AD date. At
Rogiet456 and Church Farm, Caldicot,457 both on the Gwent Levels south of Caerwent, the
majority of the ceramic assemblages consist of the same second century AD grey-ware. At
Caerleon and Cowbridge, local grey-ware has also been present in the pottery finds, although
of less density.458 As pottery of this type was probably a low-value commodity, it likely
travelled only short distances from the kilns. This implies a local ceramic industry originating
in a rural setting, with the trading centre at Caerwent (for the Caldicot kilns) and Cowbridge
(for Llanedeyrn and Llwyn-y-grant) and distribution radius not more than fifteen kilometres,
although connections with the military may have extended the circulation further. 459 This may
be seen at Neath in the single sherd of mortaria possibly of South Wales grey-ware type,
although this single example provides little conclusively. 460 The nearest known kiln for South
Wales grey-ware is at Llanedeyrn, near Cardiff, thus coming from a long distance, yet a more
local kiln nearer to Neath would have most likely resulted in a stronger presence of the grey-
ware within the ceramic assemblage. However, connections with sites across the Severn
Estuary and the kiln at Congresbury, North Somerset suggest an overlap of the pottery
distribution around the Severn.461 This overlap exists mostly through that produced at
Caldicot located roughly in between Caerwent and Sudbrook. With the kilns only 2.5 km
away south of Caerwent and 3 km west of Sudbrook, the produce could have been taken to
the town for local distribution or the estuary crossing for sale in Somerset. Yet, Barnett et al.
insist on Caldicot having an almost exclusive South Wales distribution. 462
Caerleon ware, produced particularly for Legio II Augusta while garrisoned at
Caerleon, is a highly distinctive product starting most likely in the Flavian or Flavian-Trajanic
period.463 From the late first or early second century AD to the mid-third, local kiln producing
for the garrison helped establish, along with the South Wales grey-ware, a local ceramic
industry which then lessened the dependence upon continental imports. 464 At the Mill Street
excavations of the Caerleon canabae, at least half of the ceramic assemblage was of local red-

Association, 120-9.
456
Marvell, A.G. (1996a). ‘Rogiet (Housing Allocation) H2RO1 (ST 456 829)’, Archaeology in Wales.
36, 78.
457
Insole, P. (1997). ‘Church Farm, Caldicot (ST 4816 8916)’, Archaeology in Wales. 37, 73.
458
Parkhouse and Evans (1996), 238; Compton, J. and Webster, P.V. (2000). ‘The Coarse Pottery’, in
Evans (2000), 257-9: Figs. 61-2.
459
Millet (1990), 123; Mattingly (2006), 516-8.
460
Evans, D.R. (1992). ‘The Course Pottery’, in Marvell and Heywood (1992), 263.
461
Spencer, B. (1988). ‘The Later Roman Grey-Ware Industry in South Wales’, in Vyner and Allen
(1988), 117.
462
Barnett et al. (1990), 143-4.
463
Seager Smith, R.H. (2000). ‘The Mortaria’, in Evans (2000), 279; For the initial debate over
Caerleon ware see Boon, G.C. (1966). “Legionary’ Ware at Caerleon?’, Archaeologia Cambrensis.
115, 45-66.
464
Arnold and Davies (2000), 111.
116

slipped Caerleon ware. A kiln which produced this ware, amongst other things, has been
found at Kemeys Inferior near Newport, only two kilometres west of the fortress. 465
Both the South Wales grey-ware and Caerleon ware may have partially developed
locally through the remnant pre-existing trade network from the Iron Age and the distribution
of Lydney-Llanmelin ware. As shown, Lydney-Llanmelin ware was a product exclusive to
the Silures and suggested a local ceramic production developed by the natives of South Wales
for local use. Thus, with the development of Roman culture in the area, the new ceramic
traditions may have partially originated from the traditions of the native Iron Age Silures
which persisted into the second century AD.
In the four years after AD 43, or maybe even before the invasion, the Romans may
have found out about the natural resources available in Wales particularly in the south. 466 The
Forest of Dean near Gloucester, in what may have been the eastern extent of Silurian territory,
has an abundance of iron ore, which is a necessity for the Roman military. Iron is also found
in the area around modern Monmouth, the site of one of the early Roman forts in Southern
Wales. Although evidence shows smelting from the second to fourth centuries AD, the
knowledge of iron in the area may still have lead the Romans into the area and have been the
reason for the early fort to secure the resource. 467 Lead mines are also to be found in Draethen
and Risca, 10 km and 12 km west respectively of Caerleon. These mines are connected to a
settlement at Lower Machen,468 but more importantly, they may be located within the
territorium of the new Roman Legionary Fortress at Caerleon. 469 However, more interest in
southern Wales must have stemmed from the gold mined in the Dolaucothi area; the only
known Roman gold mine in Britain.470 The Romans apparently exploited the mine quickly
after the conquest of Wales suggesting they already knew of its existence. 471 In fact, through
the writing of Strabo, the Romans knew of gold in Britain before Claudius even invaded. 472
Mining operations in Britain were initially run by legionary detachments. 473
However, by the end of the first century AD, private mining companies were not uncommon:
the names of which are still known in few instances including the Socii Novaec working in
the Mendips, first worked by Legio II Augusta. Some local ordo (council) had even gained
leases on the mines; this can be seen through lead ingots bearing the name of the tribe of the
Brigantes or Deceangli. It may be possible that a similar situation existed in South Wales
465
Marvell, A.G. (1996b). ‘Celtic Manor Golf Course, Kemeys Inferior (ST 318 910) Newport’,
Archaeology in Wales. 36, 74-75.
466
Arnold and Davies (2000), 4.
467
Brewer (2004), 215.
468
Arnold and Davies (2000), 98.
469
Boon (1972), 17; Brewer (2004), 207. The location of the mines may have been a smaller incentive
for Julius Frontinus to move the legionary fortress from Usk to Caerleon.
470
Symons (2009), 103.
471
Davies and Lynch (2000), 209.
472
Strabo. Geography. 4.5.2.
473
Wacher (1978), 170-1; 193-8; Mattingly (2006), 506-11.
117

either through the council or a private company. As well, with the only known gold mine at
Dolaucothi, some gold working must have reached Caerwent, as there is evidence for it as far
away as Yorkshire and London.474 In fact, with lead mines at Draethen and Risca and iron in
the Forest of Dean, Caerwent may have had a thriving metalworking industry which may
have included native practices used in the Iron Age.475
Existing metalworking practices, already exploiting local resources as apparent
through Iron Age archaeological evidence476, most likely flourished with the coming of the
Romans as the newly introduced building and subsistence techniques by the immigrants no
doubt involved more metal based tools and instruments. 477 Therefore, the new influences on
toolmaking may have created a desire for more advanced metalworking and in greater
numbers. Evidence of ironworking at Caergwanaf, north-east of Cowbridge, suggests work
began before the fort was abandoned but remained after, possibly through official or military
interest.478 In Cowbridge, iron slag suggests a strong metalworking industry with local
sources.479 Usk became an iron working site from the second to fourth century, well after the
abandonment of the fortress, and outside Caerleon, iron and lead objects were found of
military and civilian uses: from spearheads and armour to agricultural and metalworking tools
and slave shackles.480 The last of which most likely supports the sourcing of slaves from the
Roman Frontier, rather than the selling of slaves within South Wales, although this latter
point should not be dismissed. Thus the metalworkers within the urban centres then probably
supplied tools and household goods to support the agro-pastoral subsistence found within the
town and the surrounding countryside.
This subsistence was in no doubt exercised within or immediately outside the town
walls as farms (Insulae I.28N, II.1N, II.2N and XI.7S) have been found on the outskirts
containing barns and byres for both agriculture and animals, and along with these, farming
tools have been discovered in excavations within the town including a billhook, mattock and
shears.481 Furthermore, Venta Silurum as the market centre for the Silures may be best shown
through the originally identified amphitheatre (Insula IV), now most likely believed to be a
livestock market, in the north-east of the town. 482 This suggests that an agro-pastoral
economy was still prevalent and important in the economy of South Wales centred upon the

474
Wacher (1978), 193-4.
475
See above, page 70.
476
See above, page 56.
477
Jones, M.L. (1984). Society and Settlement in Wales and the Marches: 500 BC to AD 1100. Part I.
Oxford: British Archaeological Reports British Series 121(I), 122-3.
478
Burnham (2004), 263-4; Young, T. (2010) ‘Caergwanaf (NGR: ST 044 807; NPRN 403804)’, in
Burnham and Davies (2010), 214-216.
479
Barford, P.M. (1996). ‘The Metalworking Debris from Bear Barn and the Bear Field’, in Parkhouse
and Evans (1996), 205-9.
480
Scott, I. (2000). ‘Objects of Iron’, in Evans (2000), 386-407; Allason-Jones (2008), 13.
481
Brewer (1993a), 19; Brewer (1993b), 58; Arnold and Davies (2000), 53.
482
Brewer (2004), 222; 228.
118

civitas-capital. The nearby Gwent Levels likely played a key role in the local area as the
newly recovered landscape from the military’s drainage provided ample, high quality
grazeland. Continual practice into the fourth century AD and association with high-status
buildings within Caerwent imply the subsistence continued even after the urban immigration
of the time, explained further below.483
Numismatic and exotic goods useage similar to those of the Iron Age Silures
persisted in the Roman Period. In the north-east canabae of Caerleon, coin findings stress the
mid to late third century occupation of the area, but examples of earlier dates, from the end of
the first century AD onwards, are also prevalent. 484 Only one coin of fourth century date was
found, although structural evidence in the area and coins found at the nearby ‘Roman Gates’
give enough proof of fourth century occupation. On the excavation of the ‘Roman Gates’,
more numismatic evidence was found of the late Republican and early Imperial periods, and
of occupation until the mid-fourth century, which ends premature of the expected end of
garrison c. AD 380.485 A similar coin assemblage as that of the canabae is found at
Cowbridge, where late first century to mid to late third century coins (Illus. 6) suggests a
continual occupation.486
A total of fifty brooches were recorded at the Mill Street site, in the north-east
canabae of Caerleon.487 Corresponding with types prevalent in South Wales, two Polden Hill
and five T-shaped brooches were found. 488 Of the other finds, a La Tène II example of the
Late Iron Age is the earliest, and only of non-Roman period origin. Polden Hill and T-shaped
brooches are also present nearby within the fortress, the former fortress at Usk and at the
roadside settlement of Cowbridge (Illus. 7 & 8).489
Numerous intaglio gemstones and rings have been found in and around the Caerleon
fortress dating from the first century AD to the third.490 Depictions in the rings include Roma
and Victory, Hercules and Fortuna; all associated with the military although the last example
(the goddess of good luck) was equally prized by civilians. At Loughor, a fourth century BC
Lysippian sculpture of Achilles is cut in bloodstone (Illus. 9) continuing the trend of

483
Arnold and Davies (2000), 55.
484
Hudson, R. (2000). ‘The Coins’, in Evans (2000), 309-10.
485
McKay, B. (1992). ‘The Coins’, in Evans and Metcalf (1992), 87; Evans and Metcalf (1992), 207.
486
Lloyd-Fern et. al. (1996). 180-2.
487
Lloyd-Morgan, G. (2000). ‘Other Jewellery and Dress Accessories in Gold, Silver and Copper
Alloy’, in Evans (2000), 328-37.
488
See below, page 140.
489
Webster, J. (1992). ‘The Objects of Bronze’, in Evans and Metcalf (1992), 107-10; Webster, J.
(1995). ‘Brooches’, in Manning, W.H., Price, J. and Webster, J. (1995). Report on the Excavations at
Usk 1965-1976: The Roman Small Finds. Cardiff: University of Wales Press, 74-82; 89-90; Lloyd-
Morgan, G. and Webster, J. (1996). ‘Roman Objects of Copper Alloy’, in Parkhouse and Evans (1996),
183-8.
490
Zienkiewicz, D. (1987). Roman Gems From Caerleon. Cardiff: National Museum of Wales; Scott, I.
(1992). ‘The Objects of Iron’, in Evans and Metcalf (1992), 169; Henig, M. and Lloyd-Morgan, G.
(2000). ‘Finger-Rings, Intaglios and Other Ring-Settings’, in Evans (2000), 322-8.
119

militaristic heroic reverence.491 An iron signet ring found at Neath with an intaglio possibly
depicting a Classical Greek event also relates closely to the militaristic iconography found on
legionary personal adornment.492 Two intaglios found at the first century garrison at Usk
show another typical heroic representation and possibly an astrological connection in the
zodiac sign of Scorpio represented by a scorpion.493
Coins, brooches and intaglio rings all represent the influx of wealth that followed the
Roman military. Iron Age South Wales was within an ideally stable socio-economic system
whereby the wealth was owned and controlled by the higher classes, as shown in the elite
goods discussed above.494 Once the Roman military settled in the area, the society of the
Silures instantly changed, and the economic impact can be seen in the military’s wealth and
the vici formed by the civilian followers. It was from this point that the Silures had to choose
either to acclimatize to the new social system, or attempt to maintain the traditional lifestyle
within the new cultural surroundings.
Later Roman Caerwent seemed to maintain economic prosperity evident by the
continual construction and repairs of buildings through the third century and into the first half
of the fourth century AD.495 The basilica had extensive rebuilding in the late third century,
and remained in use in the first half of the fourth century. The substantial house in Insula I
was built in the fourth century, as was the Romano-Celtic temple. Even after the demolition
of the basilica in the AD 360s, the nave was used for metalworking and the nearby shops were
repaired, suggesting necessity and available money was still on hand for such practices. The
Pound Lane shops were altered and repaired up to the middle of the fourth century AD, and
there is even iron-working evidence there in the fifth century, although this latter subsistence
may have been on a small scale supplementary to the lower post-Roman occupation.
The town was planned from the beginning as the market centre of the Silures: the
name Venta Silurum is Latin for ‘market of the Silures’. Archaeological evidence suggests
that the prosperity of Caerwent rarely faltered over the years as the town grew into a greater
economic centre for the natives. The town incorporated the new urban commerce strategy of
the Roman Empire with their Iron Age agro-pastoral subsistence, which became more closely
connected as the number of workshops grew in the town centre and the farming buildings
multiplied in the town’s outer periphery.

491
Henig, M. (1997). ‘Intaglios’, in Marvell, A.G. and Owen-John, H.S. (1997). Leucarum:
Excavations at the Roman Auxiliary Fort at Loughor, West Glamorgan 1982-84 and 1987-88. London:
Britannia Monograph Series No. 12, 394-6.
492
Henig, M. (1992). ‘An Iron Signet Ring’, in Marvell and Heywood (1992), 250-2.
493
Henig, M. (1995). ‘Intaglios’, in Manning et al. (1995), 98-9.
494
See below, page 68.
495
Brewer (1993b), 61-3; Arnold and Davies (2000), 56-7; Brewer (2004), 230-1.
120

Urban Political Life


Civitas-capitals involved mutual understanding between provincial administration
and local tribal leaders, often with a Roman military presence nearby as can be found with
Caerleon still garrisoned as Venta Silurum was founded. In decentralised areas with multiple
auxiliary sites, such as is seen in South Wales, military commanders were initially expected to
help manage and control native populations, including census taking, tax collecting and
settling local disputes.496 These more urban areas near military settlements then brought
about different ideas of government unfamiliar to native populations. Political interactions
between Romans, whether citizens or veterans, and natives often required compromise and
adaptation of cultural customs.
In an effort to smooth the transition of the tribes to Roman lifestyles, native (Celtic)
laws were often adopted on local levels, and governors were consulted by members of staff to
help incorporate native by-laws on issues such as marriage, inheritance and landownership
into the Roman legal system. 497 This transition was also eased by granting certain tribal
leaders rights within the Roman Empire. In coloniae citizens and leading native families
were granted Roman rights, while in municipia the ruling members were given the right to
acquire Roman citizenship, but the rest of the population lived with Latin rights. 498 Since
civitates were identified as lower status than coloniae and municipia, Latin rights were
probably the highest privileges given to its citizens, and then only to those of the ruling
families, while the peregrinae (non-citizens) were without rights except for those given
through local tribal laws. This integration of Roman and native law may have given members
of the Silures opportunity for a career, as men familiar with both sets of laws would have
been useful to argue in cases.499
Although it can be expected that organisation varied with differing local or political
situations, a general rule of tribal political government is believed to have been followed
throughout Roman Britain.500 An ordo (council) usually consisted of up to 100 decuriones:
appointed free-born males of a certain age (believed over 30) with established land ownership
rights; although the number of members cannot be assumed throughout Britain. 501 The
responsibility of the decuriones was in the maintenance of the city financially, including
entertainments and taxes and at a later time the construction of the town walls. By a later
Diocletion decree, any deficit in this latter responsibility would be taken from the decuriones
personal wealth.502 Other smaller responsibilities, including building upkeep and water

496
Wacher (1978), 160-1.
497
Allason-Jones (2008), 11.
498
Wacher (1978), 66; 161; Mattingly (2006), 260-1; Allason-Jones (2008), 49-51.
499
Wacher (1978), 190-1.
500
Allason-Jones (2008), 50-1.
501
Wacher (1978), 162; Millet (1990), 66.
502
Wacher (1978), 169.
121

supply maintenance, were carried out by aediles, while the duoviri iuricundo were in charge
of the lower law courts; all of which were appointed by the ordo.503
Evidence of an established ordo, presumably of native Silurian character, can be seen
in the Paulinus inscription.504 This statue-pedestal was constructed and erected before AD
220, set up by the tribal assembly in honour of a former legate of Legio II Augusta, Tiberius
Claudius Paulinus. This date is attested to the lack of mention of Paulinus as governor of
Britannia Inferior which he was appointed to in AD 220.505 This inscription is also important
in its reference to the tribal council on behalf of the whole of the Silures, rather than the
inhabitants of the town itself.506 This suggests that the council acted, or at least intended to
act, upon the interests of the entire tribe, in the manner of a centralised government.
Although, the ordo may have included non-native members with Roman backgrounds,
including retired veterans from Legio II Augusta, and thus, the efforts of the council may have
included wholly Roman benefits. Therefore, how effective this council was for the entire
tribal territory may never be known, but nonetheless it should be presumed that at least the
local area benefited from this new Romanized political structure.
With the Constitutio Antoniniana of Caracalla in AD 212, much may have changed in
terms of the Romano-British legal system. 507 Although the incorporation of citizenship to all
free-born males would suggest a virtual elimination of the need for native laws 508, it should be
expected that this was not the case. Certain laws must have been kept for local purposes as
they probably still pertained to many matters. If this was not the case, then the passing of a
subsequent law in AD 224 allowing a governor to try cases under local law if he saw fit would
have been unnecessary.509 Moreover, much of Britain probably still spoke their native tongue,
and so were not fully Latinized in all aspects of society, leading this new act to have been
quite useful within Britannia.510 The Silurian tribal leaders, who may have only achieved
Latin rights up to this time, were afforded Roman rights and opportunities for certain
advancement within the empire. This new found opportunity within the politics of the empire
would have characterised the third century AD landscape of natives within Roman Britain, as
with Roman citizenship they could then be considered equal to those who less than two

503
Wacher (1978), 162.
504
Collingwood and Wright (1965), 107: no. 311.
505
Birley, A.R. (2005). The Roman Government of Britain. Oxford: Oxford University Press, 342-4.
506
Brewer (2004), 221.
507
Heichelheim, F.M. (1941). ‘The Text of the ‘Constitutio Antoniniana’ and the Three Other Decrees
of the Emperor Caracalla Contained in Papyrus Gissensis 40’, The Journal of Egyptian Archaeology.
26, 10-12.
508
Wacher (1978), 164.
509
Cod. Iust., 8, 53,1
510
Stevens, C.E. (1966). ‘The Social and Economic Aspects of Rural Settlement’, in Thomas, C. (ed.)
(1966). Rural Settlement in Roman Britain: Papers Given at a C.B.A. Conference Held at St. Hugh’s
College Oxford January 1 to 3, 1965. London: Council for British Archaeology Research Report 7,
109.
122

centuries earlier invaded their Iron Age subsistence with a more Roman lifestyle. It must be
presumed that from the last quarter of the first century AD, the connection of the Silures with
their native way of life would have gradually lessened. With each subsequent generation, a
more Romanized lifestyle must have seemed normal, and with the new citizenship rights,
possibly many natives chose a Roman existence, with which they felt more akin, than their
Silurian past.

The End of Roman Urban Life


The later Roman period was greatly changed from the beginning of Roman Britain.
From the first century AD, the urban centres had evolved from new locales of separate
Romans and natives to single Romano-British economic focal points. This had led many
natives to gain and maintain a wealthy lifestyle, often dependent upon the military. With later
Roman period political and civic changes in the laws and division of Britain into provinces,
the presence of the military often fluctuated. Eventually the military’s involvement with
internal struggles would leave the civilian population behind to continue on their own.
During this time, civitas centres throughout Britain show changes in settlement, building
usage and an increase in defences, all ultimately portraying the necessity for the urban areas
to become self-sufficient.
In South Wales, the continued economic uncertainty and military reduction of the
garrisons left native populations to provide their own prosperity and security from outside
invasions.511 Caerwent seemed to prosper well through the at least the middle of the fourth
century AD, as mentioned above, but Moridunum Demetarum (Carmarthen) in the tribal area
of the Demetae, west of the Silures, began a period of abandonment, neglect and demolition
in the later second century, and never fully recovered. Archaeological evidence suggests a
widespread collapse of activity soon after AD 400, which coincides with Mattingly’s
placement of the rebellion of Britain against Constantine in AD 409 as the end of Roman
Britain.512
Venta Silurum adapted to the reduction in garrison at Caerleon by constructing walls
and external towers on the north and south ends in the early to middle of the fourth century
AD in front of earthwork embankments already present by this time. 513 With the smaller
garrison of Legio II Augusta, a growing insecurity in the rural landscape resulted in more of
the surrounding population moving within the defences of Caerwent, which can be seen in the
increase in agro-pastoral houses built in the fourth century. 514 However, this could also be a
response to new lands, made up of ager publicus (public land) and/or the territorium of

511
Arnold and Davies (2000), 51-7; 143-7; Allason-Jones (2008). 130-7.
512
Arnold and Davies (2000), 57; Mattingly (2006), 530.
513
Millet (1990), 152-3: table 6.4; Brewer (2004), 229.
514
Brewer (1993a), 28; Brewer (2004), 230-1
123

Caerleon, given to the civitas upon the abandonment of the nearby fortress.515 Both pottery
centres in Caldicot were abandoned by the mid-fourth century possibly resulting from this
same pattern.516 Coinciding with the reduced military presence; there may have been a small
military garrison within the civitas during the fourth century, as evident from an increase in
military paraphernalia from this period. 517 At Cowbridge, continuity into the post-Roman
period has not yet been documented, yet the presence of a medieval community suggests
occupational continuation of some kind in the area. 518 Although some activity remained
within the boundaries of Venta Silurum, as mentioned within the Pound Lane shops and along
with local cemetery evidence, the general urban lifestyle that subsisted in the previous
centuries came to an end in the first quarter of the fifth century AD.

515
Todd, M. (1989b). ‘Villa and Fundus’, in Branigan, K. and Miles, D. (eds.) (1989). The Economies
of Romano-British Villas. Sheffield: Department of Archaeology and Prehistory, 14.
516
Insole (1997), 73.
517
Knight (1996), 46; Arnold and Davies (2000), 57.
518
Robinson (1980), 27.
124

Chapter 11: The Rural Society


Foundations for the South Wales Romano-British rural community can be found
within the landscape of the Iron Age Silures, as stressed in the beginning chapters of this
thesis. Individual settlements were often centred upon larger hillforts which provided
subsistence needs beyond what was naturally available in a system based on trade and/or
taxes, while similarly later Romano-British sites were often near vici attached to Roman
military establishments, or along Roman roads. In many ways, this Iron Age affiliation of
smaller settlements with a larger, central location mirrors the Roman Period connection
between the rural community and the urban centres.
In south-east and southern England, villas emerged in the first century AD soon after
the Roman invasion, including the well-known ‘palace’ at Fishbourne (c. AD 75).519 The
prevalence of such Roman buildings is partly due to the influx of Roman wealth that followed
the military into newly conquered lands, as well as the possibly deliberate policy of
urbanization taken by the empire. 520 Examples of the latter can be found through the
evolution of Iron Age settlements to Romano-British farmsteads and villas as seen in the
archaeological record.

Rural Settlement in South Wales


Roman villas521 and native Romano-British farmsteads are often viewed individually
and with references to wealthy Roman immigrants or the military. Yet the differences may
mostly lie within the outward appearances of wealth and Roman culture. 522 These do not
necessarily suggest the inhabitants are native or Roman. The cultural and political changes
associated with the transition from Iron Age to Roman Britain were accompanied by the
physical changes of the local environment in many ways, including the construction of
Romanized buildings. Thus, just as is found within Venta Silurum, the opportunities for
wealth and prestige often resulted in the desire of the natives to take part in a Roman lifestyle
alongside Roman immigrants or even the military. 523 As well, rural settlements must not be
viewed as entirely Roman or British, but as a mix of villas and farmsteads which both
participated in the rural economy. As a villa structurally implies wealth, different socio-
economic levels may also be presumed, yet one cannot view every Romano-British farmstead

519
Dark, K. and Dark, P. (1997). The Landscape of Roman Britain. Stroud: Sutton, 64; Cunliffe (1998).
520
Tacitus Agricola 21. For the interpretation of this passage as evidence for the Romanization
practices of Agricola see Millet (1990), 69.
521
For the varying definitions of a ‘villa’ see Rivet (1969), 175-82; Percival, J. (1989). ‘The Villa
Economy: Problems and Perspectives’, in Branigan and Miles (1989), 5; Dark and Dark (1997), 43-4;
Smith, J.T. (1997). Roman Villas: A Study in Social Structure. London: Routledge, 10-1.
522
Dark and Dark (1997), 43-64.
523
Black, E.W. (1994). ‘Villa-Owners: Romano-British Gentlemen and Officers’, Brittania. 25, 99-
110.
125

as socially or even economically inferior to villa estates. Overall, the rural community was an
ever-changing society that adjusted to the drastic differences accompanied by the Roman
invasion. The local tribal community most likely took to their new social, economic and
political environments by different means; with each site individually choosing how much it
was willing to adapt to the Roman culture or remain true to the Iron Age origins.
South Wales contains some of the best indicators of high wealth settlements in Wales.
The reasoning behind the larger number of high wealth sites in South Wales may be in the
resistance to Roman culture elsewhere, or the fact that the agricultural abilities in the lowland
coastal area of the south lends itself to more attractive settlements. 524 Davies classifies the
rural settlements of South Wales into categories which vary in defensive nature and current
state within the archaeological record. This is a very superficial organisation as what can
currently be seen at sites can be misleading towards what was present during the Iron Age and
Roman periods; this was stressed in the beginning chapters of this thesis. Furthermore, the
classifications do nothing to separate villas and farmsteads, as both settlements types can be
found within each category. It must be said here that with the various definitions of a villa
and the lack of decisive answers from the archaeological record, the differences between a
site being characterised as a Roman villa or a Romano-British farmstead can at times be hard
to recognize.
Unfortunately, there are never enough resources available to excavate or even
document sites fully. This can leave many locations destroyed further over time due to lack
of protection, such as at Hadnock Roman Villa in Monmouthshire where an ironworking
furnace base recorded in 1981 has subsequently been destroyed and scattered due to continual
ploughing.525 Furthermore, some sites recorded briefly or in obscure publications may be lost
over time, while newly documented sites are listed regularly in journals such as Archaeology
in Wales; thus the catalogue herein should not be viewed as complete (Appendix 1). Other
publications have attempted to establish contemporary lists of Romano-British settlements
and/or villas, but caution should be taken as documentary evidence can be incomplete or out
of date.526 In 2003, A Research Framework for the Archaeology of Wales released a list of
key Roman sites in the Silurian territory of South Wales. 527 Here, settlements are specified
which provide the best understanding of the rural settlement pattern in the area, stressing the
structural and artifactual assemblages which signify Iron Age and Roman occupation (Map

524
Arnold and Davies (2000), 65; 76-80.
525
Burnham (2004), 258.
526
See Davies, W. (1979). ‘Roman Settlements and Post-Roman Estates in South-east Wales’, in
Casey, P.J. (ed.) (1979). The End of Roman Britain. Oxford: British Archaeological Reports British
Series 71, 165-8: Appendix A; Scott, E. (1993). A Gazetteer of Roman Villas in Britain. Leicester:
Leicester Archaeology Monographs 1, 67-8; 80-1; 163; Brewer (2004), 234-6.
527
See http://www.archaeoleg.org.uk/pdf/roman/KEY SITES SE WALES ROMAN.pdf., Last Date of
Access: 4/17/11
126

8). From these findings, any indication of broken or continual occupation may become clear,
as well as any evidence of social or economic status of the inhabitants. Sites mentioned along
the southern coast are given in geographical order from east to west, and all locations are
considered to be Romano-British in character, unless specific indications of Roman villa
construction are found.

Map 8: Romano-British farmsteads and Roman villas


©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved.

Just off the banks of the Wye river south-east of Chepstow, excavations at Thornwell
Farm provided evidence of prehistoric and Romano-British settlements. 528 Although there are
obvious structural and artefactual changes to the site during the first century AD, a break in
the occupation cannot be ascertained, suggesting that the Roman conquest may have had little
effect on the site. As for the social status of the site, the pottery is of low quality and the
structures are not characteristically Roman. The usual evidence from a Roman villa structure
is also not found: tesserae, painted wall plaster or hypocaust tiles. This suggests that the
inhabitants of the site, whether or not they were the same as those present in the Iron Age,
were of lower social status, and did not rise in wealth through the Roman period. Occupation
of the site seems to continue into the later Romano-British period, again with no apparent
archaeological break. Thornwell Farm has a mid-fourth century end to the settlement activity,
which, as will be seen, is similar to other rural settlements in South Wales.

528
Hughes (1996).
127

In Monmouthshire, Caldicot, one kilometre south-east of Caerwent, may have a start


date in the early first century AD or a little earlier due to Iron Age evidence. The site,
however, may have quickly become abandoned around the time of contact with the
Romans.529 A complete absence of Flavian pottery within the ceramic assemblage suggests a
reoccupation in the early second century AD.530 There is little evidence for occupation after
the mid-fourth century, nor of high-status inhabitants, and the subsistence was almost fully
agricultural. But only a small area was excavated, and further work in the neighbouring area
may provide different results. This site must not be confused with the Caldicot kiln site
excavated in 1965 c. 1.5 km south. 531 The South Wales grey-ware produced there was found
at the presently discussed settlement, but consisted of only 10% of the entire grey-ware
assemblage. Just 500 m south-east of the quarry site is Church Farm, Caldicot: a low-status
Romano-British farmstead also spanning the Late Iron Age to mid-fourth century AD.532 This
site also took advantage of the nearby kilns, as the majority of the ceramic assemblage at
Church Farm is believed to be from there.
On the Gwent Levels, Romano-British settlement evidence has been dated through
pottery of second century to third or fourth century AD, and coins ending c. AD 330.533 Iron
Age pottery was also recovered, although continuation of Iron Age occupation is not
expected. It was in the Roman period that the first efforts were made to drain the Gwent
Levels for agriculture purposes, as shown by the Goldcliff Stone. This is still apparent on the
Wentlooge Level west of the Usk river, while part of the ‘planned’ landscape around
Peterstone is of Roman date. The Caldicot Level, east of the Usk river, is where Roman
occupation was first discovered and documented in three main concentrations: at Magor Pill,
near the mouth of the Usk and by Goldcliff Point. Magor Pill was a small port during the
medieval period, thus a Roman period function may be just as likely. This is also suggested
by the number of possible Roman roads that extend from Iter XIV, the Roman Road from
Caerleon through Caerwent, towards Magor Pill. As well, nearby a Roman period plank-built
boat was found dated to the mid to late third century AD. This more than likely corresponds
with cross-channel transportation like that from Sudbrook 7.5 km to the east.534
Settlement evidence on the Gwent Levels suggests humble agro-pastoral subsistence
of Romano-British character.535 On the Caldicot Level, which is in close proximity to
Caerwent, ceramic and animal bone finds of Roman date are found at the three places
mentioned above, as well as on the coast near Caldicot, Chapel Farm in Undy, west of

529
Vyner and Allen (1988).
530
Boon, G.C. (1988a). ‘The Samian Ware’, in Vyner and Allen (1988), 100.
531
Barnett et al. (1990).
532
Insole (1997), 73.
533
Rippon (1996), 25-35.
534
See above, page 94.
535
Rippon (1996), 29-35.
128

Sudbrook, Rogiet536 and Greenmoor537. The first site excavated on the Wentlooge Level was
found in 1996 at Great Pencarn Farm, south-west of Newport, although Roman period
evidence has also been recorded on a small scale at Great Warf in St. Brides, Peterstone Gout
and Rumney Great Warf.538
Just over three kilometres west of Cardiff and c. 500 m south-east of the Roman road
(Iter XII) to Carmarthen through Cowbridge, sits a Romano-British settlement at Ely with
various markers of a villa: rectilinear stone foundations, hypocaust heating, painted wall
plaster and tesserae.539 A later bath was added to the settlement as an entire wing on the south
end. This site differs from the norm as there are no associated agricultural or pastoral
buildings; instead the occupants of the villa are believed to have become wealthy through iron
working, as an area in the south-east of the complex originally marked ‘Roman Steel Furnace
(site of)’ gave a strong magnetic anomaly during survey in 2001. If this is the case, it should
be assumed that they sold their wares in the nearby vicus of the fort at Cardiff. Construction
of the villa started in the first half of the second century AD, and fortification of the site is
seen around the fourth century, in line with the need for security documented around Britain.
Close proximity of this site to the river Ely may have led the inhabitants to seek protection
from sea-borne invaders rather quickly. This defensive construction may have been a
consolidation of the land at a time when part of the settlement was already in ruin, possibly
due to the economic downturn of the third century. Thus, an early to at most mid-fourth
century date should be presumed for the abandonment of the site.
Llandough sits c. 3.5 km south-west of Cardiff and c. 3.5 km south-east of Ely. 540
Iron Age occupation of the site at or around the Roman conquest is apparent through pottery
and a curvilinear building presumed to be a roundhouse similar to those found at other sites in
the area. Later rectilinear stone buildings with painted wall plaster and a hypocaust heating
system dominate the site leaving no doubt of a villa type structure during the Roman period.
The high status of the inhabitants is furthered by the remains of a bath house within the
foundations of the building. Occupation may have been continuous, but most certainly
Roman influence at the site can be placed from the first half of the second century AD to the
early fourth century. During this time, a major expansion of the villa took place in the early
third century; it is during this construction phase that the bath house was incorporated. Due to
the limited nature of the excavation, the surrounding land could not be fully explored for

536
Marvell (1996a), 78.
537
Locock and Yates (1999), 105.
538
Marvell, A.G. (1996c). ‘Great Pencarn Farm, Coedkernew (ST 283 834) Newport’, Archaeology in
Wales. 36, 75-6; Rippon (1996), 33.
539
Wheeler, R.E.M. (1921). ‘A Roman Fortified House near Cardiff’, The Journal of Roman Studies.
11, 67-85; Young, T. (2001). ‘Ely Roman villa (ST 147 761)’, Archaeology in Wales. 41, 130-2.
540
Owen-John, H.S. (1988b). ‘Llandough: The Rescue Excavation of a Multi-Period Site Near Cardiff,
South Glamorgan’, in Robinson (1988), 123-77.
129

agricultural subsistence, but the bone assemblage suggests the cattle, sheep and pig were
evenly used; while the old age of the bones suggests the livestock were used for their milk,
wool and leather, as can be found on other sites. Because of the extreme high-status of the
site, it needs no assumption that the occupants were of great wealth, although what economic
practices they used to gain the wealth remains unknown.
In the Vale of Glamorgan, Biglis, north-east of Barry, is another example of a
Romano-British farmstead upon an Iron Age settlement. 541 The site also has origins in the
early first century AD, but occupation probably ceased before the end of the century. Re-
occupation in the later second century continued through two phases to the second quarter of
the fourth century. This site has marked differences with higher status settlements nearby
and, although of a similar mixed economy, was of a lower standard of living with none of the
characteristics of a villa. Like other South Wales settlement sites, cattle, sheep and pig are
dominant in the archaeological record, although here sheep were more prominent. Corn
dryers are found at all three phases of occupation showing a consistent agricultural practice,
although due to the relatively humble nature of the site, maybe not for more than personal
use. Six more sites near Barry, including under Barry Castle, have been recorded which show
Romano-British occupation but do not stipulate the wealth of the settlements. 542
Possibly the most widely known site of recent Romano-British rural settlement in
South Wales is Whitton in the modern Vale of Glamorgan. 543 The first phase of settlement
has been attributed to AD 30-55, quite late in the Iron Age, as this date range straddles the
initial period of the Roman invasion of Britain. However, this should not in any way suggest
a Roman origin as the archaeological record shows Roman influence structurally was not seen
until 100 years later. The site started with humble Iron Age origins which grew through the
mixed agro-pastoral farming methods, although the inhabitants may have had a connection
with the military.544 Cattle and sheep were often preferred and kept to an old age suggesting
their use for milk, wool and leather, all of which is noticeable in the bone assemblages. 545
Prosperity was achieved through their farming, as manufactured goods, i.e. pottery, glass and
bronze, were brought to the site, and a Roman high-status character was apparently strived for
as hypocausts were found under sub-rectangular stone buildings with coloured wall plaster.
Occupation at Whitton ceases around AD 340; not by violent means, but most likely through
the drying up of the well.
Whitton is central in the confusion over whether a site is a farmstead or villa. With
evidence of a hypocaust system and coloured wall plaster, preliminary publications led

541
Parkhouse (1988).
542
RCAHMW (1976b), 120: nos. 764-9.
543
Jarrett and Wrathmell (1981).
544
Black (1994), 108-9.
545
Kinnes, I.A. (1971). ‘The Animal Bones’, in Jarrett and Wrathmell (1981), 232-8.
130

RCAHMW to list the site as a villa in its inventory of Glamorgan in 1976, and more recent
sources continue to follow this classification. 546 However, Jarrett and Wrathmell argue that
Whitton does not compare to the villas at Ely or Llantwit Major, and thus was probably not
believed to be of the ‘villa’ class even to the inhabitants.547
At St. Athan, east of Llantwit Major in the Vale of Glamorgan, Iron Age occupation
ceased around the third quarter of the first century AD, but continued again after AD 140 and
activity remained until the middle to late third century. 548 This implies the new occupation
was of a different family, either Roman or Silurian. These new inhabitants remained of low-
status, however, as the quality of pottery and lack of high-wealth goods indicate. The Roman
period occupation was typical of a Romano-British farmstead, but proved the most productive
of all the periods at St. Athan. Subsistence was sustained by a mixed agricultural system
reflected in the associated field patterns which suggest pastoral farming of stock including
cattle, sheep and pig.549
Llantwit Major (Caermead), lies north-west of Llantwit Major and south-west of
Cowbridge, and has been the topic of various excavations and publications debating the
interpretations.550 A pre-Roman settlement in a ditched enclosure is suggested, followed by
an early timber villa and then a simple rectilinear stone building by the mid second century
AD. Economic decline and possibly abandonment in the third century ended with a new
phase of construction c. AD 270. From then, continual construction saw the site expand
gradually but consistently. Over time the inhabitant’s wealth diminished and use of the villa
decreased and were confined to only certain areas, leaving others to fall into ruin until
occupation ceased near the end of the fourth century. The ancillary structures interpreted as
barns and stables suggest a pastoral subsistence. Animal bone assemblages show ox, pig and
sheep to be most common. This may be attached to some small scale agricultural practices,
perhaps hay, but due to the obvious wealth of the occupants, they may have chosen to abstain
from such and purchase their feed. Supplementary to their livestock, evidence was found of
both iron and lead smelting.
Near Merthyr Mawr Warren, west of Bridgend, sits Cae Summerhouse. 551 Only
partial excavation of the settlement could be completed, thus an understanding of the entire
site is not currently available. What could be uncovered, however, suggests three phases of

546
RCAHMW (1976b), 114-7: no. 761; Arnold and Davies (2000), 77; 82-4.
547
Jarrett and Wrathmell (1981), 253.
548
Barber et al. (2007).
549
Vaughan-Williams, A. (2007). ‘Charred Plant Remains’, in Barber et al. (2007), 94-7; Higbee, L.
(2007). ‘Animal Bone’, in Barber et al. (2007), 91-4.
550
Nash-Williams, V.E. (1953). ‘The Roman Villa at Llantwit Major, Glamorgan’, Archaeologia
Cambrensis. 102, 89-163; Hogg, A.H.A. (1974). ‘The Llantwit Major Villa: A Reconsideration of the
Evidence’, Britannia. 5, 225-50.
551
Davies, J.L. (1973). ‘Cae Summerhouse, Tythegston’, Morgannwg. 17, 53-7; RCAHMW (1976b),
57: no. 691.
131

Iron Age occupation, possibly starting in the first century AD, and at least two during the
Roman period. Comparisons with Whitton portray a different settlement history, where the
inhabitants of Cae Summerhouse chose a more humble lifestyle. Both sites did maintain an
agro-pastoral subsistence and built corn-dryers, yet Cae Summerhouse seems to have been on
a much smaller scale: a less extensive timber based building tradition does not evolve to stone
as in Whitton, and the Roman characteristics of high-status life are not present. If excavations
were to continue in the uncultivated land next to the site, more may be uncovered suggesting
a more Romanised existence.
Near the coast, also west of Bridgend, the structure at Dan-y-Graig has been
interpreted alongside Llantwit Major and Whitton. 552 Thus, the site is believed to be of early
Romano-British or native construction and extending into the early fourth century AD. As at
Llantwit Major and Whitton, Roman remains characteristic of a villa settlement site exist,
including mosaic tesserae and painted wall plaster. From the results of the short excavation,
Newman did not believe the site could be classified as a villa, yet a geophysical survey over
the large area immediately west of the excavation site provided numerous villa-like features
which, along with the excavation evidence, highly suggest a villa site. Dan-y-Graig has been
interpreted as an agricultural estate with a T-shaped kiln mostly likely used for corn-drying,
although other later uses cannot be excluded.553
In Gower, a site at Oystermouth has been classed as a villa by the RCAHMW, but
only due to small evidence found during non-archaeological activities. 554 The site is currently
occupied by the All Saints Church where from the seventeenth century onwards fragments of
Roman tesserae and coins have been unearthed during grave digging in the churchyard. This
dearth of evidence cannot by any means confirm the presence of a villa at these sites, however
Roman influence is obvious and elements such as mosaic fragments are characteristic of a
Roman villa settlement. Although not much is known about this site, the evidence presented
shows that Romans did travel to Gower, and possibly further into the peninsula. At Church
Hill, in Penmaen west of Oystermouth, a small enclosure was excavated in 2008 revealing
Roman finds of non-villa occupation. 555 This is substantiated by the ceramic evidence, but
painted wall plaster and possible tesserae raise questions. Cattle and sheep bones, as
expected, are of the majority of identified animal bones, while a large quantity of oyster and
mollusc shells were recovered.556

552
Newman, R. (1990). ‘Excavations of a Romano-British Building at Dan-y-Graig, Porthcawl, Mid
Glamorgan (SS 8407 7805)’, Bulletin of the Board of Celtic Studies. 37, 247-80.
553
Caseldine, A.E. (1990). ‘The Archaeobotanical Evidence from a Roman ‘T’-shaped Kiln at Dan-y-
Graig’, in Newman (1990), 277.
554
RCAHMW (1976b), 110-1: no. 756.
555
Evans, E. (2010c). Church Hill, Penmaen, Swansea. GGAT Report No. 2010/033.
556
Locock, M. (2010). ‘The Animal Bone’, in Evans (2010c), 13-4.
132

In the Usk river valley area of modern Brecon, little evidence of Roman rural
settlement exists. With the prominence of the auxiliary fort at Brecon (Y Gaer) within the
valley, it can be assumed that the territory was under military control until at least the
Hadrianic era, if not later.557 Five miles south-east of Y Gaer lies the Roman villa of
Llanfrynach.558 That which is excavated is the bath-block of the villa, recognizable by the
mosaic pavements, hypocausts and sunken bathrooms. Coin finds cannot give a confidant
start date for the occupation of the site, but abandonment was probably in the last half of the
fourth century. This villa is the only known example of such in the Brecon area, with Roman
activity documented as persisting in the early and late Roman periods; a gap is recorded in the
archaeological record between c. AD 180 and c. AD 330.559 Four iron-smelting and forging
furnaces have been found in the area as well, most likely of first century AD date, and very
likely of military installation.560
Other possible villas have been documented over time from field surveys and small
excavations. In 1956-7, two sites at Moulton and Llanbethery, both in the Vale of
Glamorgan, were discovered to be probably villas due to traces of stone buildings, painted
plaster and hypocaust tiles.561 In a field two kilometres west of Caerwent a settlement site,
Five Lanes, is believed to be a villa due to its rectilinear shape, the presence of a ‘winged
corridor’ and possibly some associated tesserae.562 Aerial photographs of Five Lanes also led
to the discovery in the adjacent field of Oakland Farm of a rectangular building with four
rooms overlying an oval enclosure, possibly showing a Roman villa built on top of an Iron
Age site.563 In 2001, a site was discovered by metal detector enthusiasts near Langsont, east
of Newport.564 Tesserae and painted wall plaster suggest a high status site, although this may
have been in only a small area, and ceramic evidence indicates third and fourth century AD
occupation with a possible earlier activity. Near Newport, air survey in 1996 discovered the
site of Croes Carn Einion, Bassaleg, and subsequent ground survey found fragments of
imbrex and flue-tile, which led to interpretations of the site as a Roman villa. 565 However,
other sites in the area, like Rogiet and Great Pencarn Farm found on the Gwent Levels, also

557
RCAHMW (1986), 179.
558
Nash-Williams, V.E. (1950b). ‘The Roman Villa at Llanfrynach, Brecknockshire’, Bulletin of the
Board of Celtic Studies. 13, 105-8. The site is also known as the Roman Villa at Maesderwen: see
RCAHMW (1986), 179-82: RV 1
559
RCAHMW (1986), 182: RCS 1.
560
RCAHMW (1986), 183: RU 1.
561
Thomas, H. (1958). ‘Roman Discoveries in the Vale of Glamorgan’, Bulletin of the Board of Celtic
Studies. 17, 294; Richmond, I.A. (1958). ‘Roman Britain in 1957: I. Sites Explored’, The Journal of
Roman Studies. 48, 131-2; RCAHMW (1976b), 114: nos. 759-60.
562
Mein, A.G. (1995). ‘Five Lanes, Caerwent (ST4460 9100)’, Archaeology in Wales. 35, 54-5.
563
Burnham, B.C. (2000). ‘Roman Britain in 1999: 1. Wales’, Britannia. 31, 375.
564
Macdonald, P. (2001). ‘Langstone (ST 38 89)’, Archaeology in Wales. 41, 138-9.
565
Burnham (2000), 376.
133

contained masonry and tiles, but little else to suggest a high wealth settlement, therefore
bringing the identification of Croes Carn Einion as such into question. 566
A rural settlement survey of these sites shows consistent connections with the Iron
Age, although often the native occupation is of the Late Iron Age beginning in the first
century AD. At these rural sites, evidence of metalworking, ceramic production and exotic
trade goods are consistent within the Iron Age and Roman periods. A gap in archaeological
evidence around the Roman conquest is apparent and understandable through the cultural
change. However, those living in South Wales persisted in a similar subsistence pattern
during the Roman occupation as that during the Iron Age, only with many new forms of
material goods.
Where the subsistence is recorded, mixed farming is present at the majority of sites,
keeping with the agro-pastoral tradition of the Silures which fits best to the landscape and
environment of South Wales. Settlements are prevalent throughout the southern lowlands
from the Wye river in the east to the Gower. When it comes to the progression of the
residence, the diverging paths of farmstead or villa seem wholly dependent upon the
inhabitants of the site and their prosperity (cf. Cae Summerhouse and Whitton). This could
be due to a number of reasons that cannot be archaeologically determined.
The transition from Iron Age to the Roman period most likely proved to be difficult
for the majority of the Silures. This is most certainly due to the drawn out conflict lasting
over two decades, and, as mentioned, the effects of the war, treatment by the Roman
immigrants and the immediate changes associated with the forced change in lifestyle and
subsistence. In some instances prominent Iron Age sites become abandoned in the first
century AD around the final years of the Roman conquest of South Wales. Lodge Wood, the
largest hillfort in Gwent, was abandoned in the Roman conquest period, and reoccupation was
probably discouraged by the close proximity of Caerleon. 567 Similarly, the hillfort at
Llanmelin, most likely dominant in south Gwent with its connection to Sudbrook, did not last
into the Romano-British period, but was abandoned about AD 75.568 Some Iron Age
settlements ended at the same time, but were reoccupied later in the Romano-British period.
At St. Athan, Iron Age occupation ceased around the third quarter of the first century AD, but
continued again after AD 140.569 Once reoccupied, the inhabitants of St. Athan were again of
modest economic status, and subsisted through agro-pastoral farming. Second to fourth
century AD Roman pottery sherds are found at Castle Ditches (Llancarfan), suggesting
Roman occupation then, but not earlier, although no structures were found of that date during

566
Arnold and Davies (2000), 85.
567
Howell and Pollard (2000), 83.
568
Nash-Williams (1933), 290.
569
Barber et al. (2007), 66.
134

excavations.570 Other native sites remained empty until much later: although Romano-British
evidence exists at Dinas Powys, in the Vale of Glamorgan, it mostly likely came from post-
Roman occupation.571 The conquest period obviously disrupted the Silurian society, and as is
evident from the occupation patterns of sites such as these, it took years before the natives
became accustomed to the new way of life, and could return to a form of normality.
Possible continual occupation of a prominent Iron Age site may be found at Caer
Dynnaf, just south-west of Cowbridge in the Vale of Glamorgan. 572 There Iron Age pottery
dates to the mid-first century AD, while Romano-British pottery suggests occupation from the
late first or early second century AD to the fourth century. Caer Dynnaf is a ‘large’ Iron Age
enclosure of 3.8 ha in size, and may have been allowed to continue occupation after the
conquest. However, since Lodge Wood and Llanmelin both show Iron Age occupation
ending in the final years of the Roman conquest of the area, the small gap between the mid
and late first century AD in the ceramic assemblage may indicate the same. Either way, it has
also been suggested that the Romano-British occupation at Caer Dynnaf may have subsided
slowly due to the close proximity of the Roman road (Iter XII) and thus became opportunistic
by settling in what is now Cowbridge.573
Even through the apparent disruption of the decades of Silurian resistance and the
subsequent conquest, some Iron Age sites show occupation, possibly continuation, in the
Roman period. A study by Branigan of Romano-British villas with pre-conquest origins
brought about four models of settlement evolution, each characteristic of different paths taken
by the Romanised occupants.574 These models are only on the basis of continual occupation,
and examples of each are given by Branigan in his text, however discussion herein will be
solely on South Wales. The first model is of a high-status native settlement which Romanised
quickly following the invasion and then within two generations afterward built a villa. Model
two follows a pre-conquest farm slowly progressing to a wealthy villa settlement through new
agricultural practices. Branigan’s third model is of a post-conquest native farmstead rapidly
becoming a Romanised villa by the mid-second century AD at the latest. The last model has
the farmstead beginning after the conquest, but like model two, slowly progresses to a villa
status settlement.

570
Alcock, L. (1956). ‘Roman Sherds from Castle Ditches, Llancarfan (Glam.)’, Bulletin of the Board
of Celtic Studies. 16, 299-300; Hogg, A.H.A. (1976). ‘Castle Ditches, Llancarfan, Glamorgan’,
Archaeologia Cambrensis. 125, 28.
571
Alcock, L. (1963). Dinas Powys: An Iron Age, Dark Age and Early Medieval Settlement in
Glamorgan. Cardiff: University of Wales Press, 22-5.
572
Davies, J.L. (1967). ‘Caer Dynnaf, Llanblethian (SS 984744)’, Archaeology in Wales. 7, 7.
573
Robinson (1980), 26: n. 61; Arnold and Davies (2000), 88.
574
Branigan, K. (1982). ‘Celtic Farm to Roman Villa’, in Miles, D. (ed.) (1982a). The Romano-British
Countryside: Studies in Rural Settlement and Economy. Part I. Oxford: British Archaeological Reports
British Series 103(I), 81-96.
135

Evidence for a wealthy native elite class in South Wales is in the larger hillforts, as
stressed in the Iron Age chapters of this thesis. However, Lodge Wood and Llanmelin, the
two most promising sites for inclusion in the first model group, show abandonment around
the final conquest years in South Wales c. AD 75. This suggests that the Roman Period began
with severe changes to the social system of the Silures which did not allow the larger hillforts
to maintain their role within the landscape.
Whitton and Llantwit Major may fit well as model two sites with their early first
century AD native foundation and subsequent Roman construction in the first half of the
second century. However, characterising both sites as a villa by the first half of the second
century is highly interpretive, and each settlement may in fact not be considered of villa type
until the later third century AD. If this is so, then both sites may be considered characteristic
of model four sites, explained below.
At first glance, no villas in South Wales with good dating evidence fit into the third
model, post-conquest grouping. However if dating evidence is off by only a few decades in
any of the archaeological records mentioned above, then later start dates for the native
farmsteads could show exactly this rapid rise in socio-economic status. If this is the case, Ely
and possibly Llandough and Llantwit Major are prime examples.
Whitton may fit as a model four site depending upon the interpretation of the villa
structures.575 Although a rectilinear stone building appears sometime in the mid-second
century, the hypocausts may not have been built until the end of the second century or the
beginning of the third. Either way, they were abandoned shortly after, possibly unused. If
this chronology does not suggest a villa settlement at this time, and yet the site overall is
considered as a villa, it must be placed within the late third to early fourth centuries AD;
making the site an example of the fourth model. Llantwit Major may fit into this category as
well, only with the belief that the presumed abandonment was actually only a period of lower
activity, and that the sight was of continual occupancy.
Obviously there are variables existing within this theoretical framework, and
Branigan’s system of models mostly pertains to villas found within the English countryside.
Yet this type of systematic categorization of sites proved useful in bringing in overall views
of the entire rural landscape. When focused upon South Wales, examples mentioned focus
upon first century AD foundations and quick or gradual development to villa status. Ely and
Llandough seem to benefit by their close proximity to Cardiff fort and its vicus by expanding
to villas rather rapidly, and so too may have Whitton and Llantwit Major. All these sites
depend upon interpretation of the building chronology and definitions of a ‘villa’. Dan-y-
Graig and Llanfrynach, the only definite villa settlements not mentioned, are not excavated

575
Jarrett and Wrathmell (1981), 82-101.
136

well enough to deduce the entire settlement chronology, while those which possibly are villas
need more excavation work to confirm them as such.
Regionally, evidence for villa style construction along the South Wales coast is only
apparent west of Cardiff. Since settlements such as Whitton and Llantwit Major became
occupied relatively late, just before the conquest period, this suggests a somewhat new
population in the area part of which may have been more susceptible to cultural change.
Conversely, the native population in southern Gwent may have been more resistant to the
same change, deciding to keep a more traditional Siluran lifestyle as suggested by the
archaeological evidence.
Possibly the only definitive aspect of South Wales rural settlement, that can be seen
from the documentation and interpretation of the sites listed above, may be that each location
must be seen independently. In that case such a study can become quite superfluous.
However, ideas should be gathered and attempts should be made to classify the entire South
Wales rural settlement in the Roman period, if not only to prove that the individual evolution
of each site is itself the pattern. A similar situation has been stressed as having persisted in
the Iron Age through the varying settlement patterns across South Wales with regard to size,
shape, defences and locations. Since many variables were in place, from differing cultural
and economic backgrounds to desire for prosperity, what remained consistent were the vast
differences in the settlement patterns and structures before and after the Roman conquest.

The Rural Economy


Although the urban lifestyles within the civitas centre and that surrounding the
military garrisons of South Wales best exemplified the Roman impact upon Britain, the
majority of Romano-British native existence was in the surrounding countryside. 576
Furthermore, the relationship between a town (not including those surrounding military
establishments) and the local rural community has often been described as ‘parasitic’ where
the urban centres are consistently consuming the raw materials provided by the rural
population only to give little back in return. 577 This relationship can become more demanding
as settlements grow over time or are placed within a less productive countryside. For
example, at Calleva Atrebatum (Silchester), the surrounding land was of poor quality,
consisting of gravels and clay. 578 Only a pastoral subsistence could exist there, and thus the
community needed to extend its control to a much larger area to be able to meet its
agricultural requirements.

576
Miles, D. (1989). ‘The Romano-British Countryside’, in Todd (1989), 115.
577
Fulford, M. (1982). ‘Town and Country in Roman Britain – A Parasitical Relationship?’, in Miles,
D. (ed.) (1982b). The Romano-British Countryside: Studies in Rural Settlement and Economy. Part II.
Oxford: British Archaeological Reports British Series 103(II), 403.
578
Fulford (1982), 406-7.
137

Fortunately for those living in South Wales, the agro-pastoral subsistence had existed
for centuries before the Roman conquest. As mentioned above, the territorium of Caerleon
may have persisted to separate the eastern land of the Silures, including Venta Silurum, with
the western tribal territory. As is seen through the archaeological evidence, Caerwent
prospered well throughout the Roman Period, even building a few houses that continued the
rural subsistence within the town walls in the later centuries. Cowbridge was a smaller
settlement and seemingly had a much larger surrounding area from which to provide itself
with resources, as compared with Venta Silurum. Conversely, it can be presumed that
although there is evidence of some products in use which originated outside the tribal
territory, such as stamped mortaria579, much of the economy must have relied upon locally
manufactured goods, i.e. South Wales grey-ware, as it would have been more cost-effective.
Thus, the relationship between the urban community and the surrounding rural countryside in
South Wales may not have been entirely symbiotic, but there is little evidence to suggest one
which was completely ‘parasitic’. In fact, with the reliance of the town upon the resources of
the country, and the necessity of rural land ownership to participate in the civitas politics,
town life may have been subordinate to the rural settlements, at least in Roman Britain. 580
The agricultural impact of the Roman invasion on the Silures may have been almost
immediate, as the sudden increase in population due to the presence of the military would
have required more locally grown resources. 581 With the introduction of Roman tools,
farming practices, new crops and wetland reclamation efforts, increasing the subsistence
beyond that necessary to support the Iron Age populations must have been an immediate task
for the Roman army as it entered newly conquered land. Thus, it must have been that only a
short time after the arrival of the Romans, the landscape took upon a vastly different character
from that of the Iron Age. Part of this agricultural evolution can be found in the increase in
querns and corn-driers found in South Wales, including at Biglis, Cae Summerhouse, Dan-y-
Graig and Whitton. As well, with pastoral farming, locally grown hay can be presumed as a
major crop for feeding, and should be expected at most mixed farming sites.
However, as Millet stresses, archaeological evidence shows the agricultural evolution
of Late Iron Age societies did not correspond precisely with the Roman invasion, and instead
were an indirect consequence of such.582 Cooperation with the natives has already been
stressed as the most favourable route for a seamless, or at least less bloody, transition, and a
similar relationship should be expected with the agricultural regime. Yet, the Roman army as

579
Fulford (1982), 411-3. Seager Smith (2000), 277-83.
580
Perring, D. (2002). The Roman House in Britain. London: Routledge, 218.
581
Jones (1984), 101-20; Jones, M. (1989). ‘Agriculture in Roman Britain: The Dynamics of Change’,
in Todd (1989a), 127-34; Dark and Dark (1997), 93-113; Allason-Jones (2008), 81-6.
582
Millet (1990), 97-101.
138

a cultural entity and a drain on resources provided the great ‘push’ the Iron Age cultures
needed towards the subsequent technological advances.
Pastoral subsistence was always present in Silurian farmland, and the use of livestock
did not change into the Roman Period. Settlements such as Whitton and Cae Summerhouse
demonstrate strong pastoral elements as cattle and sheep are highly represented in the animal
bone assemblages. At most rural sites, cattle were the major food source with pig and sheep
secondary, but the older age of the cattle and sheep also indicates use for wool, milk and
leather. Keeping animals for long periods of time and during harsh weather necessitates
plenty of resources and accommodation for the livestock. 583 At Whitton, some iron tools
suggest leather making, but there is no evidence of tanning practices, which may have been
done elsewhere like Caerwent, but more likely Cowbridge or Cardiff. In the Gwent Levels,
cattle and sheep are prominent but horse bones also suggest a horse-rearing practice, probably
associated with the military as they organized the wetland reclamation. 584 Furthermore, a
continuation of the transhumant pastoral nature of the Iron Age may be reflected in Roman
finds recovered from upland locations.585
A predominantly agro-pastoral existence presumes a large amount of land associated
with a villa or farmstead. With the large amount of lowland arable land in modern
Monmouthshire, the Roman settlements found within, particularly the villas, could indicate
substantial agricultural exploitation involving large numbers of people. 586 Various situations
could arise from a large estate: including renting land-use to others, joint ownership, or
multiple villas, suggesting a large, wealthy family. 587 Profit from these ventures may have
come in monetary form or by goods exchange; the latter of which is possibly seen in the
presence of large barns at villa sites. As the military often taxed the locals in provisions for
the garrisons, so could have villa owners, possibly even for later use as payment to the
military in the same produce.
Todd has suggested that some smaller rural townships in Britain may have origins as
part of a large villa estate.588 As more workers are employed by the landowner, a settlement
may arise to house the entire workforce; which, after the end of the private ownership, then
continues as an independent community. This was first suggested for Roman Britain, but has
since been proposed for continental communities.
It has been shown that villas, and by association farmsteads, persisted in mixed
economies.589 Although discussed herein through the different economic ventures, the sites

583
Applebaum, S. (1966). ‘Peasant Economy and Types of Agriculture’, in Thomas (1966), 101.
584
Rippon (1996), 29.
585
Arnold and Davies (2000), 76-7.
586
Davies (1979), 154.
587
Dark and Dark (1997), 72-4.
588
Todd (1989b), 17-9.
589
Branigan, K. (1989). ‘Specialisation in Villa Economies’, in Branigan and Miles (1989), 42-50.
139

mentioned above included other agricultural and non-agricultural activities for personal and
local use; fishing, metalworking and pottery can all be attributed to rural settlements. 590
Shellfish are represented more strongly in coastal locations, such as Llantwit Major and
Whitton, however, there current evidence only points to sea-borne produce as an occasional
personal luxury.591
There was a constant demand for metalworking during the Roman period, and an
apparent wealth through iron working may be found at the villas at Ely and Llantwit Major
which presumably profited from the nearby Cardiff vicus and Cowbridge settlement,
respectively. Iron working may have continued near the remains of the fort at Caergwanaf
into the fourth century, while iron slag at Whitton has been most probably connected to the
Taffs Well-Llanharry orefield in the county borough of Rhondda Cynon Taff. 592 Caergwanaf
quite possibly forged iron from this local area as well. A Roman date is associated with iron
slag on the Wentlooge Level at Rumney Great Warf, although the quantity is of a smaller
scale.593 Similarly, lead working is thought to be a major part of Llantwit Major as well as a
Romano-British site in Wenvoe, south-east of Ely.594
Ceramic productions in South Wales have been shown to be predominately rural
based.595 The South Wales grey-ware tradition, as mentioned above, has been documented as
a product of rural based kilns in Caldicot, Llanedeyrn and Llwyn-y-grant, Penylan. Caerleon
ware, produced particularly for Legio II Augusta while garrisoned at Caerleon, has been found
produced at Kemeys Inferior near Newport, only two kilometres west of the fortress. 596
Archaeological evidence has so far pointed to production of both pottery traditions in eastern
South Wales, yet distribution has extended much further. Caerleon ware, although of a small
local production and use, and South Wales grey-ware has been found as far away as Church
hill on Gower.597 Thus the production at rural settlements was then taken to nearby market
centres, rather than the goods being produced within the urban area. The wide distribution
pattern, even if on a small scale, gives evidence for the relationship between the urban and
rural landscapes as they rely upon each other for provision of resources and economic
income. As well, this local production and long-distance trade of South Wales grey-ware
may have origins in the Lydney-Llanmelin ware production in the Iron Age, but with a wider

590
Miles (1989), 123; Todd (1989b), 19-20; Arnold and Davies (2000), 94.
591
Matheson, C. and Cowley, L.F. (1953). ‘Report on Stratified Human and Animal Remains’, in
Nash-Williams (1953), 160-3; Jarrett and Wrathmell (1981), 251-2.
592
Surman, J. (1981). ‘XI Metalwork: 5. Iron Ore and Slag’, in Jarrett and Wrathmell (1981), 208;
Young, T. and Macdonald, P. (1997). ‘Miskin, School Road (ST 0520 8144)’, Archaeology in Wales.
37, 79.
593
Rippon (1996), 29.
594
Vyner, B.E. and White, J. (1985). ‘Romano-British Fields and Farms in the Vale of Glamorgan’,
Britannia. 16, 246.
595
See above, page 114.
596
Marvell (1996b), 74.
597
Webster, P. (2010). ‘The Roman Pottery’, in Evans (2010c), 7-9.
140

distribution due to the better trade connections through the Roman urban and rural
communities.
The Gwent Levels may have also been a source for sea-borne trade. Dorset Black
Burnished ware was found in a great quantity at Rumney Great Warf on the Wentlooge Level,
suggesting cross-channel trade.598 This may have been through a larger vicus such as that in
Cardiff, but the relative lack of local pottery in the Levels assemblage suggests a more direct
interaction. A similar situation is expected of the trade in South Wales grey-ware. Of the
local ceramic evidence, a connection with the kiln at Llanedeyrn is also expected, as well as
similarly decorated ware as found in the Vale of Glamorgan, rather than towards the Gwent
area, suggesting links to the western settlements. This may be another reaction to the possible
separation caused by the Caerleon territorium. Magor Pill has been recommended as a cross-
channel trading port with the presence of a third century AD boat found nearby, and Roman
roads leading from the area to Iter XIV and Caerwent.
Numismatic evidence of rural settlement shows a similar mid-fourth century AD end
599
date. The continuation of Roman coinage in urban settings to a late fourth century date,
however, furthers more the case for rural abandonment in favour of urban protection in this
period. Iron Age coins all but cease in the Roman period as seen at Biglis and Llandough,
while a Dobunnic silver coin was found at Caldicot, although not during excavation and thus
cannot be put into any context.600 Thus the rural settlements of South Wales also used and
maintained the Roman currency, as expected.
Polden Hill and T-shaped brooches, like those found at Cowbridge (Illus. 7 & 8) are
also found elsewhere in South Wales. Polden Hill brooches date to the late first to second
century AD, and are popular in South Wales; found at Biglis, Cae Summerhouse, Caerleon,
Caerwent, Sudbrook, Usk and Whitton.601 At the last site, four examples of the Polden Hill
type have been interpreted as being from a local producer, and similar finds at Cae
Summerhouse, Caerwent and Usk may have been from the same metal smith as well. T-
shaped brooches are more commonly first century in origin, and are also present in sites
around South Wales: Biglis, Caerleon, Caerwent, Usk and Whitton. A La Tène III brooch of
Late Iron Age date was found amongst Roman objects, including a T-shaped brooch, at
Caldicot, which may suggest a continual occupation of the farmstead from the Iron Age and

598
Rippon (1996), 29-30.
599
Boon, G.C. (1988b). ‘Note on Late Roman Coinage in the Vale of Glamorgan’, in Parkhouse
(1988), 52.
600
Boon, G.C. (1988c). ‘The Coins’, in Parkhouse (1988), 51-2; Boon, G.C. (1988d). ‘The Coins’, in
Vyner and Allen (1988), 91; Boon, G.C. (1988e). ‘The Coins’, in Owen-John (1988), 158-9.
601
Webster, J. (1981). ‘XI Metalwork: 1. The Bronzes’, in Jarrett and Wrathmell (1981), 163-9;
Webster, J. (1988). ‘Bronze and Silver Objects’, in Parkhouse (1988), 53.
141

through the Roman period.602 Overall, the brooch evidence also suggests another aspect of
Silurian life that was saturated with Roman influence, for obvious reasons.
A bronze ring with a yellow glass intaglio was found at Whitton dating from the third
century AD.603 This intaglio of a man with outspread arms has been compared to solely
British examples and mostly likely was worn by members of the lower class contrary to that
found at Loughor of military origin (Illus. 9). However, a ring of military origin depicting a
pastoral scene (Illus. 10) provides a close connection to the rural landscape. Since very often
soldiers in the military are of rural backgrounds themselves, many choose to remain in the
territory at which they were stationed and continue their agricultural heritage. Perhaps the
soldier that lost this ring settled near the Loughor river upon retirement.
The presence of coins, brooches and intaglio rings like that found in the military and
urban settings of South Wales show not only the close relationship of the rural environment
with the urban, but also the exchange system in place that allowed for the native population to
provide for themselves similar signs of wealth and high status that those of Roman citizenship
could afford.
When looking at villas and farmsteads separately, some questions can arise as to why
a site developed one way or the other. A comparison between the farmstead at Cae
Summerhouse and the villa at Whitton has already briefly been made, which illustrates how
two sites with similar industrial tendencies can show such ranging socio-economic
differences. Rivet stresses that as the Roman occupation changed the British economy
drastically, the development of the villa ‘was but one manifestation of this economy and it
cannot profitably be considered in isolation from it.’ 604 Furthermore, the Roman villa, in
contrast to the native farmstead, within the Romano-British landscape was dependent on the
town for its social and economic existence and prosperity. Thus, this could have been the
deciding factor in the evolution of a settlement towards a high or low status: the desire, or
opportunity, of a rural landowner to connect with and depend upon the local civitas centre for
continual prosperity. Opportunity may have been directly related to political connections,
which is discussed below, while the personal desire may have coincided with the willingness
to Romanize and accept this new dominating culture. If this is the case, no amount of
archaeological work may be able to prove or dismiss this.
Rural settlements were an important part of the Romano-British economy. While the
urban centres provided Roman culture and ways of governing the locals, the rural economy
provided much of the resources and overall economic income of the province providing merit
to the theory of the ‘parasitic’ relationship between the two. However, in the case of South
Wales this may not have been as problematic through the prevalence of local ceramic
602
Boon, G.C. (1988f). ‘Miscellaneous Finds’, in Vyner and Allen (1988), 93.
603
Webster (1981), 183.
604
Rivet (1969), 215.
142

products. As well, the number of Roman villas present in the area suggests that a prosperous
lifestyle was to be found, and the reciprocating dependence of the civitas centre and the villa
allowed for a mutual relationship. In the third century, some chose to provide protection for
themselves or to rely upon the town, yet the subsistence and relationship remained.

Rural Political Connections


Logically, the most important aspect of rural settlement in the new Romano-British
political system would have been the prestige of landowning. As the ordo (council) in the
civitas centre consisted of the elite natives (and presumably wealthy Roman immigrants) of a
high enough wealth, most likely in the form of land, the necessity to hold land would have
been significant for those wishing to gain and maintain the local political power and clout.
Kinship has been discussed in its possible effects upon the Silures during the Iron Age and
their social and political structures. 605 How these familial ties relate to pre-Roman settlement
is still unclear, and is most likely different across the different tribes of Britain, yet some sort
of kin-based system probably existed in South Wales and persisted into the Roman period.
Thus, the acquiring and passing of land through familial and kinship ties would have also
been an aspect of local social, and by extension political, importance.
At the end of the tenth century AD, a code of Welsh Laws was codified by Hywel
Dda, and work in the past has been focused upon using these laws to try and determine native
Celtic Law of the Roman period in Wales.606 In what has been interpreted as land tenure, tir
cyfrif forced land ownership to be held in common with other members of the nearby
population, where the amount of common land tenure would change with the death or
maturity of members of the community. Upon the death of a landowner, the land would be
divided among immediate and even secondary family, easily leading over time to common
land ownership scattered amongst various areas. It is quite possible that this later code of
Welsh Laws could have origins in Iron Age native (Celtic) law and Roman laws that were
integrated during the Roman occupation.
Possible Roman South Wales kinship connections may be found at Llantwit Major
and Whitton.607 At the former, two houses separate the family groups, although over time
they tend to close in a central common courtyard. At Whitton, Smith interprets the gradual
Romanization and house groupings at the site as evidence that the settlement produced a
hierarchical establishment. Although there are arguments for this line of reasoning, this may
deserve further study, but even with more intensely specific excavations on the sites, nothing

605
See above, page 43.
606
Stevens (1966), 108-11.
607
Smith, J.T. (1978). ‘Villas as a Key to Social Structure’, in Todd, M. (ed.) (1978). Studies in the
Romano-British Villa. Leicester: Leicester University Press, 159: Fig. 52; Smith (1997), 238; Arnold
and Davies (2000), 43.
143

may be proved definitively.608 Yet speculation upon such an archaeologically absent


influence upon the site could prove useful in research on other sites or South Wales as a
whole. An Iron Age based kinship system present at a Romano-British farmstead or villa
may also have been influential upon the development of the site, as political connections to
Venta Silurum or other urban areas may have provided the desire to gain and maintain such
the luxuries of high-status settlement sites. On the other hand, as Smith implies, social and
demographic factors may have proved a hindrance to the settlement by restricting growth,
possibly through the incorporation of native and Roman culture.
A kinship hereditary system may have persisted into the Roman period, and even
been exploited by the wealthy Roman landowners (domini).609 A colonate agriculture system
developed in parts of the Roman Empire that consisted of domini exploiting tenants (coloni)
instead of slaves to cultivate their large landholdings. This arrangement over time became
somewhat hereditary, and such a rural agricultural system would have fit well within the
British provinces, such as South Wales, which were newly conquered, and thus prime for an
influx of wealthy Roman immigrants. A dominus arriving within South Wales soon after the
conquest could have gained access to vast amounts of land, and then utilised Silurian coloni,
incorporating the native kinship ties to maintain familial connections over the land.
By the third century, native law may have proven to be a source of irritation for tax
collectors. According to Dio, Caracalla’s edict in AD 212 extended Roman citizenship to all
free males for the purpose of increasing revenue from estate taxes. 610 If clan or kinship based
systems of sharing common land were in place within South Wales during the third century,
then this would have created problems allotting the duty of which to tax. Moreover, Roman
law allowed for certain joint ownerships, and with the integration of native law in Britain, this
must have been a common practice in areas where tribal kinship connections were prevalent
in the Iron Age and still relevant in the Roman period. Therefore politically, rural settlement
had a strong impact not only on the opportunities for natives to gain wealthy aristocratic
positions within the local Roman provincial government, but also in land holding and
inheritance practices.

The End of the Roman Rural Settlements


As the economy of Roman Britain was closely aligned with the relationship between
the military and urban and rural subsistence, no aspect of society was immune to the
continuing radical events during the third and fourth centuries. A presence or absence of
military garrisons greatly affects urban centres, and in turn rural production; thus the gradual
608
Perring (2002), 202.
609
Millet (1990), 203.
610
Dio 77, 9, 4-5; Perring (2002), 206. Dio’s cynical view of the edict may have been due to a bias
against Caracalla: see Keresztes, P. (1970). ‘The Constitutio Antoniniana and the Persecutions under
Caracalla’, The American Journal of Philology. 91, 449-50.
144

withdrawal of troops over time slowly led to the civilian population becoming self-sufficient
in all aspects, and eventually the rural settlement pattern reflected the contemporary climate.
With the decrease in Roman military garrisons, and the ever increasing sea-borne
threat from outsiders, South Wales was becoming ever more insecure for rural settlements.
Agro-pastoral building within the civitas centre of Caerwent in the fourth century AD
indicates a desire for some to move within the safety of the walls, which coincides with
continual archaeological evidence of Romano-British rural settlement abandonment. 611 This
can be seen in the abandonment of most South Wales rural sites in the mid-fourth century.
Additionally, excavations on the Gwent Levels indicate flooding and a return to tidal
conditions beginning around the end of the fourth century, also suggesting an abandonment of
the area.612 Contrary to the trend is Dinas Powys which shows a post-Roman settlement
reuse of the Iron Age hillfort, yet this only reinforces the apparent desire for protection. 613

611
Brewer (2004), 230-1.
612
Rippon (1996), 35.
613
Alcock (1963), 26-34.
145

Chapter 12: Silurian South Wales Conclusion


The purpose of this dissertation is to demonstrate the initial settlement, economic and
socio-political structures of Silures in the Iron Age, and the subsequent evolution and
adaptation of the native population in the Roman period. Because of the current state of
research and the available archaeological evidence on Iron Age South Wales, much of what
was presented on the Iron Age Silures was supported by interpretations of the archaeology.
In this, a society was presented with multiple local groups of settlements focused upon larger
defended hillforts which they depended upon for certain needs. This decentralised societal
structure was shown in the settlement pattern of South Wales, with the prevalence of smaller
enclosures and the placement of larger hillforts amongst the smaller sites. Interpretations
from Iron Age cultures around Britain then were suggested towards the Silurian subsistence,
yet no definitive conclusions could be reached. Overall, what can be seen about the native
inhabitants of South Wales in the Iron Age is the apparent cultural connections that link the
communities on the western side of the Wye river, the Brecon area of the Usk river valley,
and across lowland South Wales to the Gower peninsula and the Loughor estuary. Evidence
of brooches and military and equestrian artefacts within this area indicate a wealthy elite class
maintaining an aristocratic lifestyle. Suggestions have been made that they perpetuated their
social status by providing those around them with trade goods from outside of the area via
control of the local exchange network. Thus, at the time of the Roman invasion in the first
century AD, the Silures were decentrally structured, yet preserved strong cultural bonds
through the exchange networks which were maintained by the native aristocracy. The lengthy
invasion by the Romans in South Wales is mostly due to the decentralised political structure,
yet a cultural bond across the region could just as well have played a part.
Archaeological evidence of life during the Roman Period suggests that the Silures,
once pacified by the military, took well to the new Romano-British lifestyle. The last
occupation of the larger Iron Age sites at Lodge Wood and Llanmelin in the first century AD
indicates strongly that a drastic change in settlement practices had occurred from hillforts
towards smaller, rural farmsteads. If the establishment of Venta Silurum can be taken as
evidence that the local population (both native and Roman) was capable of governing
themselves, then the change from warring tribe to passive natives within a half century should
be seen as relatively quick. This is also shown in the rapid abandonment of forts within South
Wales as they became unnecessary in the second century AD. Within the Romano-British
landscape, the Silures, most likely the native Iron Age elite, participated in urban and rural
settlements following the new Roman political responsibilities and economic system. This
proved successful for those within Caerwent and the wealthy rural villa estates through the
146

third century. Although the occupants of the high status Romano-British settlements cannot
be decidedly determined as native or Roman, it should not be unexpected that some members
of the Silures became wealthy and advanced themselves and their families within the local
socio-economic classes. They most likely did this through production methods that may not
have been much different from that which they sustained within the Iron Age: agro-
pastoralism, metalworking and ceramics. The prevalence of South Wales grey-ware pottery
within the local ceramic assemblages, and the common metalworking production centres
found in urban and rural locations are only two apparent avenues by which the Silures
embraced the new Roman culture and benefited from it. It is also expected that many natives
tried to maintain traditional lifestyles akin to the Iron Age Silures, which can be seen in the
lower-status rural settlements that still persisted in Iron Age occupation methods.
Nevertheless their living conditions had most certainly been affected by the new Roman
influences, as seen in the new agricultural methods and cultural assemblages which evolved
after the invasion.
147

Appendix 1
Updated List of Defended Enclosures in South Wales614
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
450m SW of Llandewi 0.10 Level - Uni. SS45578876
Church
Allt-yr-Esgair 5.45 Promontory Circular Multi. SO12612435
Allt-yr-Esgair (W) 0.09 Hillslope Circular Uni. SO12302422
Afan Forest - Hillslope Circular - SS81419485
Afon Tarell (Waenfaeog) 0.75 Inland Promontory - Multi. SN99302463
Backingstone Farm - - - Uni. SS57498796
Bank NW of Cwm Bach - Cliff - Uni. SS89587197
Berry Wood 0.40 Hillslope Circular Bi. SS47238847
Bishop Barnet’s Wood 0.08 Hillslope Sub-Rect. Uni. ST51969419
Camp
Bishopston - Hillslope - - SS574887
Bishopston Valley Fort 0.10 Ridge Irregular Bi. SS56938780
Blackfield Wood Camp 0.46 Inland Promontory Triangular Bi. ST52909899
Blaen Cwm Bach 0.10 Ridge Circular Uni. SS79349882
Blaencamlais 0.03 Hillslope Circular Uni. SN95242610
Blue Pool Bay 0.07 Coastal Cliff Sub-Circular Uni. SS40789288
Boverton Road South - - Circular - SS97506852
(The) Briars Enclosure - Level - Uni. ST49359230
(Chepstow)
Broadway Farm (Laleston) - - Horseshoe - SS87678017
Brombil Farm (Margam) 8 - - Circular - SS78168873
Bryn-y-Gaer - Hillslope - Uni. SO194130
Bryn Melyn - Ridge - - SO10881840
Bryn Sil 0.20 Hillslope Sub-Circular Uni. SS48508736
Buarth-y-Gaer 1.10 Hilltop Sub-Circular Uni. SS76559360
Buckholt Wood - Summit Sub-Rect. Multi. SO502159
(The) Bulwark (Llanmadoc 0.90 Ridge Sub-Circular Multi. SS44309275
Hill)
(The) Bulwarks 2.06 Estuarine Cliff Sub-Circular Bi. ST53799273
Burry Holms Hillfort 1.20 Coastal Irregular Uni. SS39889258
Promontory
Cae-Camp 1.59 Ridge Sub-Circular Bi. ST35909385
Cae Summerhouse Camp 0.20 Hilltop Sub-Rect. Multi.. SS86397798
Caeau 0.13 Hillslope Circular Uni. SO16462738
Caemaen Farm - - Circular - ST04867255
Caemarchog - Hillslope Circular - SO22963566
Caer Blaen-y-Cwm 0.10 Spur Sub-Rect. Uni. SS83338807
Caer Cwmphilip (Moel Ton 0.40 Level Polygonal Bi.. SS82558702
Mawr)
Caer Dynnaf 3.80 Ridge Sub-Rect. Bi.. SS98357427
Caer Lloi - - Rectangular Uni. SS87307760

614
Wiggins (2006), Gerrard et al (2006), RCAHMW (1986) and Silvester (2007).
148

Caerau (Llanhennock) 0.70 Hilltop Sub-Circular Uni. ST32989344


Caerau (Llantrisant) 3.60 Hilltop Circular Bi. ST06458320
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Caerau Camp (Ely) 5.10 Ridge Triangular Multi. ST13357500
Caerhendy Enclosure - Summit Triangular Uni. SS78069120
Caerphilly (Gwern-y- - Hillslope Sub-Circular Uni. ST17158680
Domen)
Caldicot Castle West - Spur Sub-Circular Uni. ST48538856
Camp 530m East of 0.30 Spur Sub-Circular Multi. SS83098617
Tonmawr
Camp Hill 0.33 Spur Circular Uni. SO39130786
Campston Hill - Hilltop Sub-Rect. Uni. SO35602240
Carn Caca 0.10 Ridge Sub-Circular Uni. SN83850004
Carn Goch Common I - - Circular Uni. SS60829767
Carn Goch Common II - - Circular Uni. SS61059755
Carn Ifor - Hillslope Triangular Uni. SO16501230
Carn Nicholas 0.60 Hillslope Circular Uni. SS67559434
Carno Reservior - Hillslope Circular Uni. SO166127
Castell Dinas 1.03 Promontory Circular Multi. SO17913009
Castell Moel 0.80 Hilltop Sub-Rect. Uni. ST054734
Castell Morlais 0.80 Ridge Rectangular Uni. SO05000950
Castell Prin 0.22 Hilltop Sub-Rect. Bi. ST40989239
Castle Ditches (Llancarfan) 4.20 Ridge Sub-Rect. Uni. ST05907002
Castle Ditches (Llantwit 2.50 Coastal Irregular Bi. SS96026742
Major) Promontory
Castle Field Camp (Graig- 0.40 Ridge Sub-Rect. Bi. ST20458403
Llwyn)
Castleton North - - Circular - ST02196868
Castleton Wood - Hilltop Circular - ST02606815
Cefn Cilsanws 0.16 Ridge Sub-Rect. Uni. SO02830998
Cefn Enclosure - Spur Sub-Rect. - SO16973291
Cefn Goleu Farm - - Circular - SS57489514
Cefn-yr-Argoed 0.10 Hillslope Sub-Circular Uni. SS83029405
Cil Ifor Top 2.90 Ridge Sub-Rect. Multi. SS50559240
Cil Onen 0.20 Hillslope Sub-Circular Uni. SS54639379
Circular Enclosure - - Circular - SS96266955
(Llantwit Major)
Clawdd-y-Mynach - - Sub-Circular Uni. SS91357115
Cliff House Cropmark II - Hillslope Sub-Circular - ST04906908
Coed-y-Brenhin 0.20 Hillslope Sub-Rect. Uni. SO06592396
Coed-y-Bwnydd Camp 6.39 Hilltop Sub-Circular Multi. SO365068
Coed-y-Caerau (Langstone) 0.40 Ridge Sub-Circular Bi. ST37759145
Coed-y-Caerau 0.86 Hillslope Sub- Uni. SO06922400
(Llanfrynach) Triangular
Coed-y-Caerau Roman Fort 1.00 Ridge Sub-Rect. Bi. ST37959163
(Langstone)
Coed-y-Cymdda 0.60 Ridge Sub-Rect. Uni. ST13227397
Coed-y-Fon (Tredunnoc) - Saddle Rectangular Uni. ST37159471
Coed-y-Gaer (Homestead) 0.30 Hilltop Sub-Circular Uni. SS93918139
Coed-y-Gaer (Llanfihangel) 0.57 Hillslope Triangular Uni. SO17622402
Coed-y-Polyn - Hilltop Sub-Rect. Multi. SO18063845
149

Coed-y-Prior - Hilltop Sub-Rect. Uni. ST39259797


Coed Cefn (Llangenny) 0.29 Hilltop Sub-Rect. Uni. SO22811859
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Coed Duon - Hillslope Sub-Rect. Uni. SO35810081
Coed Fennifach Camp 0.88 Hilltop Sub-Circular Uni. SO01402945
Coed Gaer 0.45 Spur Rectangular Uni. SN98733294
Coed Mawr 0.33 Hillslope Sub-Circular Uni. SO04102512
Coed Pen-Twyn 0.89 Ridge Circular Bi. SO19351622
Coedcae’r Ychain (Vaynor) 0.18 Ridge Circular Uni. SO02830998
Coedymwstwr 2.50 Hilltop Sub-Circular Uni. SS94348099
Colomendy Farm - River Terrace Sub-Circular Uni. ST35129399
(Caerleon)
Corn-y-Fan 0.09 Promontory Sub-Circular Multi. SN98503540
Corntown Farm - - Sub-Circular - SS92457725
Crack Hill - Hillslope Sub-Circular Bi. SS94557657
Craig-ty-Isaf 0.20 Spur Sub-Circular Multi. SS75659338
Craig-y-Dinas 2.67 Inland Promontory Irregular Multi. SN91500810
Craig-y-Darren (B) - Hillslope Circular - SS79619442
Craig-y-Parc 0.20 Inland Promontory Sub-Circular Uni. ST09308082
Craig-y-Rhiwarth 0.24 Promontory Cliff Circular Uni. SN84591572
Craig Ruperra 2.00 Ridge Sub-Circular Multi. ST223867
Craig Tan-y-Lan 0.90 Spur Sub-Circular Uni. SS95857954
Crawley Rocks Fort 0.10 Coastal Irregular Bi. SS51888796
(Nicolaston Burrows) Promontory
Croft-y-Bwla - Hillslope - Uni. SO49061361
Cross Oak 0.80 Hilltop Circular Uni. SO11042355
Crug-y-Gaer 0.43 Hillslope Circular Uni. SO224153
Crug Hywel 0.63 Promontory Sub-Circular Bi. SO22552065
Cuhere Wood - Hillslope Sub-Rect. Uni. ST45659277
Cwm-Cidy - - Rectangular - ST09346781
Cwm Bach 0.30 Coastal Cliff Triangular Uni. SS89707173
Cwm Coed-Drwg - - Circular - ST09297610
Cwm Col-Huw - - Circular - SS96106790
Cwm Col-Huw NW - - Rectangular - SS95966785
Cwm Ffairty 0.10 Spur Sub-Circular Uni. SS78519022
Cwm Nash Cropmarks - Hillslope Sub-Rect. - SS90657067
Cwm Nash Enclosure - Inland Promontory Sub-Circular Multi. SS90847036
Danish Camp (Cwm Philip 0.10 Hillslope Sub-Circular Uni. SS82078706
West)
Danish Fort (Sully Island) 0.40 Coastal Cliff - Multi. ST16856699
Delly’s Brake - Hillslope Sub-Circular Uni. ST41529032
Dol-y-Gaer 0.75 Ridge Circular Uni. SO05951480
Drostre 0.65 Hilltop Circular Uni. SO10203137
Druids Moor 0.20 Hillslope Sub-Circular Uni. SS44169016
Dunraven 6.50 Coastal Cliff Irregular Bi. SS88677279
Dyffryn Farm - Valley Side Sub-Rect. Uni. ST39959550
Earthwork Near Fforest 0.50 Lip of Cwm Sub-Circular Uni. SN63630155
Newydd
East Orchard Wood - Hillslope Sub-Rect. Bi. ST02786785
East Orchard Wood West - Hillslope - - ST02546780
Farthing Hill - Hilltop Complex Uni. ST49428886
150

Ffinant Isaf 0.24 Hillslope Circular Bi. SN98193195


Ffordd Cottage - - - - ST09847593
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Ffosyddu - Hilltop Sub-Circular Multi. SO34850555
Flemings Down 0.45 Lip of Cwm Irregular Uni. SS88957680
Flemingston - - Circular - ST02366937
Fonmon Castle Wood 0.30 Hillslope Sub-Circular Uni. ST04466824
Fonmon Northeast - - Rectangular Bi. ST04866777
(The) Gaer (Aberllynfi) 0.45 Promontory Sub-Circular Bi. SO17503758
Gaer (Trellech United) 3.50 Hillslope Circular Bi. SO49300377
Gaer (Llanelly) 0.30 Hillslope Circular Uni. SO22481530
Gaer Fach 0.48 Ridge Circular Bi. SO00903664
Gaer Fawr (Llangwm) 3.50 Ridge Sub-Circular Multi. ST44149881
Gaer Fawr (Merthyr 0.48 Spur Circular Bi. SO02203805
Cynog)
Gaer Hill Camp 0.50 Hillslope Sub-Circular Bi. ST51709796
Gelli Nedd 0.41 Ridge Circular Uni. SN91741399
Geulan Las (Mynydd 0.40 Hillslope Sub-Rect. Uni. SS81298851
Margam)
Glen Usk - Hilltop Sub-Circular Multi. ST36199311
Glyn-Neath Fort 0.1 Summit Circular Uni. SN88500685
Golden Hill Earthwork 1.00 Hilltop Circular Uni. ST42589753
Graig Fawr 0.50 Ridge Sub-Circular Uni. SN61830685
Great House (Usk) - Ridge - Bi. SO387002
Great House Camp 2.37 Natural Terrace Sub-Circular Multi. SO43230337
Great House Farm - - Polygonal Uni. SS98416935
Green Down - Plateau Sub-Rect. Uni. ST06047227
Gron Gaer 0.10 Spur - Uni. SS55049476
(The) Grondre - Hilltop Sub-Circular Bi. ST49349372
Gwaunshonbrown Farm - Natural Terrace Circular Uni. ST26198435
Gwernesney Defended 1.60 Flood Plain Sub-Circular Bi. SO41350298
Enclosure (Usk)
Gwersyll Enclosure 0.20 Hilltop Sub-Circular Uni. SO02700403
Half Moon Camp I 0.20 Spur Sub-Circular Uni. SS79968673
Half Moon Enclosure - - Circular Bi. SS79908676
(Margam)
Hardings Down East 0.90 Hilltop Irregular Multi. SS43709064
Hardings Down Lower 0.20 Hillslope Circular Uni. SS43669083
(North) Camp
Hardings Down West Camp 0.60 Spur Sub-Circular Multi. SS43439078
Hatterall Hill - Summit - Uni. SO30762554
Hen Gastell (Dan-y-Lan 0.10 Hillslope Sub-Circular Uni. SS55439578
Camp)
Hendre Hafaidd 0.25 Inland Promontory Sub-Rect. Uni. SO34371277
Heol-y-Mynydd I - - Sub-Rect. - SS88417513
Heol-y-Mynydd II - - Irregular Uni. SS88417512
High Pennard Fort 0.40 Coastal Irregular Bi. SS56778663
Promontory
Hillis 3.6 Ridge Sub-Circular Multi. SO11403273
Home Farm South - - Rectangular - ST04566722
Hopkins Mount - - Rectangular - ST171678
151

(Swanbridge)
Horgrove Enclosure - Hillslope Circular Uni. SS85588150
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Horse Cliff Fort 0.40 Coastal Irregular Uni. SS43508604
Promontory
Ifton Great Wood - - Rectangular Uni. ST45798922
Ifton Manor I - Level Sub-Circular Bi. ST46808820
Ifton Manor II - Level Sub-Rect. Uni. ST46808820
Ifton Reen - Level Circular Uni. ST47158682
Ilston Parish Enclosure 0.05 Hillslope Sub-Circular Multi. SS54888921
Kemey’s Folly - Ridge Rectangular Uni. ST38429214
Kenson Wood (Concentric 0.10 Lip of Cwm Sub-Rect. Multi. ST04686876
Ditched Fort)
Kenson Wood Northwest - Level Rectangular - ST04406875
(The) Knave Fort 0.10 Coastal Cliff Sub-Circular Bi. SS43188637
Kymin Hill - Summit Sub-Circular Bi. SO52691255
Lan Fawr 0.22 Spur - - SO17071838
(The) Larches Camp 1.24 Hilltop Sub-Circular Bi. ST43298996
Lavernock - - Rectangular - ST18256855
Lavernock East - - Circular - ST18376862
Lewes Castle Fort 0.30 Coastal Irregular Bi. SS41448734
Promontory
Limpert Bay - Coastal Sub-Circular - ST01366649
Llampha Court S - - - Multi. SS92737493
Llanbethery 0.60 Lip of Cwm Sub-Rect. Bi. ST03557023
Llancadle Gorse - Level Sub-Circular - ST04076851
Llancadle South - Spur Rectangular Uni. ST03776807
Llancayo Camp 2.00 Hilltop Sub-Circular Uni. SO37900380
Llandarcy - - Circular - SS71579527
Llandough Enclosure - Spur Sub-Circular Bi. SS99447354
Llanedeyn Bridge - - Circular - ST22558230
Llanfrynach East - Hilltop Circular Bi. SS98027512
Llanfythin (Llanvithyn) 0.10 Spur Irregular Bi. ST05487193
Llanmaes East - - Circular - SS98386962
Llanmartin Enclosures - Hillslope Sub-Rect. Uni. ST396898
Llanmelin Wood Camps 6.56 Hilltop Complex Multi. ST46109257
Llanmelin Wood Outwork - Ridge Sub-Rect. Bi. ST46309385
Llanquian Wood 1.00 Hillslope Circular Bi. ST02147446
Llantrithyd Camp 0.40 Hillslope Sub-Circular Bi. ST03867318
(Llancarfan)
Llantwit Major Bypass - Summit Sub-Rect. - SS96856960
Lle’r Gaer 0.50 Spur Polygonal Uni. ST05018703
Llwyfen - Hillslope Circular Uni. SO13133169
Llwyn-Llwyd - Hillslope Circular Uni. SN99963869
Llwyn-y-Gaer - Hillslope Irregular Uni. SO406107
Llwyn Bedw 0.10 Hillslope Circular Multi. SO00202424
Llwyn on Village I - Hillslope Sub-Rect. - SO024113
Llwyn on Village II - Hillslope Sub-Rect. - SO024113
Llwynda-Ddu Camp 0.40 Hilltop Sub-Circular Bi. ST10858100
Llwynheiernin 0.25 Level Sub-Circular Uni. SS67379472
Llys-Wen 0.70 Hillslope Circular Multi. SO12773790
152

Lodge Wood 9.58 Ridge Sub-Rect. Multi. ST32309130


Lower Tredogan North - - Circular Multi. ST06816810
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Maendy Camp 0.20 Spur Sub-Circular Bi. SS95739551
Maes-Llech - Hillslope Sub-Rect. Uni. ST37569522
Maes-y-Felin - Hillslope Rectangular Uni. ST03408085
Maiden Castle Fort 0.60 Hillslope Polygonal Bi. SS509854
(Oxwich Point)
Merthyr Cynog 0.66 Ridge Circular Uni. SN984375
Minorca (Michaelstone-y- - - Rectangular - ST24078603
Fedw)
Mitchel-Troy - Hillslope - Multi. SO49001070
Moorcorner Farm - - Sub-Rect. Uni. SS46718638
Moulton - - Sub-Rect. - ST07757009
(The) Mount (Pen-y-Lan) 1.56 Hilltop Circular Uni. ST25858487
Myarth (Penmyarth) - Promontory - Uni. SO17202070
Mynydd-y-Castell 2.70 Hilltop Sub-Circular Uni. SS80618655
Mynydd-y-Fforest 0.30 Hillslope Circular Uni. ST00747835
Mynydd-y-Fforest - - Rectangular Uni. ST00857840
(Enclosure E of Hillfort)
Mynydd-y-Gaer (Coed Cae 1.00 Spur Polygonal Uni. SS97358495
Gaer)
Mynydd-y-Gaer (Gaer 0.10 Natural Terrace Sub-Circular Multi. SS76579425
Fawr) Lower Camp
Mynydd Bychan 0.30 Hillslope Triangular Multi. SS96307560
Mynydd Llangorse (Cockit 0.09 Ridge Triangular Uni. SO16032774
Hill)
Mynydd Pen-Hydd (Nant 0.05 Hillslope Sub-Circular Uni. SS820940
Herbert)
Mynydd Portref I - - Circular - SS99778627
Mynydd Portref II - - Circular - SS99778627
Mynydd Twmpathyddaear 1.20 Hilltop Circular Uni. SS84058037
Nant-y-Durfol I - Hillslope Rectangular Uni. SS90207280
Nant-y-Durfol II - Hillslope Circular Uni. SS90207280
Nant Cwm Moel 0.12 Hillslope Circular Uni. SO039113
Nant Tarthwyni I 0.26 Hillslope Sub-Rect. Uni. SO08951944
Nant Tarthwynni II 0.38 Hillslope Circular Bi. SO09111951
Nash Point Camp 0.40 Coastal Irregular Multi. SS91486848
Promontory
New Breach Farm (Nash) - - Circular - SS97207300
New Inn Brake - Hillslope Sub-Circular Uni. SO47960101
New Mill Farm (Monknash) 0.40 Level Circular Bi. SS91156980
New Park - Hillslope Sub-Rect. Uni. ST25288419
Newhouse (Llantrisant) 0.20 River Terrace Circular Uni. ST40889555
North Hill Tor (Nottle Tor 1.85 Coastal Cliff Sub-Circular Uni. SS45309381
Camp)
Ogmore 0.80 Hilltop Sub-Circular Bi. SS86757580
Old Castle Fort (Rhossili 0.70 Coastal Irregular Uni. SS40938798
Cliffs) Promontory
Palmerston - Hilltop Sub-Rect. - ST140695
Pant-yr-Hyl (Mount - Summit Circular - SS83317921
153

Pleasant)
Pant-y-Pyllau (Parcnewydd) 0.40 Level Sub-Circular Uni. SS92738245
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Pantywenallt 0.18 Hillslope Sub-Rect. Uni. SO11182209
Parc Coed Machen 0.60 Hillslope Rectangular Uni. ST08507910
Paviland Manor Cropmark - - Sub-Circular - SS44818611
Pen-Coed - Hillslope Sub-Circular Uni. ST40808930
Pen-Llwyn-Mawr Fort - Hillslope Rectangular Bi. ST17429535
Pen-Toppen-Ash Camp 0.45 Ridge Circular Bi. ST37859152
Pen-y-Castell (Cwmavon) 0.20 Hilltop Sub-Circular Multi. SS78859174
Pen-y-Castell (Kenfig Hill) 0.40 Ridge - Bi. SS84228270
Pen-y-Crug 1.86 Promontory Sub-Circular Multi. SO02933037
Pen-y-Gaer 0.90 Ridge Sub-Circular Uni. SS53659554
Pen-y-Garn 0.06 Hillslope Rectangular Uni. SO06042703
Pen-y-Parc (Raglan) - Hilltop Sub-Circular Uni. SO44090894
Pen-yr-Allt 0.23 Hilltop Sub-Circular Uni. SO11083269
Pen Twyn 4.17 Spur Complex Multi. SO32112303
Pencaerau 0.10 Level Sub-Circular Uni. SS74849548
Pencoed Lane (Llanmartin) - - Sub-Rect. Uni. ST40308951
Pendre 1.57 Promontory Sub-Circular Bi. SO15583263
Penffawyddog 0.40 Hillslope Circular Uni. SO19571845
Pengwern Common (Ilston) - - - - SS54369152
Penmark - Hillslope Circular - ST04616892
Penmark Place - - Trapezoidal - ST05246838
Penmark Place (Cowshed - - Sub-Rect. - ST05406847
Field)
Pennard Pill SS53828851
Pentwyn - Ridge Sub-Circular Uni. ST41459971
Pentwyn Triley - Level - Uni. SO31161758
Perth-Hir West - Ridge - Uni. SO482159
Pierce Wood Camps 0.20 Inland Promontory Sub-Rect. Uni. ST53259572
Piercefield Camp - Inland Promontory Sub-Rect. Uni. ST53639596
Pill Farm - Level - Uni. ST41438569
Pilton Green Cropmark - Hillslope - - SS44708770
Plas-y-Gaer 0.58 Hillslope Sub-Circular Bi. SO03292462
Pontsticill - - Rectangular - SO05411174
Porthkerry Bulwarks 4.10 Coastal Cliff Sub-Rect. Multi. ST08206630
Priory Wood Camp 3.87 Hilltop Sub-Rect. Bi. ST36089043
Pwll-y-Cwrw 0.80 Ridge Sub-Circular Bi. SO09303510
Rectilinear Enclosures 0.50 Level Rectangular Uni. ST05226847
(Kenson South)
Redley (Caswell) Cliff 0.20 Coastal Irregular Bi. SS58808756
Camp Promontory
Reynoldston Camp 0.20 Hillslope Circular Uni. SS48348991
Rhiwderin Camp 1.19 Hilltop Circular Uni. ST26408773
Rhondda Earthwork - - Circular - SN92310435
Rhyd Uchaf 0.01 Hillslope Triangular Multi. SN98741776
Rhyle Cropmark I - Hillslope Sub- - SS91677245
Triangular
Rhyle Cropmark II - Hillslope Rectangular - SS91677245
Rills Valley - - Sub-Circular Bi. ST02626838
154

Rogiet - - - Uni. ST45728777


Sewage Farm (Wick) - - Rectangular - SS925727
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Sluxton Cropmark - Hillslope Sub-Circular - SS43188966
Slwch Tump 3.80 Promontory Sub-Circular Uni. SO05622840
Stembridge Camp 0.20 Inland Promontory Triangular Bi. SS46979145
St. Athan - - Sub-Circular Uni. ST00186727
St. Illtyd’s (Llanilltyd) 0.57 Hilltop Circular Uni. SN97122611
Churchyard
St. Julian’s Wood Camp 0.15 Ridge Sub-Rect. Uni. ST34038917
St. Mary Hill Down 0.20 Hillslope Circular Uni. SS96727905
St. Pierre’s Wood - Hillslope Sub-Rect. Uni. ST49749238
Sudbrook Camp 3.21 Coastal Triangular Multi. ST50558732
Promontory
Sully Enclosure - Hillslope Sub-Rect. - ST17336872
Summerhouse Camp 1.30 Coastal Cliff Sub-Circular Multi. SS99456647
Tair Croes Down - - Circular Bi. SS915766
Talaches Farm - Hillslope Sub-Rect. Bi. SO48451109
Taloches Farm - Hillslope - Uni. SO484115
Thurba Head Camp 0.30 Coastal Irregular Multi. SS42208705
Promontory
Tor-Gro 0.40 Ridge Sub-Circular Uni. SS46109355
Tre Adam - Hillslope - Uni. SO38191580
Tre Durn Wood - - Sub-Rect. - SO10673472
Tredegar Hillfort 9.28 Hilltop Sub-Rect. Multi. ST28958684
Tredogan (NW) 0.15 Level Sub-Circular Bi. ST06816798
Tregare 0.20 Hilltop Sub-Circular Bi. SO41761027
Trelleck Grange - Hilltop Polygonal Uni. SO48920188
Trelleck Grange - Ridge Sub-Circular Uni. SO49330273
(Parkhouse)
Tump Wood 0.63 Promontory Circular Bi. SO11272149
Twm Barlwm 4.20 Ridge Sub-Circular Uni. ST24219261
Twyn-y-Gaer (Cwmyoy) 2.98 Summit Complex Bi. SO29402195
Twyn-y-Gaer (Llanfihangel 0.38 Hilltop Sub-Circular Uni. SO05443526
Fechan)
Twyn-y-Gaer (Llansyddid) 0.70 Hilltop Sub-Circular Uni. SN99002805
Twyn-y-Gaer (Trallong) 0.45 Spur Circular Uni. SN96953060
Twyn-yr-Allt 0.10 Summit Sub-Rect. Uni. SO296163
Twyn Bell Camp 0.06 Inland Promontory Sub-Circular Bi. SO37530004
Twyn Llechfaen 0.37 Promontory Triangular Uni. SO08212911
Ty Bryn (Clemenstone) - Hillslope - Multi. SS91687342
Tyle Clydach 1.00 Spur Sub-Circular Uni. SO09742175
Tynewydd-y-Bryn - - - Multi. ST07287560
Tythegston Enclosure - - Polygonal - SS86677908
Ty’n-y-Coed - Hillslope Rectangular Uni. ST06777295
Ty’n-y-Waun 0.40 Spur Sub-Circular Uni. SS94858527
Unknown (Clyne and - - - - SS81689971
Melincourt)
Unknown (Craig-y-Darren - Ridge Circular Uni. SS79539405
A)
Unknown (Hirwaun) I - - Circular - SN94591061
155

Unknown (Hirwaun) II - - Circular - SN96800985


Unknown (Penllyn) - - - Bi. SS99957645
Name Internal Topographic Shape Vallation NGR
Size (ha) Location
Unknown (Penrice) - - - - SS50908495
Unknown (Rhigos) - - Circular - SN92320431
Unknown (Vaynor) 0.12 Hillslope Sub-Circular Uni. SO03961131
Upper Pink - Hillslope Sub-Circular Uni. ST45159615
Walnut Tree - Hillslope Sub-Rect. Uni. ST33449682
Walterston Farm - - Sub-Circular Multi. ST06827118
Warren Hill 0.80 Hilltop Sub-Rect. Uni. SS73659410
Waun-Arw (Langstone) - - Sub-Rect. Uni. ST40948862
Waun-y-Pwll Farm - Hilltop Sub-Circular Uni. ST313954
Welsh St. Donats - Hillslope Circular Uni. ST03407670
Wenallt Camp 0.20 Hillslope Sub-Circular Uni. ST15228276
Westwood Corner - Promontory - - ST09196689
White Castle Farm/Tredam 1.50 Hilltop Circular Multi. SO38071618
Enclosure
Whitmore Stairs - Promontory Fort Rectangular - SS89887145
Whitton Lodge 0.30 Level Sub-Rect. Uni. ST08367127
Wig Fach - - Circular - SS85387793
Wilcrick Hill Camp 1.10 Hilltop Circular Multi. ST41118780
Willis Hill - Hilltop Sub-Rect. Uni. ST505922
Willoxton Cwm Cropmark - - Circular - SS54308980
Windmill East - - Circular - SS97406927
Windmill Farm (Llanfair - Hillslope Rectangular Uni. ST42929203
Discoed)
Windmill Lane - Hillslope Sub-Circular Bi. SS99507410
(Llanblethian)
Windmill North - - - - SS97396948
Worm’s Head Fort 0.60 Coastal Cliff Sub-Circular Uni. SS39358755
Y Bwlwarcau 7.20 Spur Sub-Circular Multi. SS83888855
Y Gaer (Bonvilston Gaer) 1.00 Level Sub-Circular Bi. ST06357473
Y Gaer 1.00 Hilltop Sub-Circular Uni. ST27338620
(Graig-y-Saeson)
Y Graig (Porthcawl) - - Sub-Circular - SS843779
Y Gaer (Twyn-y-Gaer 0.70 Ridge Circular Uni. SN92252630
(Defynnog))
Yellow Top Fort 0.10 Coastal Irregular Multi. SS43708596
Promontory
Ysgyryd (Skirrid) Fawr 1.40 Ridge Complex Multi. SO33111827
Hillfort
156

Appendix 2
Roman Military Occupation in South Wales615

Map 9: Pre-Flavian Roman sites to AD 60.

615
All maps are ©Joanne Edwards/Crown Copyright Ordnance Survey. All rights reserved.
157

Map 10: Roman sites to AD 80.

Map 11: Roman sites to AD 150.


158

Map 12: Roman sites c. AD 300.


159

Illustrations

Illus. 2: Terrets – Chariot Rein-Guides (c. 300 BC –


AD 100).
Illus. 1: Seven Sisters Hoard (c. AD 1-150).
© Copyright The National Museum of Wales, Cardiff Num. 9: Castle Ditches Hillfort, Llancarfan, Vale of
Glamorgan. Num. 10: Windmill Farm, Llansannor, Vale
of Glamorgan.
© Copyright The National Museum of Wales, Cardiff

Illus. 3: Bronze Strap-Union with Horse Bell of La Tène or Illus. 4: Bronze Moulded Tankard Handles (c. AD 1-
Celtic Art style (c. AD 1-120). 150).
Found in Maindy Camp Hillfort, Rhondda Cynon Taf. Found near Coelbren Roman Fort.
© Copyright The National Museum of Wales, Cardiff © Copyright The National Museum of Wales, Cardiff

Illus. 6: A dupondius of Antoninus Pius (c. AD 140-4).


Illus. 5: The Goldcliff Stone. Excavated at Cowbridge.
© Copyright The National Museum of Wales, Cardiff © Copyright GGAT HER Charitable Trust
160

Illus. 8: T-Shaped fibula (c. late first to


middle second century AD)
Illus. 7: Light Poldon Fibula (c. first century Excavated at Cowbridge.
AD)
© Copyright GGAT HER Charitable Trust
Excavated at Cowbridge.
© Copyright GGAT HER Charitable Trust

Illus. 10: Iron Ring with intaglio of a pastoral scene (c.


first or early second century AD)
Excavated at Loughor.
Illus. 9: Iron Intaglio Ring of © Copyright GGAT HER Charitable Trust
Achilles (c. first half of the
second century AD)
Excavated at Loughor.
© Copyright GGAT HER Charitable Trust
161

Works Cited

Ancient Authors
Caesar. Gallic Wars. (trans: H.J. Edwards. Cambridge, Mass.: Harvard University Press.
1917).
Cassius Dio. Roman History. (trans: E. Cary. London: William Heinemann. 1917).
Historia Augusta. Hadrian. (trans: D. Magie. Cambridge, Mass.: Harvard University Press.
1991).
Strabo. Geography. (trans: H.L. Jones. Cambridge Mass.: Harvard University Press. 1917).
Suetonius. The Twelve Caesars. (trans. R. Graves. London: Penguin Books. 1957).
Tacitus. Agricola. (trans: M. Hutton. Cambridge, Mass.: Harvard University Press. 1970).
Tacitus. Annals. (trans: J. Jackson. Cambridge, Mass.: Harvard University Press. 1937)
Tacitus. Histories. (trans: K. Wellesley. London: Penguin Books. 1995).

Modern Authors
Alcock, L. (1955). ‘The Hill-Fort in Cwrt-yr-Ala Park, near Dinas Powis (Glam.): I. The
Defences’, Bulletin of the Board of Celtic Studies. 16, 242-250.
Alcock, L. (1956). ‘Roman Sherds from Castle Ditches, Llancarfan (Glam.)’, Bulletin of the
Board of Celtic Studies. 16, 299-300.
Alcock, L. (1963). Dinas Powys: An Iron Age, Dark Age and Early Medieval Settlement in
Glamorgan. Cardiff: University of Wales Press.
Alcock, L. (1980). ‘The Cadbury Castle Sequence in the First Millennium B.C.’, Bulletin of
the Board of Celtic Studies. 28, 656-718.
Aldhouse-Green, M. (2004). ‘The Iron Age: Art, Ritual and Society’, in Aldhouse-Green, M.
and Howell, R. (eds.) (2004). The Gwent County History Vol. 1: Gwent in Prehistory
and Early History. Cardiff: University of Wales Press, 160-77.
Aldhouse-Green, M. and Howell, R. (eds.) (2004). The Gwent County History Vol. 1: Gwent
in Prehistory and Early History. Cardiff: University of Wales Press.
Allason-Jones, L. (2008). Daily Life in Roman Britain. Oxford: Greenwood World.
Allen, J.R.L. (1998). ‘Late Iron Age and Earliest Roman Calcite-tempered Ware from Sites
on the Severn Estuary Levels: Character and Distribution’, Studia Celtica. 32, 27-41.
Allen, J.R.L. and Fulford, M.G. (1986). ‘The Wentlooge Level: A Romano-British Saltmarsh
Reclamation in Southeast Wales’, Britannia. 17, 91-117.
162

Allen, S.J. and Bell, M. (2000). ‘Rectangular Buildings of the Later 1 st Millennium BC at
Goldcliff: Dating, Character and Reconstruction’, in Bell, M., Caseldine, A. and
Neumann, H. (2000). Prehistoric Intertidal Archaeology in the Welsh Severn
Estuary. York: Council for British Archaeology Research Report 120, 128-35.
Applebaum, S. (1966). ‘Peasant Economy and Types of Agriculture’, in Thomas, C. (ed.)
(1966). Rural Settlement in Roman Britain: Papers Given at a C.B.A. Conference
Held at St. Hugh’s College Oxford January 1 to 3, 1965. London: Council for British
Archaeology Research Report 7, 99-107.
Arnold, C.J. and Davies, J.L. (2000). Rome and Early Medieval Wales. Stroud: Sutton.
Barber, A., Cox, S. and Hancocks, A. (2007). ‘A Late Iron Age and Roman Farmstead at
RAF St. Athan, Vale of Glamorgan. Evaluation and Excavation 2002-03’,
Archaeologia Cambrensis. 155, 49-115.
Barford, P.M. (1996). ‘The Metalworking Debris from Bear Barn and the Bear Field’, in
Parkhouse, J. and Evans, E. (eds.) (1996). Excavations in Cowbridge, South
Glamorgan, 1977-88. Oxford: British Archaeological Reports, British Series 245,
205-9.
Barnett, C., Stanley, P., Trett, R. and Webster, P.V. (1990). ‘Romano-British Pottery Kilns at
Caldicot, Gwent’, Archaeological Journal. 147, 118-47.
Bell, M. (2001). ‘Interim Report on the Excavation of a Middle Bronze Age Settlement at
Redwick 2000-1’, Archaeology in the Severn Estuary.12, 99-117.
Bell, M and Neumann, H. (1997). ‘Prehistoric Intertidal Archaeology and Environments in
the Severn Estuary, Wales’, World Archaeology. 29(1): Riverine Archaeology, 95-
113.
Bell, M. and Neumann, H. (1999). ‘Intertidal Survey, Assessment and Excavation of a Bronze
Age Site at Redwick, Gwent, 1999’, Archaeology in the Severn Estuary. 10, 25-37.
Bell, M., Caseldine, A. and Neumann, H. (2000). Prehistoric Intertidal Archaeology in the
Welsh Severn Estuary. York: Council for British Archaeology Research Report 120.
Bevan, B. (ed.) (1999). Northern Exposure: Interpretive Devolution and the Iron Ages in
Britain. Leicester: Leicester University Archaeology Monograph 4.
Binford, L.R. (1971). ‘Mortuary Practices: Their Study and Their Potential’, Memoirs of the
Society for American Archaeology, No. 25: Approaches to the Social Dimensions of
Mortuary Practices, 6-29.
Bird, D.G. (2000). ‘The Claudian Invasion Campaign Reconsidered’, Oxford Journal of
Archaeology. 19, 91-104.
Birley, A.R. (2005). The Roman Government of Britain. Oxford: Oxford University Press.
Birley, A.R. (ed.) (1988). Roman Papers V. Oxford: Clarendon Press.
163

Black, E.W. (1994). ‘Villa-Owners: Romano-British Gentlemen and Officers’, Britannia. 25,
99-110.
Boon, G.C. (1966). “Legionary’ Ware at Caerleon?’, Archaeologia Cambrensis. 115, 45-66.
Boon, G.C. (1972). Isca: The Roman Legionary Fortress at Caerleon, Mon. 3rd ed. Cardiff:
National Museum of Wales.
Boon, G.C. (1978). Monographs and Collections: Relating to Excavations Financed by H.M.
Department of the Environment in Wales. I. Roman Sites. Cardiff: Cambrian
Archaeological Association.
Boon, G.C. (1988a). ‘The Samian Ware’, in Vyner, B.E. and Allen, D.W.H. (1988). ‘A
Romano-British Settlement at Caldicot, Gwent’, in Robinson, D.M. (ed.) (1988).
Biglis, Caldicot and Llandough: Three Late Iron Age and Romano-British Sites in
South-East Wales. Oxford: British Archaeological Reports British Series 188, 100-1.
Boon, G.C. (1988b). ‘Note on Late Roman Coinage in the Vale of Glamorgan’, in Parkhouse,
J. (1988). ‘Excavations at Biglis, South Glamorgan’, in Robinson, D.M. (ed.) (1988).
Biglis, Caldicot and Llandough: Three Late Iron Age and Romano-British Sites in
South-East Wales. Oxford: British Archaeological Reports British Series 188, 52.
Boon, G.C. (1988c). ‘The Coins’, in Parkhouse, J. (1988). ‘Excavations at Biglis, South
Glamorgan’, in Robinson, D.M. (ed.) (1988). Biglis, Caldicot and Llandough: Three
Late Iron Age and Romano-British Sites in South-East Wales. Oxford: British
Archaeological Reports British Series 188, 51-2;
Boon, G.C. (1988d). ‘The Coins’, in Vyner, B.E. and Allen, D.W.H. (1988). ‘A Romano-
British Settlement at Caldicot, Gwent’, in Robinson, D.M. (ed.) (1988). Biglis,
Caldicot and Llandough: Three Late Iron Age and Romano-British Sites in South-
East Wales. Oxford: British Archaeological Reports British Series 188, 91;
Boon, G.C. (1988e). ‘The Coins’, in Owen-John, H.S. (1988b). ‘Llandough: The Rescue
Excavation of a Multi-Period Site Near Cardiff, South Glamorgan’, in Robinson,
D.M. (ed.) (1988). Biglis, Caldicot and Llandough: Three Late Iron Age and
Romano-British Sites in South-East Wales. Oxford: British Archaeological Reports
British Series 188, 158-9.
Boon, G.C. (1988f). ‘Miscellaneous Finds’, in Vyner, B.E. and Allen, D.W.H. (1988). ‘A
Romano-British Settlement at Caldicot, Gwent’, in Robinson, D.M. (ed.) (1988).
Biglis, Caldicot and Llandough: Three Late Iron Age and Romano-British Sites in
South-East Wales. Oxford: British Archaeological Reports British Series 188, 93-99.
Boon, G. C. and Lewis, J. (eds.) (1976). Welsh Antiquity. Cardiff: National Museum of
Wales.
Bowden, M. and McOmish, D. (1987). ‘The Required Barrier’, Scottish Archaeological
Review. 4(2), 76-84.
164

Branigan, K. (1982). ‘Celtic Farm to Roman Villa’, in Miles, D. (ed.) (1982a). The Romano-
British Countryside: Studies in Rural Settlement and Economy. Part I. Oxford: British
Archaeological Reports British Series 103(I), 81-96.
Branigan, K. (1989). ‘Specialisation in Villa Economies’, in Branigan, K. and Miles, D. (eds.)
(1989). The Economies of Romano-British Villas. Sheffield: Department of
Archaeology and Prehistory, 42-50.
Branigan, K. and Miles, D. (eds.) (1989). The Economies of Romano-British Villas. Sheffield:
Department of Archaeology and Prehistory.
Braun, D. P. (1981). ‘A Critique of Some Recent North American Mortuary Studies’,
American Antiquity. 46(2), 398-416.
Brewer, R.J. (1993a). Caerwent Roman Town. Cardiff: Cadw: Welsh Historic Monuments.
Brewer, R.J. (1993b). ‘Venta Silurum: A Civitas-Capital’, in Greep, S.J. (1993). Roman
Towns: The Wheeler Inheritance. York: Council for British Archaeology Research
Report 93.
Brewer, R.J. (1996). Gelligaer Roman Fort. Caerphilly: Rhymney Valley District Council.
Brewer, R.J. (2004). ‘The Romans in Gwent’, in Aldhouse-Green, M. and Howell, R. (eds.)
(2004). The Gwent County History Vol. 1: Gwent in Prehistory and Early History.
Cardiff: University of Wales Press, 205-243.
Briggs, C.S. (1985). ‘Problems of the Early Agricultural Landscape in Upland Wales, as
Illustrated by an Example from the Brecon Beacons’, in Spratt, D. and Burgess, C.
(1985). Upland Settlement in Britain: The Second Millennium B.C. and After. British
Archaeological Reports British Series 143, 285-316.
Brothwell, D. and Bourke, J.B. (1995). ‘The Human Remains from Lindow Moss 1987-8’, in
Turner, R.C. and Scaife, R.G. (eds.) (1995). Bog Babies: New Discoveries and New
Perspectives. London: British Museum Press, 52-8.
Bulleid, A. (1917). ‘Pottery’, in Bulleid, A. and Gray, H.S.G. (1917). The Glastonbury Lake
Village: A Full Description of the Excavations and the Relics Discovered, 1982-1907.
v. 2. Glastonbury: Antiquarian Society, 486-553.
Bulleid, A. and Gray, H.S.G. (1917). The Glastonbury Lake Village: A Full Description of
the Excavations and the Relics Discovered, 1982-1907. v. 2. Glastonbury:
Antiquarian Society.
Burgess, C. (1985). ‘Population, Climate and Upland Settlement’, in Spratt, D. and Burgess,
C. (1985). Upland Settlement in Britain: The Second Millennium B.C. and After.
British Archaeological Reports British Series 143, 195-230.
Burnham, H. (1995). A Guide to Ancient and Historic Wales: Clwyd and Powys. London:
HMSO.
Burnham, B.C. (2000). ‘Roman Britain in 1999: 1. Wales’, Britannia. 31, 372-9.
165

Burnham, B.C. (2004). ‘Roman Britain in 2003: 1. Wales’, Britannia. 35, 254-65.
Burnham, B.C. (2005). ‘Roman Britain in 2004: 1. Wales’, Britannia. 36, 384-2.
Burnham, B.C. and Davies, J.L. (eds.) (1990). Conquest, Co-existence and Change: Recent
Work in Roman Wales. Lampeter: Trivium 25.
Burnham, B.C. and Davies J.L. (eds.) (2010). Roman Frontiers in Wales and the Marches.
Aberystwyth: RCAHMW.
Burnham, H. (1995). A Guide to Ancient and Historic Wales: Clwyd and Powys. London:
HMSO.
Cancian, F. (1976). ‘Social Stratification’, Annual Review of Anthropology. 5, 227-48.
Caseldine, A.E. (1990). ‘The Archaeobotanical Evidence from a Roman ‘T’-shaped Kiln at
Dan-y-Graig’, in Newman, R. (1990). ‘Excavations of a Romano-British Building at
Dan-y-Graig, Porthcawl, Mid Glamorgan (SS 8407 7805)’, Bulletin of the Board of
Celtic Studies. 37, 274-7.
Casey, P.J. (ed.) (1979). The End of Roman Britain. Oxford: British Archaeological Reports
British Series 71.
Casey, P. J. (2010). ‘The end of the Roman army in Wales and the Marches’, in Burnham,
B.C. and Davies J.L. (eds.) (2010). Roman Frontiers in Wales and the Marches.
Aberystwyth: RCAHMW, 62-6.
Champion, T.C. and Collis, J.R. (eds.) (1996). The Iron Age in Britain and Ireland: Recent
Trends. Sheffield: J.R. Collis Publications.
Children, G. and Nash, G. (2001). Prehistoric Sites of Breconshire: Ideology, Power and
Monument Symbolism. Monuments in the Landscape Vol. 9. Hereford: Logasten
Press.
Chouls, W.H.M. (1993). ‘Rheda Forest, Hirfynydd (SN 8124 0401)’, Archaeology in Wales.
33, 59.
Clarke, S. (1993). ‘Monmouth, The Gloucestershire House, St. Mary’s St. (SO 505 126)’,
Archaeology in Wales. 33, 61.
Clarke, S., Jackson, R. and Jackson, P. (1992). ‘Archaeological Evidence for Monmouth’s
Roman and Early Medieval Defences’, Archaeology in Wales. 32, 1-2.
Collingwood, R.G. and Wright, R.P. (1965). The Roman Inscriptions of Britain I:
Inscriptions on Stone. Oxford: Oxford University Press.
Compton, J. and Webster, P.V. (2000). ‘The Coarse Pottery’, in Evans, E. (2000). The
Caerleon Canabae: Excavations in the Civil Settlement 1984-90. London: Britannia
Monograph Series No. 16, 198-264.
Costin, C.L. and Earle, T. (1989). ‘Status Distinction and Legitimation of Power as Reflected
in Changing Patterns of Consumption in Late Prehispanic Peru’, American Antiquity.
54(4), 691-714.
166

Cunliffe, B. (1971). ‘Some Aspects of Hill-forts and Their Cultural Environments’, in Hill, D.
and Jesson, M. (eds.) (1971). The Iron Age and its Hill-forts: Papers Presented to Sir
Mortimer Wheeler on the Occasion of his Eightieth Year, at a Conference Held by the
Southampton University Archaeological Society, 5th-7th March, 1971. Southampton:
Millbrook Press Ltd, 53-69.
Cunliffe, B. (1984). Danebury: An Iron Age Hillfort in Hampshire. 2 vols. London: Council
for British Archaeology.
Cunliffe, B. (1991). Iron Age Communities in Britain: An Account of England, Scotland and
Wales from the Seventh Century BC until the Roman Conquest. 3rd Edition. London:
Routledge.
Cunliffe, B. (1997). ‘Armorica and Britain: The Ceramic Evidence’, in Cunliffe, B. and De
Jersey, P. (1997). Armorica and Britain: Cross-Channel Relationships in the Late
First Millennium BC. Studies in Celtic Coinage 3. Oxford: Oxford University
Committee for Archaeology Monograph 45, 2-71.
Cunliffe, B. (1998). Fishbourne Roman Palace. Stroud: Tempus.
Cunliffe, B. (2003). Danebury Hillfort. Stroud: Tempus.
Cunliffe, B. (2005). Iron Age Communities in Britain: An Account of England, Scotland and
Wales from the Seventh Century BC until the Roman Conquest. 4th Edition. London:
Routledge.
Cunliffe, B. and De Jersey, P. (1997). Armorica and Britain: Cross-Channel Relationships in
the Late First Millennium BC. Studies in Celtic Coinage 3. Oxford: Oxford
University Committee for Archaeology Monograph 45.
Cunliffe, B. and Miles, D. (1984). Aspects of the Iron Age in Central Southern Britain.
Oxford: Oxford University Committee for Archaeology. Monograph 2.
Dark, K. and Dark, P. (1997). The Landscape of Roman Britain. Stroud: Sutton.
David, N. and Kramer, C. (2001). Ethnoarchaeology in Action. Cambridge: Cambridge
University Press.
Davies, J.L. (1967). ‘Caer Dynnaf, Llanblethian (SS 984744)’, Archaeology in Wales. 7, 7.
Davies, J.L. (1973). ‘Cae Summerhouse, Tythegston’, Morgannwg. 17, 53-7.
Davies, J.L. (1995). ‘The Early Celts in Wales’, in Green, M.J. (1995).The Celtic World.
London: Routledge, 671-700.
Davies, J.L. (2004). ‘Soldier and Civilian in Wales’, in Todd, M. (ed.) (2004). A Companion
to Roman Britain. Oxford: Blackwell, 91-113.
Davies, J.L. and Lynch, F. (2000). ‘The Late Bronze Age and Iron Age’, in Lynch, F.,
Aldhouse-Green, S. and Davies, J.L. (2000). Prehistoric Wales. Stroud: Sutton, 139-
219.
167

Davies, W. (1979). ‘Roman Settlements and Post-Roman Estates in South-east Wales’, in


Casey, P.J. (ed.) (1979). The End of Roman Britain. Oxford: British Archaeological
Reports British Series 71, 153-73.
De Jersey, P. (1997). ‘Armorica and Britain: The Numismatic Evidence’, in Cunliffe, B. and
De Jersey, P. (1997). Armorica and Britain: Cross-Channel Relationships in the Late
First Millennium BC. Studies in Celtic Coinage 3. Oxford: Oxford University
Committee for Archaeology Monograph 45, 72-103.
de la Bédoyère, G. (2003a). Defying Rome: The Rebels of Roman Britain. Stroud: Tempus.
de la Bédoyère, G. (2003b). Roman Towns in Britain. Stroud: Tempus.
Deroux, C. (ed.) (1997). Studies in Latin Literature and Roman History VIII. Bruxelles:
Latomus Revue d'Études Latines, Collection Latomus 239.
Earle, T.K. (1987). ‘Chiefdoms in Archaeological and Ethnohistorical Perspective’, Annual
Review of Anthropology. 16, 279-308.
Earle, T.K. (2002). Bronze Age Economics: The Beginnings of Political Economies. Boulder:
Westview Press.
Egloff, S. (2009) Later Prehistoric Undefended and Roman Rural Settlements in Southeast
Wales. GGAT Unpublished Report No.2009/013.
Egloff, S. and Evans, E. M. (2010) GGAT Unpublished Report No.2010/019.
Evans, D.R. (1992). ‘The Course Pottery’, in Marvell, A.G. and Heywood, B. (1992).
‘Excavations at Neath’, Bulletin of the Board of Celtic Studies, 263-76.
Evans, D.R. and Metcalf, V.M. (1992). Roman Gates Caerleon. Oxford: Oxbow Monograph
15.
Evans, E. (2000). The Caerleon Canabae: Excavations in the Civil Settlement 1984-90.
London: Britannia Monograph Series No. 16.
Evans, E. (2010a). ‘Hirfynydd (NGR: SN 8284 0663; NPRN 93165)’, in Burnham, B.C. and
Davies J.L. (eds.) (2010). Roman Frontiers in Wales and the Marches. Aberystwyth:
RCAHMW, 296.
Evans, E. (2010b). ‘Rheola Forest (NGR: ST 8127 0404; NPRN 301353)’, in Burnham, B.C.
and Davies J.L. (eds.) (2010). Roman Frontiers in Wales and the Marches.
Aberystwyth: RCAHMW, 298.
Evans, E. (2010c). Church Hill, Penmaen, Swansea. GGAT Report No. 2010/033.
Flannery, K.V. (1972). ‘The Origins of the Village as a Settlement Type in Mesoamerica and
the Near East: A Comparative Study’, in Ucko, P.J., Tringham, R. and Dimbleby,
G.W. (eds.) (1972). Man, Settlement and Urbanism. London: Duckworth, 23-53.
Flannery, K.V. (2002). ‘The Origins of the Village Revisited: From Nuclear to Extended
Households’, American Antiquity. 67(3), 417-33.
Frere, S. (1987). Britannia. 3rd Edition. London: Routledge and Kegan Paul, 175.
168

Fulford, M. (1982). ‘Town and Country in Roman Britain – A Parasitical Relationship?’, in


Miles, D. (ed.)(1982). The Romano-British Countryside: Studies in Rural Settlement
and Economy. Part II. Oxford: British Archaeological Reports British Series 103(II),
403-19.
Gerrard C., Wiggins H. and Evans, E M. (2006). Prehistoric Defended Enclosures in
Glamorgan with Recommendations for Fieldwork. GGAT Unpublished Report No.
2006/089.
Green, M.J. (1995).The Celtic World. London: Routledge.
Green, M. and Howell, R. (2000). A Pocket Guide: Celtic Wales. Cardiff: University of Wales
Press.
Greep, S.J. (ed.) (1993). Roman Towns: The Wheeler Inheritance. York: Council for British
Archaeology Research Report 93.
Gwilt, A. (2004). ‘Late Bronze Age Societies (1150-600 BC): Tools and Weapons’, in
Aldhouse-Green, M. and Howell, R. (eds.) (2004). The Gwent County History Vol. 1:
Gwent in Prehistory and Early History. Cardiff: University of Wales Press, 111-39.
Gwilt, A. (2007). ‘Silent Silures? Locating People and Places in the Iron Age of South
Wales’, in Haselgrove, C. and Moore, T. (eds.) (2007). The Later Iron Age in Britain
and Beyond. Oxford: Oxbow Books, 297-328.
Gwilt, A. and Haselgrove, C. (eds.) (1997). Reconstructing Iron Age Societies. Oxford:
Oxbow Monograph 71.
Gwilt, A., Bell, M., Cardy, B., Davis, M., Lodwick, M., Makepeace, G., Northover, P.,
Olding, F., Sharples, N. and Yates, A. (2003). A Research Framework for the
Archaeology of Wales; South-east Wales – Later Prehistory.
http://www.cpat.org.uk/research/selpre.htm.
Hamilton, M.A. (2004). ‘The Bronze Age’, in Aldhouse-Green, M. and Howell, R. (eds.)
(2004). The Gwent County History Vol. 1: Gwent in Prehistory and Early History.
Cardiff: University of Wales Press, 84-110.
Haselgrove, C. and Moore, T. (eds.) (2007). The Later Iron Age in Britain and Beyond.
Oxford: Oxbow Books.
Hayden, B. (2001). ‘Richman, Poorman, Beggarman, Chief: The Dynamics of Social
Inequality’, in Price, T.D. and Feinman, G.M. (2001). Archaeology at the
Millennium: A Sourcebook. New York: Springer, 231-272.
Heichelheim, F.M. (1941). ‘The Text of the ‘Constitutio Antoniniana’ and the Three Other
Decrees of the Emperor Caracalla Contained in Papyrus Gissensis 40’, The Journal of
Egyptian Archaeology. 26, 10-12.
Henig, M. (1992). ‘An Iron Signet Ring’, in Marvell, A.G. and Heywood, B. (1992).
‘Excavations at Neath’, Bulletin of the Board of Celtic Studies, 250-2.
169

Henig, M. (1995). ‘Intaglios’, in Manning, W.H., Price, J. and Webster, J. (1995). Report on
the Excavations at Usk 1965-1976: The Roman Small Finds. Cardiff: University of
Wales Press, 98-9.
Henig, M. (1997). ‘Intaglios’, in Marvell, A.G. and Owen-John, H.S. (1997). Leucarum:
Excavations at the Roman Auxiliary Fort at Loughor, West Glamorgan 1982-84 and
1987-88. London: Britannia Monograph Series No. 12, 394-6.
Henig, M. and Lloyd-Morgan, G. (2000). ‘Finger-Rings, Intaglios and Other Ring-Settings’,
in Evans, E. (2000). The Caerleon Canabae: Excavations in the Civil Settlement
1984-90. London: Britannia Monograph Series No. 16, 322-8.
Higbee, L. (2007). ‘Animal Bone’, in Barber, A., Cox, S. and Hancocks, A. (2007). ‘A Late
Iron Age and Roman Farmstead at RAF St. Athan, Vale of Glamorgan. Evaluation
and Excavation 2002-03’, Archaeologia Cambrensis. 155, 91-4.
Higham, N.J. (1987). ‘Brigantia Revisited’, Northern History. 23, 1-19.
Hill, D. and Jesson, M. (eds.) (1971). The Iron Age and its Hill-forts: Papers Presented to Sir
Mortimer Wheeler on the Occasion of his Eightieth Year, at a Conference Held by the
Southampton University Archaeological Society, 5th-7th March, 1971. Southampton:
Millbrook Press Ltd.
Hill, J.D. (1995). ‘How Should We Understand Iron Age Societies and Hillforts? A
Contextual Study from Southern Britain’, in Hill, J.D. and Cumberpatch, C.G. (eds.)
(1995). Different Iron Ages: Studies on the Iron Age in Temperate Europe. Oxford:
British Archaeological Reports International Series 602, 45-66.
Hill, J.D. (1997). ‘‘The End of One Kind of Body and the Beginning of Another Kind of
Body?’ Toilet Instruments and Romanization’, in Gwilt, A. and Haselgrove, C. (eds.)
(1997). Reconstructing Iron Age Societies. Oxford: Oxbow Monograph 71, 96-107.
Hill, J.D. and Cumberpatch, C.G. (eds.) (1995). Different Iron Ages: Studies on the Iron Age
in Temperate Europe. Oxford: British Archaeological Reports International Series
602.
Hingley, R. (1984). ‘Towards Social Analysis in Archaeology: Celtic Society in the Iron Age
of the Upper Thames Valley (400-0 BC)’, in Cunliffe, B. and Miles, D. (eds.) (1984).
Aspects of the Iron Age in Central Southern Britain. Oxford: Oxford University
Committee for Archaeology Monograph 2, 72-88.
Hingley, R. (2006). ‘The Deposition of Iron Objects during the Later Prehistoric and Roman
Periods: Contextual Analysis and the Significance of Iron’, Britannia. 37, 213-57.
Hogg, A.H.A. (1972). ‘The Size-Distribution of Hillforts in Wales and the Marches’, in
Lynch, F. and Burgess, C. (eds.) (1972). Prehistoric Man in Wales and the West:
essays in honour of Lily F. Chitty. Bath: Adams and Dart, 295-305.
170

Hogg, A.H.A. (1974). ‘The Llantwit Major Villa: A Reconsideration of the Evidence’,
Britannia. 5, 225-50.
Hogg, A.H.A. (1976). ‘Castle Ditches, Llancarfan, Glamorgan’, Archaeologia Cambrensis.
125, 13-39.
Howell, R. (2009). Searching for the Silures: An Iron Age Tribe in South-east Wales. The
History Press.
Howell, R. and Pollard, J. (2000) ‘Caerleon, Lodge Wood Camp (ST 323 914)’, Archaeology
in Wales. 40, 81-3.
Howell, R. and Pollard, J. (2004). ‘The Iron Age: Settlement and Material Culture’, in
Aldhouse-Green, M. and Howell, R. (eds.) (2004). The Gwent County History Vol. 1:
Gwent in Prehistory and Early History. Cardiff: University of Wales Press, 140-59.
Hudson, R. (2000). ‘The Coins’, in Evans, E. (2000). The Caerleon Canabae: Excavations in
the Civil Settlement 1984-90. London: Britannia Monograph Series No. 16, 309-22.
Hughes, G. (1996). The Excavation of a Late Prehistoric and Romano-British Settlement at
Thornwell Farm, Chepstow, Gwent, 1992. Oxford: British Archaeological Reports
British Series 244.
Hunter, F. (2007). ‘Artefacts, Regions and Identities in the Northern British Iron Age’, in
Haselgrove, C. and Moore, T. (eds.) (2007). The Later Iron Age in Britain and
Beyond. Oxford: Oxbow Books, 286-96.
Hutcheson, N. (2007). ‘An Archaeological Investigation of Later Iron Age Norfolk:
Analysing Hoarding Patterns across the Landscape’, in Haselgrove, C. and Moore, T.
(eds.) (2007). The Later Iron Age in Britain and Beyond. Oxford: Oxbow Books, 358-
70.
Insole, P. (1997). ‘Church Farm, Caldicot (ST 4816 8916)’, Archaeology in Wales. 37, 72-4.
Insole, P. (2000). ‘The Archaeological Excavation of a Romano-British Farmstead at Church
Farm, Church Road, Caldicot, Monmouthshire’, Archaeology in Wales. 40, 20-33.
Jackson, D. (1999). ‘Variation in the Size Distribution of Hillforts in the Welsh Marches and
the Implication for Social Organisation’, in Bevan, B. (ed.) (1999). Northern
Exposure: Interpretive Devolution and the Iron Ages in Britain. Leicester: Leicester
University Archaeology Monograph 4, 197-216.
Jarrett, M.G. (ed.) (1969). The Roman Frontier in Wales. 2nd Edition. Cardiff: University of
Wales Press.
Jarrett, M.G. (1994). Early Roman Campaigns in Wales. The Seventh Annual Caerleon
Lecture. Cardiff: National Museum of Wales.
Jarrett, M.G. and Wrathmell, S. (1981). Whitton: An Iron Age and Roman Farmstead in
South Glamorgan. Cardiff: University of Wales Press.
171

Johnson, A.W. and Earle, T.K. (2000). The Evolution of Human Societies: From Foraging
Groups to Agrarian State. Stanford: Stanford University Press.
Jones, C. (2000). The Roman Auxiliary Fort of Leucarum. Swansea: University of Wales
Swansea.
Jones, M. (1989). ‘Agriculture in Roman Britain: The Dynamics of Change’, in Todd, M.
(ed.) (1989a). Research on Roman Britain: 1960-89. London: Britannia Monograph
Series 11, 127-34.
Jones, M.J. (2004), ‘Cities and Urban Life’, in Todd, M. (ed.) (2004). A Companion to
Roman Britain. Oxford: Blackwell , 162-92.
Jones, M.L. (1984). Society and Settlement in Wales and the Marches: 500 BC to AD 1100.
Part I. Oxford: British Archaeological Reports British Series 121(I).
Keresztes, P. (1970). ‘The Constitutio Antoniniana and the Persecutions under Caracalla’,
The American Journal of Philology. 91, 446-59.
Kidder, T.R. (2006). ‘Climate Change and the Archaic to Woodland Transition (3000-2500
Cal B.P.) in the Mississippi River Basin’, American Antiquity. 71(2), 195-231.
Kimes, T., Haselgrove, C., and Hodder, I. (1982). ‘A Method for the Identification of the
Location of Regional Cultural Boundaries’, Journal of Anthropological Archaeology.
1, 113-31.
Kinnes, I.A. (1971). ‘The Animal Bones’, in Jarrett, M.G. and Wrathmell, S. (1981). Whitton:
An Iron Age and Roman Farmstead in South Glamorgan. Cardiff: University of
Wales Press, 232-8.
Knight, J.K. (1996). ‘Late Roman and Post-Roman Caerwent: Some Evidence from
Metalwork’, Archaeologia Cambrensis. 145, 34-66.
Lawler, M. and Marvell, A. G. (1994). ‘Neath, Dwr-y-Felin Road’, Archaeology in Wales. 34,
54.
Lloyd-Fern, S., Boon, G.C. and Sell, S.H. (1996). ‘The Coins’, in Parkhouse, J. and Evans, E.
(eds.) (1996). Excavations in Cowbridge, South Glamorgan, 1977-88. Oxford: British
Archaeological Reports, British Series 245, 180-2.
Lloyd-Morgan, G. (2000). ‘Other Jewellery and Dress Accessories in Gold, Silver and
Copper Alloy’, in Evans, E. (2000). The Caerleon Canabae: Excavations in the Civil
Settlement 1984-90. London: Britannia Monograph Series No. 16, 328-44.
Lloyd-Morgan, G. and Webster, J. (1996). ‘Roman Objects of Copper Alloy’, in Parkhouse, J.
and Evans, E. (eds.) (1996). Excavations in Cowbridge, South Glamorgan, 1977-88.
Oxford: British Archaeological Reports, British Series 245, 182-97.
Locock, M. (2010). ‘The Animal Bone’, in Evans, E. (2010). Church Hill, Penmaen,
Swansea. GGAT Report No. 2010/033, 13-5.
172

Locock, M. and Yates, A. (1999). ‘Greenmoor Arch, Bishton (ST 400 866)’, Archaeology in
Wales. 39, 88; 105.
Lynch, F. and Burgess, C. (eds.) (1972). Prehistoric Man in Wales and the West: Essays in
Honour of Lily F. Chitty. Bath: Adams and Dart.
Lynch, F., Aldhouse-Green, S. and Davies, J.L. (2000). Prehistoric Wales. Stroud: Sutton.
Macdonald, P. (2001). ‘Langstone (ST 38 89)’, Archaeology in Wales. 41, 138-9.
Malinowski, B. (1922). Argonauts of the Western Pacific. New York: Dutton.
Manning, W.H. (1981). Report on the Excavations at Usk 1965-1976: The Fortress
Excavations 1968-1971. Cardiff: University of Wales Press.
Manning, W.H. (1989). Report on the Excavations at Usk 1965-1976: The Fortress
Excavations 1972-1974 and Minor Excavations on the Fortress and Flavian Fort.
Cardiff: University of Wales Press.
Manning, W.H. (1993). Report on the Excavations at Usk, 1965-1976: The Roman Pottery.
Cardiff: University of Wales Press.
Manning, W.H. (2004). ‘The Romans: Conquest and the Army’, in Aldhouse-Green, M. and
Howell, R. (eds.) (2004). The Gwent County History Vol. 1: Gwent in Prehistory and
Early History. Cardiff: University of Wales Press, 178-204.
Manning, W.H., Price, J. and Webster, J. (1995). Report on the Excavations at Usk 1965-
1976: The Roman Small Finds. Cardiff: University of Wales Press.
Mason, D.J.P. (1988). “Prata Legionis’ in Britain’, Britannia. 19, 163-89.
Martin, C. (2003). Iron Age Artefacts in Wales: An Investigation into the Material Culture of
South-east Wales during the Pre-Roman Iron Age. Oxford: British Archaeological
Reports British Series 353.
Marvell, A.G. (1996a). ‘Rogiet (Housing Allocation) H2RO1 (ST 456 829)’, Archaeology in
Wales. 36, 78.
Marvell, A.G. (1996b). ‘Celtic Manor Golf Course, Kemeys Inferior (ST 318 910) Newport’,
Archaeology in Wales. 36, 74-75.
Marvell, A.G. (1996c). ‘Great Pencarn Farm, Coedkernew (ST 283 834) Newport’,
Archaeology in Wales. 36, 75-6.
Marvell, A.G. and Heywood, B. (1992). ‘Excavations at Neath’, Bulletin of the Board of
Celtic Studies. 171-298.
Marvell, A.G. and Maynard, D.J. (1998). ‘Excavations South of the Legionary Fortress at
Usk, Gwent, 1994’, Britannia. 29, 247-67.
Marvell, A.G. and Owen-John, H.S. (1997). Leucarum: Excavations at the Roman Auxiliary
Fort at Loughor, West Glamorgan 1982-84 and 1987-88. London: Britannia
Monograph Series No. 12.
173

Matheson, C. and Cowley, L.F. (1953). ‘Report on Stratified Human and Animal Remains’,
in Nash-Williams, V.E. (1953). ‘The Roman Villa at Llantwit Major, Glamorgan’,
Archaeologia Cambrensis. 102, 158-63.
Mattingly, D. (2006). An Imperial Possession: Britain in the Roman Empire. London:
Penguin Books.
Maynard, D.J. and Marvell A.G. (1993). ‘Cwrt Herbert, Neath, Glamorgan (SS 745 979)’,
Archaeology in Wales. 33, 60.
McKay, B. (1992). ‘The Coins’, in Evans, D.R. and Metcalf, V.M. (1992). Roman Gates
Caerleon. Oxford: Oxbow Monograph 15, 87-96.
Mein, A.G. (1995). ‘Five Lanes, Caerwent (ST4460 9100)’, Archaeology in Wales. 35, 54-5.
Miles, D. (ed.) (1982a). The Romano-British Countryside: Studies in Rural Settlement and
Economy. Part I. Oxford: British Archaeological Reports British Series 103(I).
Miles, D. (ed.) (1982b). The Romano-British Countryside: Studies in Rural Settlement and
Economy. Part II. Oxford: British Archaeological Reports British Series 103(II).
Miles, D. (1989). ‘The Romano-British Countryside’, in Todd, M. (ed.) (1989). Research on
Roman Britain: 1960-89. London: Britannia Monograph Series 11, 115-126.
Millet, M. (1990). The Romanization of Britain. Cambridge: Cambridge University Press.
Moore, T. (2007). ‘Perceiving Communities: Exchange, Landscapes and Social Networks in
the Later Iron Age of Western Britain’, Oxford Journal of Archaeology. 26(1), 79-
102.
Morris, E.L. (1996). ‘Iron Age Artefact Production and Exchange’, in Champion, T.C. and
Collis, J.R. (eds.) (1996). The Iron Age in Britain and Ireland: Recent Trends.
Sheffield: J.R. Collis Publications, 41-66.
Nash-Williams, V.E. (1930). ‘Further Excavations at Caerwent, Monmouthshire, 1923-5’,
Archaeologia. 80, 229-88.
Nash-Williams, V.E. (1933). ‘An Early Iron Age Hill-fort at Llanmelin, near Caerwent,
Monmouthshire’, Archaeologia Cambrensis. 88, 237-315.
Nash-Williams, V. E. (1939a). ‘An Early Iron Age Coastal Camp at Sudbrook, Near the
Severn Tunnel, Monmouthshire’, Archaeologia Cambrensis. 94, 42-79.
Nash-Williams, V.E. (1939b). ‘A New Roman Mining Site at Lower Machen,
Monmouthshire’, Archaeologica Cambrensis. 94, 108-10.
Nash-Williams, V.E. (1950a). ‘The Roman Stations at Neath (Glamorgan) and Caer Gai
(Mer.)’, Bulletin of the Board of Celtic Studies. 13, 239-45.
Nash-Williams, V.E. (1950b). ‘The Roman Villa at Llanfrynach, Brecknockshire’, Bulletin of
the Board of Celtic Studies. 13, 105-8.
Nash-Williams, V.E. (1952). ‘The Roman Station at Neath, Further Discoveries’, Bulletin of
the Board of Celtic Studies. 14, 76-9.
174

Nash-Williams, V.E. (1953). ‘The Roman Villa at Llantwit Major, Glamorgan’,


Archaeologia Cambrensis. 102, 89-163.
Nash-Williams, V.E. (1954a). The Roman Frontier in Wales. Cardiff: University of Wales
Press.
Nash-Williams, V.E. (1954b). ‘The Forum-and-basilica and Public Baths of the Roman Town
of Venta Silurum at Caerwent in Monmouthshire’, Bulletin of the Board of Celtic
Studies. 15, 159-68.
Nash-Williams, V.E. (1956). ‘The Roman Town of Venta Silurum and Its Defences’, in
Swoboda, E. (ed.) (1956). Carnuntina: Ergebnisse der Forschung über die
Grenzprovinzen des römischen Reiches Vorträge beim internationalen Kongreß der
Altertumsforscher, 1955. Graz: H. Böhlaus Nachf., 100-16.
Newman, R. (1990). ‘Excavations of a Romano-British Building at Dan-y-Graig, Porthcawl,
Mid Glamorgan (SS 8407 7805)’, Bulletin of the Board of Celtic Studies. 37, 247-80.
Olding, F. (2000). The Prehistoric Landscapes of the Eastern Black Mountains. Oxford:
British Archaeological Reports British Series 297.
Ordnance Survey. (1962). Map of Southern Britain in the Iron Age, 1:625,000. Chessington,
Surrey: Ordnance Survey.
Owen-John, H.S. (1988a). ‘A Hill-slope Enclosure in Coed y Cymdda, Near Wenvoe, South
Glamorgan’, Archaeologia Cambrensis. 137, 43-98.
Owen-John, H.S. (1988b). ‘Llandough: The Rescue Excavation of a Multi-Period Site Near
Cardiff, South Glamorgan’, in Robinson, D.M. (ed.) (1988). Biglis, Caldicot and
Llandough: Three Late Iron Age and Romano-British Sites in South-East Wales.
Oxford: British Archaeological Reports British Series 188, 123-77.
Owens, E.J. (1997). ‘Zosimus, the Roman Empire, and the End of Roman Britain’, in Deroux,
C. (ed.) (1997). Studies in Latin Literature and Roman History VIII. Bruxelles:
Latomus Revue d'Études Latines, Collection Latomus 239, 478-504.
Parkhouse, J. (1988). ‘Excavations at Biglis, South Glamorgan’, in Robinson, D.M. (ed.)
(1988). Biglis, Caldicot and Llandough: Three Late Iron Age and Romano-British
Sites in South-East Wales. Oxford: British Archaeological Reports British Series 188,
3-64.
Parkhouse, J. and Evans, E. (eds.) (1996). Excavations in Cowbridge, South Glamorgan,
1977-88. Oxford: British Archaeological Reports, British Series 245.
Percival, J. (1989). ‘The Villa Economy: Problems and Perspectives’, in Branigan, K. and
Miles, D. (eds.) (1989). The Economies of Romano-British Villas. Sheffield:
Department of Archaeology and Prehistory, 5-13.
Perring, D. (2002). The Roman House in Britain. London: Routledge.
175

Peterson, R., Pollard, J. and Morris, E. (2006). ‘Prehistoric Pottery’, in Pollard, J., Howell, R.,
Chadwick, A. and Leaver, A. (2006). Lodge Hill Camp, Caerleon and the Hillforts of
Gwent. Oxford: British Archaeological Reports British Series 407, 33-37.
Plog, F. and Upham, S. (1983). ‘The Analysis of Prehistoric Political Organization’, in
Tooker, E. (ed.) (1983). The Development of Political Organization in Native North
America. Seattle: University of Washington Press, 199-213.
Pollard, J., Howell, R., Chadwick, A. and Leaver, A. (2006). Lodge Hill Camp, Caerleon and
the Hillforts of Gwent. Oxford: British Archaeological Reports British Series 407.
Ponsford, M.W. (2006). ‘Archaeological Excavations at Castle Street Car Park (1999) and
Ewers’ Garden (1968-72), Abergavenny, with a Reconsideration of the Early History
of the Roman Fort’, Archaeology in Wales. 46, 49-78.
Price, T.D. and Feinman, G.M. (2001). Archaeology at the Millennium: A Sourcebook. New
York: Springer.
Probert, L.A. (1976). ‘Twyn-y-Gaer hill-fort, Gwent: An Interim Assessment’, in Boon, G.
and Lewis, J. (eds.) (1976). Welsh Antiquity. Cardiff: National Museum of Wales,
105-19.
RCAHMW. (1976a). An Inventory of the Ancient Monuments in Glamorgan, Vol. I: Pre-
Norman. Pt. I: The Stone and Bronze Ages. Cardiff: HMSO.
RCAHMW. (1976b). An Inventory of the Ancient Monuments in Glamorgan, Vol. I: Pre-
Norman. Pt. II: The Iron Age and the Roman Occupation. Cardiff: HMSO.
RCAHMW. (1986). An Inventory of the Ancient Monuments in Brecknock (Brycheiniog):
Vol. I: The Prehistoric and Roman Monuments. Pt. II: Hill-forts and Roman
Remains. London: HMSO.
Richardson, R.E. (1992). ‘Iron Age and Romano-British Farmland in the Herefordshire Area’,
Transactions of the Woolhope Naturalists’ Field Club. 47, 144-161.
Richmond, I.A. (1958). ‘Roman Britain in 1957: I. Sites Explored’, The Journal of Roman
Studies. 48, 130-49.
Rippon, S. (1996). Gwent Levels: The Evolution of a Wetland Landscape. York: Council for
British Archaeology Research Report 105.
Rivet, A.L.F. (1969). The Roman Villa in Britain. London: Routledge and Kegan Paul.
Rivet, A.L.F. (1970). ‘The British Section of the Antonine Itinerary’, Britannia. 1, 34-82.
Rivet, A.L.F. and Smith, C. (1979). The Place-Names of Roman Britain. London: Batsford.
Robinson, D. M. (1980). Cowbridge: The Archaeology and Topography of a Small Market
Town in the Vale of Glamorgan. Swansea: The Glamorgan-Gwent Archaeological
Trust Limited Town Survey No. 1.
176

Robinson, D.M. (ed.) (1988). Biglis, Caldicot and Llandough: Three Late Iron Age and
Romano-British Sites in South-East Wales. Oxford: British Archaeological Reports
British Series 188.
Salway, P. (1981). Roman Britain. Oxford: Oxford University Press.
Savory, H. N. (1952). ‘List of Hill-forts and Other Earthworks in Wales: III. Brecknockshire’,
Bulletin of the Board of Celtic Studies. 14, 69-75.
Savory, H.N. (1954a). ‘The Excavation of an Early Iron Age Fortified Settlement on Mynydd
Bychan, Llysworney (Glam.), 1949-50: Part I’, Archaeologia Cambrensis. 103, 85-
108.
Savory, H.N. (1954b). ‘Early Iron Age Discoveries on Merthyr Mawr Warren (Glam.)’,
Bulletin of the Board of Celtic Studies. 16(1), 53-54.
Savory, H. N. (1955). ‘Prehistoric Brecknock’, Brycheiniog. 1, 79-127.
Savory, H. N. (1961). ‘Twyn Lechfaen hill-fort excavations 1959’, Bulletin of the Board of
Celtic Studies. 19, 173-176.
Savory, H.N. (ed.) (1984). Glamorgan County History, Vol. II: Early Glamorgan: Pre-
History and Early History. Cardiff: Glamorgan County History Trust.
Savory, H.N. (1988). ‘Prehistoric Pottery (Figs. 17-18, Nos. 1-15)’, in Owen-John, H. (1988).
‘A Hill-slope Enclosure in Coed y Cymdda, Near Wenvoe, South Glamorgan’,
Archaeologia Cambrensis. 137, 87-93.
Scott, E. (1993). A Gazetteer of Roman Villas in Britain. Leicester: Leicester Archaeology
Monographs 1.
Scott, I. (1992). ‘The Objects of Iron’, in Evans, D.R. and Metcalf, V.M. (1992). Roman
Gates Caerleon. Oxford: Oxbow Monograph 15, 164-74.
Scott, I. (2000). ‘Objects of Iron’, in Evans, E. (2000). The Caerleon Canabae: Excavations
in the Civil Settlement 1984-90. London: Britannia Monograph Series No. 16, 386-
407.
Seager Smith, R.H. (2000). ‘The Mortaria’, in Evans, E. (2000). The Caerleon Canabae:
Excavations in the Civil Settlement 1984-90. London: Britannia Monograph Series
No. 16, 264-83.
Sealey, P.R. (2004). The Boudican Revolt Against Rome. 2nd Edition. Princes Risborough:
Shire.
Sellwood, L. (1984). ‘Tribal Boundaries Viewed from the Perspective of Numismatic
Evidence’, in Cunliffe, B. and Miles, D. (1984). Aspects of the Iron Age in Central
Southern Britain. Oxford: Oxford University Committee for Archaeology.
Monograph 2, 191-204.
Sherman, A. and Evans E. (2004). Roman Roads in Southeast Wales. GGAT Report No.
2004/073.
177

Shoesmith, R. (1991). Excavations at Chepstow 1973-1974. Bangor: Cambrian


Archaeological Association.
Silvester, R.J. (2007). Defended Enclosures in Brecknock. CPAT Report 859.
Silvester, R.J. and Toller, H. (2010). ‘Roman Roads in Wales and the Marches’, in Burnham,
B.C. and Davies J.L. (eds.) (2010). Roman Frontiers in Wales and the Marches.
Aberystwyth: RCAHMW, 91-7.
Simpson, G. (1963). ‘Caerleon and the Roman Forts in Wales in the Second Century A.D.’,
Archaeologia Cambrensis. 112, 13-76.
Smith, J.T. (1978). ‘Villas as a Key to Social Structure’, in Todd, M. (ed.) (1978). Studies in
the Romano-British Villa. Leicester: Leicester University Press, 149-85.
Smith, J.T. (1997). Roman Villas: A Study in Social Structure. London: Routledge.
Southern, P. (2004). ‘The Army in Late Roman Britain’, in Todd, M. (ed.) (2004). A
Companion to Roman Britain. Oxford: Blackwell, 393-408.
Spencer, B. (1983). ‘Limestone-Tempered Pottery from South Wales in the Late Iron Age and
Early Roman Period’, Bulletin of the Board of Celtic Studies. 30, 405-19.
Spencer, B. (1988). ‘The Later Roman Grey-Ware Industry in South Wales’, in Vyner, B.E.
and Allen, D.W.H. (1988). ‘A Romano-British Settlement at Caldicot, Gwent’, in
Robinson, D.M. (ed.) (1988). Biglis, Caldicot and Llandough: Three Late Iron Age
and Romano-British Sites in South-East Wales. Oxford: British Archaeological
Reports British Series 188, 114-8.
Spratt, D. and Burgess, C. (1985). Upland Settlement in Britain: The Second Millennium B.C.
and After. British Archaeological Reports British Series 143.
Stanford, S.C. (1971). ‘Invention, Adoption and Imposition – The Evidence of the Hill-forts’,
in Hill, D. and Jesson, M. (eds.) (1971). The Iron Age and its Hill-forts: Papers
Presented to Sir Mortimer Wheeler on the Occasion of his Eightieth Year, at a
Conference Held by the Southampton University Archaeological Society, 5 th-7th
March, 1971. Southampton: Millbrook Press Ltd, 41-52.
Stein, G.J. (2002). ‘From Passive Periphery to Active Agents: Emerging Perspectives in the
Archaeology of Interregional Interactions: Archaeology Division Distinguished
Lecture AAA Annual Meeting, Philadelphia, December 5, 1998’, American
Anthropologist. New Series, 104(3), 903-16.
Stevens, C.E. (1966). ‘The Social and Economic Aspects of Rural Settlement’, in Thomas, C.
(ed.) (1966). Rural Settlement in Roman Britain: Papers Given at a C.B.A.
Conference Held at St. Hugh’s College Oxford January 1 to 3, 1965. London:
Council for British Archaeology Research Report 7, 108-28.
178

Surman, J. (1981). ‘XI Metalwork: 5. Iron Ore and Slag’, in Jarrett, M.G. and Wrathmell, S.
(1981). Whitton: An Iron Age and Roman Farmstead in South Glamorgan. Cardiff:
University of Wales Press, 204-10.
Swoboda, E. (ed.) (1956). Carnuntina: Ergebnisse der Forschung über die Grenzprovinzen
des römischen Reiches Vorträge beim internationalen Kongreß der
Altertumsforscher, 1955. Graz: H. Böhlaus Nachf.
Syme, R. (1988). ‘The Subjugation of Mountain Zones’, in Birley, A.R. (ed.) (1988). Roman
Papers V. Oxford: Clarendon Press, 648-60.
Symons, S. (2009). Fortresses and Treasures of Roman Wales. Derby: Breedon Books.
Tainter, J.A. (1980). ‘Behavior and Status in a Middle Woodland Mortuary Population from
the Illinois Valley’, American Antiquity. 45(2), 308-13.
Tainter, J.A. and Cordy, R.H. (1977). ‘An Archaeological Analysis of Social Ranking and
Residence Groups in Prehistoric Hawaii’, World Archaeology. 9(1): Island
Archaeology, 95-112.
Thomas, B. (2000). Communication and Lines of Sight: An Interpretation of the Intervisibility
of Hillforts and Related Earthworks of Usk and Lower Wye River Valleys. MA
Dissertation in Celto-Roman Studies: University of Wales, Newport.
Thomas, C. (ed.) (1966). Rural Settlement in Roman Britain: Papers Given at a C.B.A.
Conference Held at St. Hugh’s College Oxford January 1 to 3, 1965. London:
Council for British Archaeology Research Report 7.
Thomas, H. (1958). ‘Roman Discoveries in the Vale of Glamorgan’, Bulletin of the Board of
Celtic Studies. 17, 293-6.
Todd, M. (ed.) (1978). Studies in the Romano-British Villa. Leicester: Leicester University
Press.
Todd, M. (ed.) (1989a). Research on Roman Britain: 1960-89. London: Britannia Monograph
Series 11.
Todd, M. (1989b). ‘Villa and Fundus’, in Branigan, K. and Miles, D. (eds.) (1989). The
Economies of Romano-British Villas. Sheffield: Department of Archaeology and
Prehistory, 14-20.
Todd, M. (ed.) (2004). A Companion to Roman Britain. Oxford: Blackwell.
Tooker, E. (ed.) (1983). The Development of Political Organization in Native North America.
Seattle: University of Washington Press.
Turner, R.C. and Scaife, R.G. (eds.) (1995). Bog Babies: New Discoveries and New
Perspectives. London: British Museum Press.
Ucko, P.J., Tringham, R. and Dimbleby, G.W. (eds.) (1972). Man, Settlement and Urbanism.
London: Duckworth.
179

Van Ardsell, R. (1994). The Coinage of the Dobunni: Money Supply and Coin Circulation in
Dobunnic Territory. Studies in Celtic Coinage, 1. Oxford: Oxford University
Committee for Archaeology, Monograph 38.
Vaughan-Williams, A. (2007). ‘Charred Plant Remains’, in Barber, A., Cox, S. and Hancocks,
A. (2007). ‘A Late Iron Age and Roman Farmstead at RAF St. Athan, Vale of
Glamorgan. Evaluation and Excavation 2002-03’, Archaeologia Cambrensis. 155,
94-7.
Vyner, B.E. and Allen, D.W.H. (1988). ‘A Romano-British Settlement at Caldicot, Gwent’, in
Robinson, D.M. (ed.) (1988). Biglis, Caldicot and Llandough: Three Late Iron Age
and Romano-British Sites in South-East Wales. Oxford: British Archaeological
Reports British Series 188, 65-122.
Vyner, B.E. and Evans, G.C. (1978). ‘The Excavation of a Roman Pottery Kiln at Llanedeyrn,
Cardiff’, in Boon, G.C. (1978). Monographs and Collections: Relating to
Excavations Financed by H.M. Department of the Environment in Wales. I. Roman
Sites. Cardiff: Cambrian Archaeological Association, 120-9.
Vyner, B.E. and White, J. (1985). ‘Romano-British Fields and Farms in the Vale of
Glamorgan’, Britannia. 16, 245-6.
Wacher, J. S. (1978). Roman Britain. London: Dent.
Wacher, J.S. (1989). ‘Cities from the Second to Fourth Centuries’, in Todd, M. (ed.) (1989).
Research on Roman Britain: 1960-89. London: Britannia Monograph Series 11, 91-
114.
Wacher, J. S. (1995). The Towns of Roman Britain. 2nd Edition. London: BCA.
Walters, B. and Walters, M. (1989) ‘Ariconium Military Site Excavation Interim Report’,
Dean Archaeology. 2, 33-45.
Warner, P. (1996). The Origins of Suffolk. Manchester: Manchester University Press.
Webster, G. (1993). The Roman Invasion of Britain. London: Batsford.
Webster, G. (2003). Rome Against Caratacus: The Roman Campaigns in Britain AD 48-58.
London: Routledge.
Webster, J. (1981). ‘XI Metalwork: 1. The Bronzes’, in Jarrett, M.G. and Wrathmell, S.
(1981). Whitton: An Iron Age and Roman Farmstead in South Glamorgan. Cardiff:
University of Wales Press, 163-88.
Webster, J. (1988). ‘Bronze and Silver Objects’, in Parkhouse, J. (1988). ‘Excavations at
Biglis, South Glamorgan’, in Robinson, D.M. (ed.) (1988). Biglis, Caldicot and
Llandough: Three Late Iron Age and Romano-British Sites in South-East Wales.
Oxford: British Archaeological Reports British Series 188, 53-6.
Webster, J. (1992). ‘The Objects of Bronze’, in Evans, D.R. and Metcalf, V.M. (1992).
Roman Gates Caerleon. Oxford: Oxbow Monograph 15, 103-63.
180

Webster, J. (1995). ‘Brooches’, in Manning, W.H., Price, J. and Webster, J. (1995). Report on
the Excavations at Usk 1965-1976: The Roman Small Finds. Cardiff: University of
Wales Press, 60-96.
Webster, P.V. (1984). ‘The Roman Period’, in Savory, H.N. (ed.) (1984). Glamorgan County
History, Vol. II: Early Glamorgan: Pre-History and Early History. Cardiff:
Glamorgan County History Trust, 277-313.
Webster, P. (1989) ‘A Natural Channel at Magor Pill’, in Whittle, A.W.R. (1989). ‘Two Later
Bronze Age Occupations and an Iron Age Channel on the Gwent Foreshore’, Bulletin
of the Board of Celtic Studies. 36, 217-21.
Webster, P. V. (1993). ‘The Post-Fortress Coarsewares’, in Manning, W.H. (1993). Report on
the Excavations at Usk, 1965-1976: The Roman Pottery. Cardiff: University of Wales
Press, 227-361.
Webster, P. (2003). ‘An Early Fort at Caerwent? A Review of the Evidence’, in Wilson, P.
(ed.) (2003). The Archaeology of Roman Towns: Studies in Honour of John S.
Wacher. Oxford: Oxbow, 214-20.
Webster, P. (2010). ‘The Roman Pottery’, in Evans, E. (2010). Church Hill, Penmaen,
Swansea. GGAT Report No. 2010/033, 7-9.
Weissner, P. (1982). ‘Beyond Willow Smoke and Dog’s Tails: A Comment on Binford’s
Analysis of Hunter-gatherer Settlement Systems’, American Antiquity. 47, 171-8.
Wheeler, R.E.M. (1921). ‘A Roman Fortified House near Cardiff’, The Journal of Roman
Studies. 11, 67-85.
Whittle, A.W.R. (1989). ‘Two Later Bronze Age Occupations and an Iron Age Channel on
the Gwent Foreshore’, Bulletin of the Board of Celtic Studies. 36, 200-23.
Wiggins, H. (2006). Southeast Wales Prehistoric Defended Enclosures: Prehistoric Defended
Enclosures in Gwent. GGAT Unpublished Report No.2006/021.
Wigley, A. (2007). ‘Rooted to the Spot: The ‘Smaller Enclosures’ of the Later First
Millennium BC in the Central Welsh Marches’, in Haselgrove, C. and Moore, T.
(eds.) (2007). The Later Iron Age in Britain and Beyond. Oxford: Oxbow Books, 173-
89.
Williams, G. (1990). ‘Recent Work on Rural Settlement in South-west Wales’, in Burnham,
B.C. and Davies, J.L. (eds.) (1991). Conquest, Co-existence and Change: Recent
Work in Roman Wales. Lampeter: Trivium 25, 112-122.
Willis, S. (2007). ‘Sea, Coast, Estuary, Land and Culture in Iron Age Britain’, in Haselgrove,
C. and Moore, T. (eds.) (2007) The Later Iron Age in Britain and Beyond. Oxford:
Oxbow Books, 107-129.
Wilson, D.R. (1970). ‘Roman Britain in 1969: I. Sites Explored’, Britannia. 1, 269-305.
181

Wilson, P. (ed.) (2003). The Archaeology of Roman Towns: Studies in Honour of John S.
Wacher. Oxford: Oxbow.
Woodward, A. (1996). ‘The Prehistoric and Native Pottery’, in Hughes, G. (1996). The
Excavation of a Late Prehistoric and Romano-British Settlement at Thornwell Farm,
Chepstow, Gwent, 1992. Oxford: British Archaeological Reports British Series 244,
36-45.
Worrell, S. (2007). ‘Detecting the Later Iron Age: A View from the Portable Antiquities
Scheme’, in Haselgrove, C. and Moore, T. (eds.) (2007). The Later Iron Age in
Britain and Beyond. Oxford: Oxbow Books, 371-88.
Young, T. (2001). ‘Ely Roman villa (ST 147 761)’, Archaeology in Wales. 41, 130-2.
Young, T. (2003). ‘Caergwanaf, Pont-y-Clun, Rhondda Cynon Taff (ST 04 80)’, Morgannwg.
47, 67-9.
Young, T. (2010) ‘Caergwanaf (NGR: ST 044 807; NPRN 403804)’, in Burnham, B.C. and
Davies J.L. (eds.) (2010). Roman Frontiers in Wales and the Marches. Aberystwyth:
RCAHMW, 214-216.
Young, T. and Guest, P. (2004). ‘Caergwanaf, Pont-y-Clun, Rhondda Cynon Taff (ST 04
80)’, Morgannwg. 48, 73-4.
Young, T. and Macdonald, P. (1997). ‘Miskin, School Road (ST 0520 8144)’, Archaeology in
Wales. 37, 79.
Zienkiewicz, D. (1983). ‘Great Bulmore (ST 360 915)’, Archaeology in Wales. 23, 47-9.
Zienkiewicz, D. (1984). ‘Great Bulmore (ST 360 915)’, Archaeology in Wales. 24, 57-8.
Zienkiewicz, D. (1987). Roman Gems From Caerleon. Cardiff: National Museum of Wales.
Zienkiewicz, D. (1993). ‘Excavations in the Scamnum Tribunorum at Caerleon: the Legionary
Museum site 1983-5’, Britannia. 24, 27-140.

You might also like