Escolar Documentos
Profissional Documentos
Cultura Documentos
Artist's note
Readers may care to note that the original paintings from which
the colour plates in th is book were prepared are available for
pr ivate sale.All reproduction copyright whatsoever is retained by
the Publishers .All enquiries should be addressed to :
A ll rights reserved .Apart from any fair dealing fo r the purpose of private study.
research. criticism o r review. as per mitt ed under the C o pyright. D esigns and Patents
Act.
1988. no part of this publication may be reprodu ced. stored in a re trieval system.
or transm itte d in any for m or by any means. electronic. electri cal. chemical. mechanical.
op tical. photocopying. reco rding or otherwise. w ithout the pr ior wr itten permission of
the copyright owner. Enquir ies should be addressed to the Publisher s.
Measurements
06 07 08 09 10
10 9 8 7 6 S 4 3 2 1
A CIP catalogue recor d fo r this book is available from the Briti sh Library.
O sprey D irect.
W estminste r. MD 21 157
Email: info@os preydirect.co m
www.ospreyp ub lishing.com
Distanc es. ranges . and dimen sions are mostly given in metric.To
convert t hese figures to Imperial values. th e following co nvers io n
formulae are provided :
I metre (m)
1.0936 yards
I kilometre (km) 0.6214 miles
Contents
Introduction
Chronology
Locatio n
14
H ill -forts: form and function Building t he hill-forts Th e design of bro chs
24
30
Maiden Castle The development of t he fortificati ons The settle ment T he econom ic and polit ical centre
42
Aftermath
52
56
Bibliography
62
Glossary
63
Index
64
Introduction
Unfor tunately neither Caesar, Strabo, nor an y other classical writer had mu ch
to say of the Celts who lived in Britain. Instead we have to rely on linguistic or
archaeological evidence. Elements of what was once a pan- European Celtic
lang uage still survive on the 'Celtic fringe' of Europe, wher e Welsh , Irish and
Scots Gaelic all share th e same linguis tic roo ts. It was th e 18th-century Welsh
scho lar Edward Lh uyd who first iden tified thi s Celtic lin guistic tradition, and
who first esta blishe d th e existence of a finite Celtic culture. Indeed it was Lh uyd
wh o resurr ected th e word 'Ce ltic', which co inci ded with th e eme rging evidence
produ ced through th e new science of archaeology.
From th e mid-1 9th century onwards, archa eologists began to un earth artefacts
th at were att ributed to th e Celts - as defined by Lhuyd. Two sites in particular
came to be associated with particular ph ases of Celtic cultura l developmen t: the
Late Bron ze Age site at Hallstatt in Austria, and th e Iron Age religious site at La
Tene in Switzerland, whe re Celtic votive offerings were recove red from th e waters
of Lake Neufcha tel. Subsequently both sites gave th eir nam e to cultura l ph ases
into wh ich all material evide nce att ributed to th e Celts was placed. However, th is
was not th e wh ole sto ry. As evide nce of th e earlier Hallstatt ph ase can be found
in some part s of Euro pe and not in others, archa eologists presum ed th at th e
Celtic sph ere of influence expanded during th e Iron Age to cover all of France,
Spain, Britain , Ireland, Switzerland, Austria and part s of Italy and a swathe of
Eastern Euro pean countries traversed by th e River Danube. However, recent
archaeological evide nce ha s sho wn th at even dur ing th e Late Bron ze Age th e
indige no us (and presumably non- Celtic) peoples of Britain ma intained cultura l
an d com mercial links with th e Celtic peopl es on th e Euro pea n mainl an d. Wh ile
artefacts help exp lain links between different region s in th e Celtic world, th ey do
litt le to help us understand how th e indi gen ou s populat ion of Britain inte racted
with th e Celtic inco me rs from th e Late Bron ze Age on wards. In other words, the
division between th e Celts and th ose th at came before th em is somew hat blurr ed .
The on ly clear archaeological evide nce we can find is th e rem ains of pre-Celtic
and post-Celtic sett lemen ts, religiou s centres and defensiv e works.
While in recent years archa eologists and historians have becom e mor e hesitant
in defining exactly who or what th e Celts were, th ere is at least some agreeme n t
on whe n th ey lived in Britain. The Celts were essentially an Iron Age culture, a
term first devised by Dani sh an tiquarians to help th em catalogue the ir mu seum
co llections . Today th e British Iron Age is used as shortha nd for th e period from
aro und 700 BC (whe n the produ ct ion of iron first took place in Britain) until just
after th e Rom an invasion of southe rn England in AD 43. Even th ese paramete rs
are far from fixed : for exam ple, it is gene rally held th at in Scotland, where th e
Rom an pene t ratio n was limited , th e Iron Age co n tin ued un til the 5th cen tury AD.
For th e purposes of th is book we will limi t the study of Celtic fortifications to an
even briefer period - from aro und sao BC un til a few decad es afte r the first Rom an
in vasion of Britain , c. AD 80, wh en most of main land Britain had fallen under
Rom an co n trol. Wh ile th e Celtic period co n tin ued mu ch later in Scotland, Wales
and Irelan d, the nature of 'Dark Age' Celtic fo rtificatio ns is a su bject wor th y o f
ano the r study, and will th erefore rem ain outside th e bounda ries o f this book .
Sim ilarly, the aut hor has avoi ded cove rage of Irelan d, a lan d with its ow n Celtic
tale to tell, whic h is o nce again deserving of its own book.
No part of the survivi ng ph ysical rem ains o f Iron Age Celtic Britain is more
spectac ular that th e fortificatio ns that still dot th e British lan dscape, from
th e great earthe n hill- fort s o f so ut hern Eng land (such as Maid en Castle) to th e
im posing sto ne-built broch s of no rth ern Scotland . While none of th ese were
stro ng enough to keep out a determined att ac k by th e Roman war ma chi ne, th ey
still domi na ted th e landscape, and to th e pre-Roma n peopl e o f Celtic Britain th ey
wo uld have represented th e ultim ate state me n t in politi cal, mi litary and social
po wer. This book will provide a brief survey of th e typ es of fortificatio ns used, and
will show how th ey develop ed ove r time and how th ey change d from region to
region . It is also hoped th at , in some way, it will explain how th ese great
fortifi cati ons were defended, in th eir role as th e last bastions of Celtic civilizatio n
in Britain .
Chronology
Note: all th e dates provided below for th e gene ral ph ases of archaeological
periods are open to interpretation . They represen t th e broad conse ns us of
archaeologists and h istorian s.
2100
2000
1750
1323
1000
800
700
BC
BC
BC
BC
BC
BC
BC
600 BC
500 BC
400 BC
200 BC
150 BC
100 BC
60 BC
AD
AD
AD
AD
AD
AD
AD
43
47
49
60
82
122
410
A D 500
\ -
Altho ugh hill-forts and brochs are the most commonly found type of Celtic
fortification in Brita in, ot her types of fortified sites ex isted in ta ndem wit h
them . The scope of th is boo k precludes a study of th e less defensible of these,
suc h as lake Villages and cran nogs, largely becau se th ese struc tures lacked any
obv ious mean s of defen ce save th eir encircleme n t by water. Altho ugh th e lake
Village at Glasto n bury, Som erset was su rro un ded by a wood en palisade, it was
prob abl y not designed to keep in trude rs out but rather as a protecti ve barri er to
prevent ch ildren and anim als fallin g into the sur roun ding lake. Sim ilarl y wh ile
buildings built over th e wat er, such as Oakbank Cran nog in Scotland 's Loch
Tay, ma y well have been defensible for a sho rt time, it lacked the protection to
keep determ in ed attackers at bay. For the purposes of thi s study we sha ll
conce n trate on th e Celtic sites th at appear to have been desig ned with defence
in min d .
Celtic fortifi cations come in a variety of types, altho ugh with th e possib le
exception of broc hs they all share certain characteristics. Over th e years
archa eologists have developed termin ology that he lps them classify the intricate
systems of ditch es and banks th ey encoun ter. These are often augmented with
more Wide ly understood fortification term s to help explain how th ese features
were supposed to work. For exam ple, many arch aeologists use th e word s 'ban k'
and 'ra m part' interchangeably, but to be more accurate th e fortification term
'ra m part' sho uld really on ly be appli ed to th e innerm ost bank surro unding th e
fort's enclosure. If a fort is enclosed by a single circuit of bank and ditch, it is
described as a 'univa llate' fort . More com plex fort ificati on s are described as
'bivallate' (if th ey have two such lines of defenc e), 'trivallate' (if th ey have three
circuits of bank and dit ch ), and 'm ultivallate' (if th e fort is defended by more
than three lines of defence). If th e banks are set close
together th ey are described as 'com pact', whil e th e
opposite are regarded as 'dispersed' .
Given th e range of different types of fortificati ons ,
most fall into four general categories. The first are the
pure hill-forts, whose defences are placed to make
best use of th e terrain. The perim eter of the enclosure
follows th e co nto urs of the hilltop on whic h the fort
is built, and consequent ly th ese fortificatio ns are
rarely circular or even regular. Instead th e lines bend
in accordance with th e shape of th e hill. The result
is an irregularly shaped defensive position, but one
that makes best possib le use of the lie of the land. A
variation on th is is th e headland or promontory forts
found on rocky spurs of coastline wh ere th e site
could be made defensible with relatively little work .
In sites such as Burghead in Moray, Scotland, or Rame
Head in Cornwall these fortification s close off th e
landward side of a headland by mean s of a defen sive
bank and ditch system similar to that found in hillfort s. The only difference is that on th e rem aining
sides of th e defended position th e sea itself provides
a natural barrier to attackers. It is wo rth noting that
forts of thi s kind are not always found on th e coast.
On rare occasion s wh ere the terrain provid es a similar
adva ntage far from the sea, such as at the confluence of two large rivers, a similar
position could also be established. An example of this is th e Iron Age sett lement
at Dyke Hills in Oxfordsh ire, where th e River Thame meets the River Thames near
mo de rn-day Dorcheste r-on-Thames. There a bivallate defens ive line was created to
pro tect th e sett lement, while the two rivers protected th e remaining three sides of
the site.
A variant on th e h ill-fort is what has unsat isfactorily been described as the
plateau fort, or valley fort. These are similar to the more common hill-forts, but
were buil t on sites th at possessed no defen sive adva ntages such as slopes or even
rivers. Inst ead th ey had to rely on their own man -made defences to keep attac kers
at bay. The on ly real adva ntage of this type of fortificat ion seems to be th at in
ma ny cases th e enclosure inclu ded a natu ral spring, so tha t, unlike man y hillforts, th e defenders had access to water and thus in th eory could bett er with stand
a siege. Ofte n these appea r to have been built in areas of goo d farml and, in valleys
or on broad ridges wh ere no more obviously defensible feature was available. An
exam ple of thi s type of early Iron Age fort is Rain sborough in Northa mptonsh ire,
wh ich was excavated dur ing th e 1960 s. There th e fortification stoo d on th e edge
of a plateau, whe re th e ground th en fell away gen tly int o th e Che rwell Valley
below. It enclosed an area of some 2.5 hectares, and altho ugh its defen ces were
uni vallat e, th e archaeologists un covered traces of an outer bank th at had been
filled in at some stage during th e fort's occupa tion. The occupan ts may have come
to regret th e siting of th eir fort : archaeologists also un covered evide nce th at th e
gateway had been destroyed by fire, and a skeleto n was found amid th e burned
ruins of a guardho use. The inference is of course th at the fort was attacked and
captured probably at some point in th e 3rd cen tury BC
Ano ther distin ct group of forts com prises th ose th at were clearly built as nondefensible enclosures, probably to house livestock or to provide a seasona l hom e
for a farming comm unity. Sites of this kind were ofte n built on th e sides of a hill
or in a flat area, and often invo lved m ultipl e enclosures encircling a central area.
Wh ile not primarily designed as a defen sive position, th ese could serve as an
emergency refuge in tim e of dan ger. Examples of forts of th is type include
Lorde ns haws in Northumberland and Clovelly Dykes in Devon , and th ey are
generally locat ed in either the north of Eng land, the south-west, or in th e western
. "
Hill-forts
Major areas of hill-fortconcentration
50miles
lOOkm
(
, ~
A.'~ .~
v';' A'<
-: .;
"
\) ,J"~ ~ /'
Burghead
~~ .
~
~a!:,O 'Noth
COTLANO;:- ~~
-,',~f;
:;:;;r';/S~
ca~eh~~un ~own
~~ _
e
/..
..
~'
Caterthun
"--,
. Barry Hill
. Ounsinane/
d ::;""" 'b" ,~
0V;/;~!
~,.
.,~
r,( '
tJ
lj
IRE L:AND
0"
NOR T H
- , Woden Law
SEA
.U
IR I SH SEA
.r~J,
. Almondbury
J)\
~,.....'
jr
'I
ENGLAND
~~
Caer Caradoc
. Crott Ambrey
~edon Hill
r~./ ~~
\
'--..r'
--
j Rainsborough
ifi nghoe Beacon
. Uffington
Worlebury
Bratton Camp.
Beacon
~I
Ir
Ladle Hill
Mt~~~~
. ~
, 0
10
ENGLIS H CHANNEL
srocns
Major areasof brocn concentration
50miles
!
100km
Shetland
Islands "
A T LA N T IC
C EA N
I
Midhowe
Gurness .
Howe
,.
~~
~~
1\
NOR T H
5 EA
'.~ p
0
o'
II
Location
12
The h ill-forts of Celtic Britain , th e large earth-bu ilt structures th at can be found
th rou gh out th e land, are mostly con centrated in two broad areas. The first of
th ese areas forms an elongated trian gle running from th e island of Anglesey in
Wales southwa rds th rou gh th e rest of Wales, beyond th e Welsh mountains, th en
across th e River Severn in to mod ern-day Eng land . From th ere th e swathe runs
southwards to th e sou th Devo n coas t, an d south -east and east to th e head wat ers
of th e River Th am es, co ntin uing to th e Kentish shore. The seco nd large band of
hill- forts runs th rou gh th e easte rn lowlands of Scotland, from th e Beaul y Firth
near Invern ess round th e coast to Aberdee n, th en south th rou gh Angus and Fife
to th e Loth ians . From th ere it th en runs in a south-easte rly direction toward s
the Solway Firth, wh ich marks the Eng lish border. Hill-forts did exist outside
these areas, but not in such great num bers. Within these genera l bands the size
and sty le of the forts varied co nsiderably. In very broad ter ms the forts found
in Wales, Scotland and the north of England tend to be smaller and more
numerous than the hill-for ts found in the south of England . There are of course
exceptio ns: Traprain Law and Tre'r Ceiri were as impressively large as many
found further south.
These fortifications were built over a span of seve ral centuries, and not all
remained in continuous occupation throughout the Iron Age. Th e fact that
certain areas of mainland Britain seem almost devoid of hill -for ts while others
have a profusion has never been fully exp lained, but may well represen t so me
form of socia l difference betwee n th e in habi tants . Some archaeologists have
argued th at man y of th ese forts represent territorial divisions, whe re th e forts
represen t poli tica l sta te me nts of land owners hi p an d bo un da ries as mu ch as
fortificat ions in thei r ow n right - akin to th e feuda l castles of th e medi eval
period . Certain ly so me of th ese forts might well have served as political cen tres
of po wer; th e proximity of man y to lat er Rom an provin cial cen tres is often
used to suppo rt this argument. Un fortu na te ly we kn ow all too littl e abo ut th e
distribution and political organizati on of Britain 's Iro n Age populati on, so th e
terr itor ial importance of hill -fort s remains a matter of speculation .
The sto ne-bu ilt brochs are concentrated in th e north of Scot land, parti cu larly
in th e northern isles of Orkn ey and She tlan d, and th e Western Isles. However,
isolated exa mples can also be found as far south as th e Scottish Bord ers. Th eir
construction is unique: th e largest surviving structures look more like giant kiln s
tha n forti fications. As not ed previou sly, th ey also resembl e th e sto ne-built keeps
built in th e century afte r th e Norma n Co nquest of Eng land. Due to limit ation s
of space, any descripti on of th ese two fortifi cation types ha s to remai n fairly
general; however, each n eeds to be dea lt with separate ly as, in all but thei r
purpose, hill -for ts and broc hs have litt le or nothing in co m mo n. A related
for m of Iron Age fortificatio n is th e 'dun ', wh ich in essence repr esents a sma ller
version of a broch - more like a sma ll sto ne enclosure th an a tower. Wh ile th ese
may have been used as nothing more th an fortified hou ses or farm comp lexes,
some may well have served as im portant forti fied sites in th eir own right.
13
14
The ter m hill -fort is defined as a fortified enclosure , designed to take advantage
of a h ill or rise for its defensive advan tage . The fortificatio n could consis t of one
or more circular or part-circular ea rt he n or even sto ne ramparts, built to fo llow
th e con to urs of th e hill th e fort was sited on. In man y cases th ese rampart s are
often associated with atte ndan t externa l dit ch es. Th e structures date from th e
Late Bron ze Age and th e Iron Age.
Beyond th is rathe r gene ral definition , th e variation in types an d size of hillforts is con siderable, as are th e dates given by archaeo logists for th e occupa tion.
These fortificatio ns varied from what amo unted to little more th an a sma ll
cattle enclos ure to a fort big eno ugh to acco mmodate a reason ably sized town .
The larger settleme n t h ill-fort s - th e type Ju lius Caesar called oppida whe n he
enco untered th em in Gaul - were clearly of great imp ortance to th e people wh o
bui lt them , and fun ction ed as muc h as a statemen t of regiona l power as a place
of refuge in time of conflict. Few British sites mat ched th e scale of th ese Gallic
forts, alth ough hill-forts like Maiden Castle came close. Some hill-forts fun ction ed
as perm an ent settleme nts whilst othe rs appear on ly to have been occup ied in
tim e of dan ger, during a particular season, or for some specia l even t. Some show
little or no evidence of milit ary use, and sho uld th erefore be seen more as
enclosures to pen in dom estic ani mals rath er than as fortified stro ng points. The
only broad definition of use is that the hill-forts were built to prot ect settlem ents
or livestock from atta ck.
Inevitabl y th ere is some debate over how and why th ese hill-fort s were used,
and whe n. Most appear to have been used for at least tem porary habitat ion or as
fortified military enca mpme nts during th e th ree centuries before
th e Rom an conq uest of Britain, whic h constitute th e Midd le and
Late Iron Age, and all but a few fell int o disuse after th e Rom an
occupatio n. However, a few were re-occupied in th e 5th century AD
after th e Rom an s left, and a few were even used as field fortification s
by th e Ang lo-Saxons during th e Viking incursion s thr ee centuries
later. In areas such as northe rn Scotland and Wales, whe re th e
Rom an influen ce was less pro no unced , hill-fort s tended to remain
in use for longer.
Inevitably th ese multip le periods of occupation and d ifferent
types of use have tended to mask th e real purp ose behind th ese
fortified enclosures, and to enco urage vigorous excha nges between
I5
16
17
18
19
The same equation can be exte nded to incl ude othe r sim ilar hill-forts,
includ ing Maide n Castle. The on ly differen ce would have bee n that in severa l of
th ese sites the defences were built in severa l phases, ofte n a cen tury or more
apart. These earlier works were ofte n inco rporated int o th e new design , wh ich
reduced th e amoun t of wor k involved. However, in its fina l phase Maiden Castle
was pro tected by no fewer th an three ban ks and a countersca rp bank, and three
ditch es. In additio n the gate ho use defences were significan tly grander th an those
th at would have bee n built at Ladle Hill. The hill-fort historian A.H.A. Hogg
estima ted th at in time of peace th e defen ces of Maiden Castle could have bee n
im proved from one building phase to ano ther in a period of approx ima te ly
five to seven months. As thi s would have involved an un acceptable degree of
disru pti on to th e farm ing life of th e com m un ity, he argues that th e wo rk would
have been spread out over two or even th ree seasons. In many ways Ladle Castle
was the ideal size - a sma ller hill-fort would almos t certain ly have served a
sma ller population, wh ich meant th e work invo lved would have taken lon ger.
Here again th e wor k would have im posed on th e agrar ian calenda r, and so would
prob ably have been undertaken ove r at least two yea rs.
20
The com mu nities served by th ese fortifi cati on s mu st have been sma ll
compared to th e Iron Age com mun ities who built hill-fort s, but th en th e two
types of fort ification do not bear close comparison. Given th e correlatio n between
hill-fort size and th e estimated number of peopl e wh o lived in or beneath it, on ly
the smallest h ill-fort s can be compa red in scale to th ese broch s. Both involved a
co nsiderable amo unt of effort for th e sma ll agrarian com m un ity who built th em ,
particularly whe n some hill-fort s or broch s seem to have been little more th an
fortified farms, serving at most one or two extended famili es. It has th erefore been
suggested th at in places like Orkney or She tland, wh ere these broch structures are
common, neighbour in g com m un ities pooled th eir efforts and helped out in th e
construction of each othe r's tower - similar to th e com muna l barn -building
tradition of th e Amish co mm un ity in th e Un ited States.
A typical bro ch was built using dr y-ston e walling, and was approx ima tely
20m in diam eter. The broch at Mou sa sta nds some 13m high , altho ugh othe rs
migh t well have been lower structures - no more th an five metres in height. The
walls were doubl e-skinned, with a cavity between th em wide eno ugh to fit a
staircase th at wo und its way up to th e to p of th e tower. In most large brochs th e
walls are approx imately three metres thick. The structure was en tered th rou gh a
small, low doorw ay; th ese were ofte n flanked by guard cha mbers to improve th e
secur ity of th e fortificat ion . The in ter ior was almos t certain ly divided into floors,
each level bein g accessed fro m th e sta irway and its atten da nt galleries. In broc hs
such as Gurness, Midhowe and Mousa th e lower floor co nta ined a sto ne-lined
well and food storage pit s, whi ch meant th e defenders could withsta nd a len gth y
siege. Archaeo logica l evidence suggests th at th e who le struct ure was cap ped by a
con ical pit ched roo f of th at ch or turf, with timber or even wha lebo ne suppo rts.
The same dry-ston e walling techniques have been practised in the north of
Scotland ever since, as exemp lified by old cro fts and farm boundary walls.
Broch s always seem to have been buil t in easily defensible locati ons, but
they wou ld also h ave had to be co ns tructed close to th e ara ble land worked by
21
22
the com m un ity who built th e struc ture . In man y cases th e struc tures were built
close to th e sea: Gurness and Midhowe were both built on th e Orkney seashore,
separated fro m each ot he r by th e wat ers of Eyn ha llow Sou nd. The Broc h of
Gurness is parti cul arl y impressive, becau se a sma ll sett leme n t of at least six sets
of hou ses was built ben eath th e tower itself. Each was ente red fro m a main
passageway that led th rou gh th e village to th e broch . Some of th ese buildings
were entered from sma ll co ur tya rds, flanked by sto rage she ds. Th e result
resembled a sto ne -built warren . Surro und ing th e village was a substan tial sto ne
wall at least two met res high , which may h ave been surmo un ted by a sto ne
walkway and parap et. Th e who le co m plex was en te red through a double gate,
reach ed by a sto ne- lined approach that span ne d a dit ch running around th e
landward s side of th e village . Beyond this a series of two banks and dit ch es
co m pleted wha t mu st have appeared a most formidable site. The neighbouring
broch of Midh owe was smaller, with a less develop ed defen sive syste m
surro unding th e tower and sett leme n t.
Structures such as th ese suggested th at th e society wh o built th em was one
under threat , eithe r from n eighbouring co m mun ities or more likely from
outsiders. Were th ey built as a reaction to a wave of arme d Celtic settlers from
th e south, or were th ey sym bo ls of com m uni ty power in an othe rwise stable
society? So far arch aeo logists have failed to provid e a clear answer. Wh oever built
th em and for wha teve r reason , th ey remain as pot ent reminders of th e Iron Age
com m un ities wh o felt th e need to defend th em selves in such a dramat ic fashi on.
23
24
Without exception Ce lt ic
fo rt ificat io ns in Britain prove d
unab le to protect t hemselves
fro m the Ro mans. who used siege
engine s firing stone shot to batte r
a path t hrough their fragile t imber
palisades .These red sandstone
exam ples of ballista shot were
recove red from the Burnswar k
hill-fort in th e Sco ttish Bo rd ers.
(N atio nal Museum s of Scot land)
25
26
were less substa ntial th an on th e northern face, whil e th e ditch itself narr owed
slightly to around eight metres across.
Visitors would approach th e easte rn en trance to th e fort und er th e gaze of
sentinels on th e ramparts , wh ere th e guards co uld easily close th e large wood en
gates at sh ort notice if required, or drive intruders away with a hail of slings ho ts.
The chalky approach road was nine metr es wide - eno ugh for two chariots to
eme rge from th e fort side by side . The gate itself was a co m plicated affair, with
two sections of outworks lyin g in front of th e gateway. Even before a visitor
reach ed th e entrance an outlyin g dit ch had to be crossed, a wide but shallow
tren ch that encircled th e wh ole hill -fort. Once past thi s obstacle th e visitor
wou ld be faced with a series of small banks - essentially th e coun terscarp of th e
main dit ch th at surrounded th e fort. These projected out to flank th e path, a
litt le like th e claws of an insect. However, th e ditch ended at th e main gate itself,
so th ese spurs form ed what is best described as a semi-circular oute r enclosure of
th e fort (see page 27).
Th e excava tions revea led that the gap betw een th e jaws of th e bank was
sealed by an oute r gate way, th e exac t nature of whi ch is still uncl ear. Inside this
enclosure two second section s of bank pro vid ed a furth er o bstacle befor e th e
gatehous e was reached, protecting th e flank s of th e en clos ure.
Where th e road cross ed the inner rampart th e banks turned outwards for
20-30m, creatin g a funn el through which an y visitor would have had to pass.
This is wh ere the main gateway stood , a stru cture that was about six metres
wid e, its approac h cove red by the proj ecting hornwork s of th e ram parts. It was
c
27
Dunsin ane Hill. Perthshire the lite rary last bast ion of Macb eth
- is the site of a sma ll Iron Age
hill-fort overlooking Str athmore.
A th ick. stone-fi lled inne r wall was
surro unded by two outer ra mpar ts ,
alth o ugh many of the fort's featu res
we re destroyed by exte nsive digging
by 19th-ce ntu ry antiquaria ns.
(RC AHMS)
28
a double gate, its two wings o pen ing inwards int o th e fort itself. Th ere is
evide nce th at some form of tow er or gallery structu re stoo d ove r th e gateway,
probably linking th e two en ds of th e in ne r rampart s to form a co ntinuo us band
of defe nces . A visitor wo uld have had to pass under th is struct ure to en ter th e
fort itself.
Once th rough the defe nces the wide cha lky road wo uld lead off in to th e
interior of the fort, narrowing as it wen t un til it reached th e far side of th e
enclosure. Other smaller roads forked off from th is ma in th orou ghfare like
branches from a tree, two to the left and th ree to th e right. The roads to th e left
were flanked by rows of square-shaped gra naries , probably raised off the gro und
slightly on nine posts - three on each face and one supporti ng the centre of
the structure. Each side of the building was
approximately six metres across, alt ho ugh a few
smaller, two me tre -wide, four-post grana ries or
hayrick platfo rms lay scattered between th e
larger buildings. Almost certai n ly each structure
contained baskets or cera mic jars filled with
wheat, oa ts or barley. To the righ t th e visitor
would have seen th e main sett leme nt area of
Dane bu ry, with a seeming ly irregular scatte ring
of th atch ed roundho uses th at extended as far as
the northern side of th e ea rthwor ks.
The last fork in th e road led off towards a
sh rine or temple structu re, built in what was
rough ly th e cen tre o f th e who le enclosure. It
was a squa re structure, approxi ma tely th e same
size as th e nine-post grana ry build ings, but of
more solid construction, with its walls set int o
the ground. Three sma ller sh rines lay a few
yards down the path towards the fork in the
road , a larger one on the left of the pat h and
two smaller ones to the right. All three were
square, the largest being just under three me tres
across , and th e entrances of all three st ruct ures
faced the visitor as he approached the main
temple 10m behind t he m on the pa t h.
Althoug h archaeology can not defini tely say
what was he ld in th ese shri nes, we know from
other evidence that th e Celtic peopl e who lived in Dane bury wo uld h ave
worsh ipped a ran ge of deiti es, an d surro unded images o f th eir gods with votive
offerings of food , dr ink or even scraps of cloth . Th e impo rtance of th ese
structures wou ld have been em phasi zed by th eir cen tral position with in th e
Dan ebu ry enclos ure .
The one thing missin g from th e picture created by th e arc haeology of th e site
is the im pression a visitor would have had of the peopl e who lived th ere. At its
he igh t Daneb ury had a populati on of between 300 and 500 peo ple, as well as
their do mes tic anima ls and pets. Chic kens would have scratc hed aro und th e
feet of any visitor, whil e th e sme ll of an ima ls held in wicker enclosures next to
the roundho uses would have been pervasive. San ita tion was largely unkn own ,
and hu ma n waste was sim ply dep osited in sma ll pit s th en loosely cove red ove r.
From ot he r sources we kn ow wha t th ese peo ple would have looked like, and
the appearance of th e Celts of th e Late Iron Age has been cove red in other
Osprey books (see Men -at -Arms 158, Rom e's Enem ies (2): Gallic and British Celts).
However, th e gene ral appea rance would have been of a fairly wealthy ru ral
economy, and a com m un ity th at en joyed both a stable po litical structure and
one wh ich could easily prov ide for th e welfare of its own people. Far from being
a primitive sett leme n t, Dan ebury would have been a well-organ ized farming
co m mun ity, whose inh abit ants en joyed a reason ably h igh sta ndard of livin g,
and whose safety was assure d by th e impressive defe ns ive works tha t protected
their village.
29
Maiden Castle
To better understand th e way hill-forts developed with the passage of time we
could do worse th an to look at probably th e most famo us examp le: Maiden Castle
in Dorset (see pages 38-39). This, th e largest hill-fort site in Britain , was subject to
two large-scale scientific investigat ion s, allowing us to trace its develop men t and to
und erstand the way it fun ction ed as a settle ment with more certainty than man y
ot he r sma ller sites. Maiden Castle was first excavated by Sir Mortimer Whee ler
between 1934 and 193 7. Further excavations were carried out in 1985-86. It has
been proposed th at th e name probably derived from th e pre-Celtic nam e for th e
hill-fort, 'Mai Dun ', wh ich approximates to 'big fort '. It occupies a prominent
saddle-sha ped ridge two miles from th e town of Dorchester, presenting a striking
30
ap pearance to the visitor. The novelist Tho mas Hardy described its app earance with
considerable eloq uence:
At one's every step forward it rises high er aga inst th e sout h sky, wit h an
obtrusive person ality th at com pels th e sense s to regard it and cons ider ... The
pro file of th e who le stupendo us ruin , as seen at a distan ce of a mile eastwa rds,
is clearly cut as th at of a marbl e inlay. It is varied with protruberan ces, wh ich
from hereabo uts have the anima l aspect of warts, wens, kn uckles and hi ps. It
may indeed be liken ed to an eno rmo us man y-limbed o rganism of an
an tediluvia n tim e ... lyin g lifeless, and cove red with a thin green cloth,
wh ich hides its substance, wh ile revealin g its conto ur. I
Tho m as Hardy. 'A Tryst in an Anci en t Earthwork', from A Cllanged "'JIUI, und Other Tilles (Londo n, 19 13).
31
32
two cha n ne ls by a median bank. The rampart s and outlying ditch were exten ded
to th e western side of th e ridge, wh ere a second en trance was co nstructed, again
with a semi-circular lin e of oute r work s. Both ram parts were pierced by two
gateways approx imate ly 50m apart. Th is ph ase of th e development of Maid en
Castle ha s been link ed to th e cultura l ph ase kn own in Iron Age archaeology as
'A Culture', namely th e first identifi able British culture of th e Iron Age. Althou gh
th ese peopl e took advantage of th e new iron-making technology import ed from
th e mainland of Euro pe, it is now th ou ght th at th e majorit y of th ese peopl e were
indi gen ou s inhabit ants of th e region.
Aroun d th e same tim e as Maiden Castle was expa nded a new group app eared
in southe rn Eng land, th eir route traced by th e rem ains of th eir distin ctive
pottery, th e use of th e sling and th eir ow n parti cular ideas about fortifi cati on.
These peopl e were clearly identifi able as Celts.
Arou nd 250 BC th ese members of ' B Cu lture', as th ey are known, began to
make th eir mark on Maid en Castle; its defen ces went th rou gh an ex te ns ive
revision . As an offens ive wea po n th e sling proved superior in both ran ge and
firep ower to th e javelin used by th e 'A Culture' inhabitants. Wh ether thi s
defen sive improvem ent was mad e by th e old gro up or th e new is largely
unknown, but th e new sche me was certa in ly design ed to coun te r an d to take
adva ntage of th e sling's capabilit ies. On th e north ern side of th e ridge a
seco nda ry bank and dit ch was created, suppo rted by a sma ller spur bank on th e
n orth-western face of th e ridge to scree n th e western gateway. On th e south
side of th e ridge two banks and dit ch es were added, in addition to additiona l
bank defen ces in front of both gateways, design ed to fun ne l attac kers into a
'killing zo ne' for slings hot. Fina lly the origina l inner ramp art was repaired an d
hei ghten ed , the extra scale of the bank su ppo rted by a sto ne revetm ent buri ed
on th e in ne r face of th e ea rthe n ba n k.
Th e final form of Maid en Castle's form idable array of defen ces was prob abl y
co m pleted at so me stage during th e ea rly 1st cen tury Be. The rampart s were
en larged once again , wh ile a substa ntial co unte rsca rp bank was ad ded wh ich
enc ircled th e who le ridge . Wha t was most sign ificant about thi s th ird phase of
improvem ents was th e streng the n ing of th e two gate ways. On th e easte rn en d
th e outlying defen ces of th e o ld gateway were filled in or dem olish ed, an d in
th eir place a series of two large fortifi ed rampart s were added, both fro n ted
by a dit ch an d a sma ller co un te rscarp bank. On th e weste rn end of Maide n
Castle the old defences were greatly strengthened, and augmented by ano ther
outlying rampart, ditc h and cou n terscarp ditch. In addition smaller banks
with in the gateway co mplex acted as barriers to funnel attackers trying to
round th e last outlying rampart before th e gateho use.
The twin gateways th em selves were greatly streng the ne d with sto ne
revetmen ts, as was th e north-western en trance to th e gateway where it passed
the first line of th e outer defen ces. Betwee n th e twin gates and th e first outer
rampart a row of guard huts hou sed th e gate garr ison , suggesti ng a level of
m ilitary organizatio n th at had been lacking in previous defensive systems . In
mo st of th e outer dit ch es leading to th e gate ho use (and possibly elsewhe re
aroun d th e fort perim eter) woo de n sta kes were emplaced as chevaux de {rise,
obstacles designed to deter or hinder an y attacker. Finally, firin g platforms ma y
well ha ve been installed at various points along th e oute r rampart s, leading to
the gatehouse, so that a 'for lorn hop e' o f slinge rs could shoot into th e flanks or
rear of an assault part y. By now Maid en Castle had develop ed int o what was
probably th e best-d efended hill- fort in Britain .
Around the time th ese fina l improveme nts were being mad e to th e defences
a new gro up arr ived in southe rn England. Known as 'C Cultu re' peopl e, th ese
in comers were Celts of th e tribe known as th e Belgae, wh ose origins lay in what
is now northern France and southern Belgium . Within half a century these
inc omers had exte nded th eir con trol over most of south-east England, an d by
AD 25 at th e latest thi s had extended as far as Maiden Castle, wh ich th ey may
well have occ upied. Certain ly their influ en ce was felt wit hi n th e fort. The
Belgae, or those wh o ado pted th eir ways, repaired th e ramparts by rein forcin g
the ba n ks with a layer of earth and streng the ned th e wall walk an d palisade
th at surmo u nted it. Strangely eno ugh thi s palisad e was mounted on th e
in ner side of the rampart, leaving th e men who patrolled it exposed to fire
from outside the fort. This has been explaine d as being mor e of a secur ity
barrier th an a defen sive work - controlling access to th e rampart s as a privilege
reserved for th e warrior elite of th e garrison. The posts su ppo rting this palisade
were sunk deep into th e outer edge of th e ba n k, and so in effect they dou bled
as a reinfo rceme nt for th e sto ne revetment bu ried in th e bank itself.
These imp rovem ents did not help th e defenders
when th ey encountered Vespasian's II Legion in
AD 43. Maiden Castle was captured without much
difficulty by th e legionary commander, wh o would
soo n become a Roman empero r. After capturing
the hi ll-fort th e Rom an s destroyed th e fort 's
gateways, leaving it defen celess. The site rem ained
occupied for ano the r three decades, until th e
Romans established a new region al capital two
miles away in Durnovaria (Dorchester), named
after th e local tribe known as th e Durotrigii.
Around AD 70 Maiden Castle was aband on ed, and
its once formidable defen ces became a place of
pasture. The Rom an s had one fina l humiliation for
th is great symbol of Celtic power. In AD 3 13
Ch ristianity becam e th e official religion of the
Roman Empire, and around AD 380 a small square
temple was built on the eastern side of th e ridge,
within th e bounds of th e original fort. A large
circular sh rine was th en built beside what was once
the main th orough fare of th e hill-fort, occupying a
site that may once have belon ged to th e prin cipal
Celtic roundho use in th e fort. Both structures
had fallen int o disuse by th e end of th e 5th
cen tury AD.
33
34
LEFT
T he settlement
The one feature th at bo th excava tions at Maide n Castle failed to reveal in any
detail was the configura tion of the hill -fort 's inte rior. Afte r all, th e reason the
fortifications existed in th e first place was to protect the Iron Age community
who lived and wo rked th ere. Archaeology has revealed a littl e abo ut how th ese
people lived , and h ow th eir settleme n t was organized. Th e first settleme nt
there was a Neo lith ic one, establishe d aro und 4000 Be. However, th e
commu n ity th ere was relati vely small - prob ably no more th an 100 peopl e,
based on th e size of th e causeway cam p th ey built . It seems to have rem ain ed
a focus for the Neo lithic people of South Dorset, as abo ut 3 50 0 BC the long
ban k barrow was built, suggesting th e ridge was seen as a centre of religio us
importance. The area was aba ndo ned a few centu ries lat er, and it was no t until
the very end of th e Neo lith ic period , aro und 22 50 BC, that archaeologists have
been able to trace any furth er act ivity. During th e Bron ze Age there seems to
have been little occupation, although th e nearb y Frome Valley becam e a
relatively well-p opulated area. Th is all cha nge d aro und 50 0 BC, whe n th e Iron
Age peo ple of th e region bu ilt th eir hill -fort th ere.
Although little trace rem ains of th e settleme nts created by either th e Neolithic
sett lers or th eir early Iron Age ancestors, by comparing Maid en Castle with ot he r
sites in th e area we can see th at th ere seemed to be a tenden cy for communities
to congregate int o easily defended settleme n ts dur ing thi s period , prob ably du e
to an increased level of social unrest. The trouble with Maiden Castle is that th e
35
36
excavation co n ducted during th e 1980s also revealed a littl e m ore abo ut the
occupation of th e fort, and a number of these rou nd houses were excavated .
The rema ins of three huts were d iscovered in 1986, the largest of whic h
measur ed almost six metres in diameter. These struc tures were typical of the hu ts
associated wit h the Iron Age in Britain: wooden circular structures, surrounded by
a sligh t ditch and ba nk . The build ing was centred on an open hearth, while traces
of an oven were found to one side of this . The main st ructural timbers of the
dw ellin g co ns iste d of a circle of upr ight posts hold in g up a frame of timber bea ms .
The en trance to the hut faced so uth , an d was delineat ed by a sma ll fen ce leading
o nto a lim estone wa lkway. A post-h ole suggests th at the en trance was o nce
secur ed by a substa n tia l woode n door. The excavating team also d iscovered that
th e stru cture was in use for a lon g time, and was rebu ilt at least th ree times during
its occu pan cy.
Cu lt ivatio n a n d qu arrying has des troyed m uc h of the evidence of other huts
o n to p o f the plat eau , so all we ca n do is ima gin e th at d uri ng t he Late Iron Age
th e h ill-fo rt co n ta ined numerou s st ruc tures of this typ e. After all, ex pe rime ntal
arch aeo logists and re-enact or s h ave proved that a hut of th e size m entioned
above co u ld co m for ta bly h ou se an ex te n de d famil y grou p o f arou nd six ad u lts
an d ch ildren . Give n th e co rre latio n bet ween th e size of th e h ill-fort and the
popu latio n it h oused di scu ssed previously, we arrive at a projected to tal of
just over 180 huts, h ou sin g over a tho usand people. This nu m ber of buildings
seem s hi gh , give n the n eed to pr ovid e add it ion al space for sto rage facilit ies and
refu se pit s, n ot to m ention worksh op s, co m m u na l buildings an d rel igiou s
ce n tres within th e same enclosure.
In the de cades before the Roman in vasion these underground sto rage an d
refuse pit s we re filled in , an ac tio n whic h m ay well reflect the in flue nce of the
'C Cu ltur e' peop le wit h in th e for t. They were replaced with storage barns, which
im proved the ca pac ity an d the su ita bility of grain storage within th e hill-fort.
Even m o re spectac ula rly, the huts appea r to h ave bee n reorgan ized . Rather than
bein g scattered aro un d the rid ge, they we re co ncen t rated into rows, a littl e like
modern subur ba n st reets, with eac h roundhou se en clos ure space d eve n ly, a nd far
closer to each oth er than before. A row of three suc h h ou ses ha s been unco vered ,
includ in g o ne built using sto ne . Th e middle h ou se o f the three was built o n
th e foundati ons of an ea rlier struc ture, and was eve n te rraced sligh tly to take
advan tage of the nat ur al slope. The final h ou se of th e th ree was surrounded by
a sm all gu lly. All three st ruc tures a ppea r to h ave bee n repaired during their
occu pancy, wh ich suggests they rem aine d in use for so m e considerable t ime p robably surv iving beyond th e period of Roman in vasion in th e mid-1 st ce n tury
AD, roughly a cen tury after th e stru ctures were first built.
37
38
House
40
to have broken down . For the next few decades the population appeared to
decline steadily, and buildings were once again scattered across the site rather
than grouped together in streets. Only one house from this period was firmly
identified in the 1980s, but it is clear that some if not mos t of the lat er
'suburban ' houses fell into disuse. Other evidence of occupation is sparse,
although Professo r Wheeler uncovered the remains of five ho uses and several
storage pits as well as an iron-working area that he associated with the period
im mediately after the Roman invasion . Th ere was also evidence that the
sett lemen t spilled out through the disused eastern gateway and that buildings
were established within th e banks of the fort's outworks beyond the gate. At
the same time many of the out lying ditches were filled, suggesti ng a change of
em phasis from defence to accessibility. The settleme n t was abandoned a few
decades lat er, as th e regio nal cen tre of power shifte d two mil es east to th e n ew
Roma n civitas pe regrina (regional cap ita l) of Durnovaria (Dorc hes ter).
41
Celtic fortifications In
operation
Hill-forts today are deceptive places: the smooth, grass-covered banks and
ditches have been rounded and weat he red over the centuries, and th e approach
to th em, alt ho ugh often somethi ng of a climb, is no longer an expe rience fraugh t
with dan ger. It wo uld have been a di fferent sto ry at the height of th e Late
Iron Age, where th e ramparts wo uld h ave been stee per an d high er, the dit ch es
impassa ble, and th e rampart s lined with well-armed defen ders. Sta kes and
im pedime n ta wou ld h in der any approac h over th e oute r banks and dit ch es
of the fort, wh ile th e mai n gate itself would present a labyrinth ine trap to an
un wary ene my.
42
o rigin ally bui lt usin g th is bo x m eth od , but at a lat er d at e the o rigina l st ruc ture
was rep laced by a la rge eart he n bank.
Buildi ng a ram pa rt withou t a box struc t ure appe ars to h ave developed
du rin g th e Early Iron Age, alt h ough a sim ple ti mber revetmen t was someti mes
used to h ol d th e ea rth o r sto ne in pla ce w he n the bank was being built .
An exam ple o f this typ e o f co ns t ruc t io n is found at Cissbu ry, Sussex, a large
20-hectare site whe re the palisad e for m ed th e o ute r retain ing wa ll o f an eart h en
ba n k behind it . Thes e earth en ba nk s ten d ed to be hi gh er than earlier st ruc t ures,
with a steeper ou ter face to ma ke it h ard er for an att ack er to rea ch t h e pa lisad e
or breastwork at the to p of th e ra mpart . Similarly a st eep in ne r face to the
ram part wa s ea sier to co n st ru ct, as it reduced t he amou n t of so il th at h ad to
be moved. At Wand lebury, Cam b ridgesh ire, th e ear lie r box st ruc tu re was
co n vert ed into an oute r de fensive lin e, wh ile a larger t im ber-fronte d bank was
bu ilt t hat rese m bled the o n e fou n d at Cissbur y. What these rem ains fail t o
p rovide is any so lid ev ide nc e fo r the palisad e or br eastwor k th at ran alo ng the
to p o f th e ra m part. Presu mably the timber revetment doubled as a pa lisade, as
it was h igh er th an th e bank for me d behind it.
Alth ough sto n e-built rampart s we re ver y different in appearance to eart he n
ones, a sim ilar approach was ado pted. The hi ll-for t at Tre'r Ceiri in Caern arfo n
was sur rou n ded by a sto ne -bu ilt ram pa rt whe re th e oute r face also form ed th e
palisade. A stone wa lkway ran behind this out er face, and the wa ll the n descend ed
toward s t he interior of th e enclosure by m eans of two ste pped revetments or
te rraces, wh ich im pa rte d greater streng th to th e who le struc ture . A m ore co m plex
stone ram pa rt is fou nd at Worlebury in Some rset,
wh ere archaeologists h ave shown th at th e origina l
com plex o f wa lls once stood to a height of ove r 10m.
As at Tre'r Ceiri th e main ram part was supported
by a series of sto ne revetmen ts o r terraces stepping
down toward s th e interior of th e fort, wh ile th e
outer face presen te d a n ear ve rtica l face to any
att acker. In effect thi s stre ng the ne d th e defen ce,
because eve n if th e outer wa ll were da maged, th e
stone revetments beh ind wo uld serve as seco ndary
walls, thus m aintaining th e in teg rity o f th e defen ce.
A var iatio n of th e co nve n tiona l sto ne ram part is
found in th e 'vitrified ' forts of Scotlan d, whe re the
walls were delib erat ely sub jecte d to th e effects of fire.
The process of setti ng fire to th e st ruc ture fused th e
rocks together in va rious degrees, an d in some
cases prod uced a di stincti ve glass-like coa ting th at
served as a binding age n t. Exam ples include th e
Tap O'Noth on Bennachi e, Gra m pian; Barr y Hill,
Perthsh ire; and Craig Phadrig outside Inverness.
Sim ilar vitrified struc tures were found in Ireland an d
in Cen t ral Europe, but outside Scotlan d th ey are no t
fou n d elsewhe re on th e British m ainland. In man y
cases th e Vitrified wa lls were th en reinfo rced or
revett ed by un vitrified sto ne, ofte n built up on both
the outer an d inner faces of th e vitri fied rampart.
However strong th e ram pa rts of an Iron Age fort
were, th e wea kest point of th e site was always th e
gateway - an d it was presumab ly th ere th at an
attacker wo u ld conce n tra te h is effor ts. Obviously
the builde rs placed great em phas is on strengthe ning
th e gateway defen ces, usually by placin g obstacles in
fron t of th em , wh ich would cha n ne l an attack in to
killin g zones whe re the defenders cou ld shower th e
43
attackers with javelin s and slingsho t sto nes. The simplest form of gateway would
be an open entrance, sealed with a temporary barrier such as cut logs or felled
trees. Usually th e passage th rough th e rampart was faced with stone, the n blocked
by one or more wooden gates. Maiden Castle appears to have had two gateways,
separated by a sho rt length of rampart. In some fort s, such as Dino rben, a
gateho use or guard post lay beh ind th e gate itself, wh ich suggests th e presenc e of
a perm an ent gate garrison. At Maiden Castle one of two small guard posts located
at th e eastern en trance was equip ped with a hearth, whi ch supports the the ory
th at th ese posts were perm an ently manned.
One surprising aspect is th e lack of iron fittings associated with Iron Age
gateways. At Sou th Cadbur y, Som erset, and Hembury, Devon , iron rings were
found whi ch might well have form ed part of a gate h inge. It appears th at in most
cases these h inges, like th e gates and gateposts themselves, were constructed using
wood. The width of th e gateway seems to have varied considerab ly; at Danebury
in Hamps hire th e gate itself had two leaves or sides, and was sup ported in th e
centre by a post set int o a stone-lined post-hole. At Breden Hill, Gloucestershire
th e gateway span ned a Sm-wide road, and was constructed in a similar fashion
to Danebury. What is int erestin g about both th ese gateways is th at both were
set between lon g passageways, formed by th e ramp art s turning outward (at
Danebury) or inward (at Bredon Hill). Defend ers on the ramp arts could th erefore
sho wer th e approac h to th e gate with sto nes even more effectively th an th ey
could when defending a more conve ntiona l gateway system. At Breden Hill it
app ears that th e line of rampa rts was spanned by a foot bridge whic h crossed the
approach road, wh ile th e gatewa y itself was set some 20m further back, at the end
of the inwa rd-turning horns of th e bank. Some archae olog ists have suggested th at
these gateways or footbridges were decorate d with troph ies of war such as severed
hea ds or skulls, or augmented by some form of triumphal arch. However, the
evidence for these features is either circums tan tial or no n-existen t.
The gateway itself was usually approached th rough a series of outer works,
which were designed to make th e attackers turn and expose their side to the
defenders. If the attacker was equipped with a shield then a left-hand turn in the
44
T he defence of a hill-fort
The Celtic fortification s of Britain were certain ly not designed to withstand an
A section of th e Bro ch of Mousa
att ack by a professional standing army suc h as th e one fielded by Rom e in th e
shows how the walls were dou bleIst cen tury AD. The Greek historian Strab o said of th e Celts that th ey 'were war
skinned above groun d level. allowing
mad, high spi rited, and quick to battl e, but othe rwise straightforward, and not
space for t he stairwa y between
of evil character'. By nec essity th e way th ey designed th eir fortifi cati ons was
th e tw o walls.As at Gurne ss.
in fluen ced by the manner in which th ey waged war. In particular, th eir ability
the groun d floo r co ntained wells
and storage pits. (Stratford Arch ive)
to defe nd brochs, hill-f orts or oth er fortifi ed sites depended on th eir tactical
ability, their available weapons, and th eir skill in
using th em .
It is important to cons ider th e weapons at th eir
disposal in order to und erstand wha t part th ese
played in the defence of a fortified position. We
know from th e writings of Rom an historians such as
Caesar, Tacitus, Dio and Sueto n ius amo ng others
how the Celts fought , and what weapons th ey used.
The principa l missile weapon in use in Britain was
the sling, which fired a sma ll rou nd sto ne a distance
of up to 60m . Altho ugh primaril y a hun ting weapon,
it could also be used in time of war. The large caches
of slingshot stones recovered from several hill-forts
are clear in dication s that thes e weap ons were
considered crucial in th e defen ce of a fortified
position. An opponent would rarely be killed by a
slingshot, but th e sto ne it fired could break bon es or
crack skulls. A hara ssing fire could be aimed at an
approach ing force, and as th e attackers clamb ered
over th e outer lines of banks and ditches they would
c-o
SfCOON
be slowed, thus rem aining in optimum range for
zo
I
longer than if their approach went unhampered.
45
~
C1'
A
B
Ramparts
Ditches
Gateway
Guard cham bers
E
F
Glacis
Parapet
G
H
Outer ditches
App ro ach roa d
Once th e ene my reach ed th e last ditch th ey would be with in javeli n ran ge.
Although th ey were slow in flight, if eno ugh light th rowin g spea rs were th rown
at a target th e barra ge wo uld almos t certa in ly cause casua lties, bein g di fficult to
avoid . Given that th ey were usually th rown ove rarm from a rampart aga inst
troo ps approaching th e firer from below, th ey were difficult weap on s to aim
with an y effect. Instead th ey should be co ns ide red mor e of an indirect fire
weapo n - a last form of defen sive fire befor e th e attackers reach ed th e rampart
or gateway . Finally, th e defenders would throw wha teve r th ey had to hand,
such as piles of roc ks. Once th e attac kers h ad scaled th e bank and had reac he d
t he rampart, th e fighting wou ld be hand-to-h and, usin g spear, swo rd an d
sh ield . Bows were almos t certai n ly used as hunting wea po ns by th e Celts in
Britain , altho ugh th ey appea r to h ave been fairly rare. Strange ly th ere is no
accoun t of th em bei ng used as a military wea po n .
The practical limitat ion s of these weapons influenced th e way th e Celts bu ilt
fortifications. After all, th e peo ple who built the brochs and hill-forts of Celtic
Britain almost certa inly had no experience of the Roma n way of
war, and had no in kling of the vast techn ological gap between
their defens ive methods and the Rom an form of siege warfare,
with its secure fortified camps, siege engines and concen trated
bombardments. They built to defend them selves against wha t
they knew - raids by fellow Celts, or even large-scale assaults
a determined tribal enemy. Th is meant maintaining a welldefended perim eter, and encircling thi s with man-m ade or
natural terrain designed to hinder an attacker, eithe r by forcing
the m to endure a rain of missiles as th ey approach ed th e
inner ramparts, or by tirin g them as they struggled to climb
up towards th e waiting defenders. Given th ese parameters,
hill-forts app ear to have been successful in doing what th ey
were designed for. Although we know less about how brochs
were defend ed, th eir similarity to later Norman keeps or even
Border Reiver stro ngho lds spea ks volumes about th e practicality
of their design .
We know a littl e abo ut th e type of wa rfare for wh ich th e
great hill -forts were built fro m Julius Caesar, wh o describ ed
Celtic siege tact ics as th ey existe d in th e mid-l st cen tury Be.
His com men ts are brief but revealing:
There was a town of th e Rem i, by nam e Bibrax, eight miles
dista nt from th is camp. This th e Belgae on th eir march began
to attack with great vigour. [The assault] was with difficult y
sustained for th at da y. The Gauls' mode of besieging is th e
same as th at of the Belgae: when after having drawn a large
numb er of men around th e whole of th e fortifications, sto nes
47
have begun to be cast against th e wall on all sides, and th e wall has been
stripped of its defenders, [then ], form ing a testudo, th ey advance to th e gates
and undermine th e wall: which was easily effected on thi s occasion; for whil e
so large a nu mber were casting sto nes and darts, no on e was able to ma intain
his position upon th e wall. When nig ht had put an end to th e assault, lccius,
who was then in comma nd of the tow n, one of the Rem i, a man of the
highest ran k and influen ce amo ng h is people, and one of th ose who had
come to Caesar as ambassado r Ito sue] for peace, sends messenge rs to hi m, Ito
report] 'That, unl ess assista nce were sent to h im he could not hold out any
lon ger.' (Til e Gallic Wars)
Caesar duly march ed to th e aid of th e Remi and dest royed th e Belgae in
battle. The acco un t m ight well have been written a cen tu ry befor e th e hill -fort s
of Britain faced an attack by th e Rom an s, but th e tribal warfare for wh ich th e
forts were design ed would h ardl y ha ve cha nge d much . Th e mention that
th e attackers assaulted the gat e is particu larly revea ling, as the evide nce from
Dan ebury and seve ral othe r fort s suggests that th e gate way was th e weak point
of th e defen ces. Once th e defenders cou ld reach it they wou ld be able to set it
on fire, which mig ht well ha ve been wh at happen ed at Dan ebury. Of co urse
Caesar's co m me nt that th e Celts form ed a testudo (or 'to rto ise', an attacking
formati on used by th e Roman army) is misleading. It was sim ply the best
means he had of describing a den se assa ult co lum n of Celtic warri ors.
A problem with descriptions of the Celtic forts of Britain and th e way th ey
were att acked or defended is th at we mu st rely on a mi xture of n on- Celti c
observers, and ofte n fragm entary or misleading archaeo logical evide nce. Th e
combination of the two can some times h ave dr amatically misleading result s.
Dur in g h is excavatio ns of Maid en Castle Sir Mortimer Wh eeler becam e
co nvin ced that th e h ill-fort had been attacked by th e Rom ans , wh o stormed
th eir way into th e fort 's easte rn gateway. He had goo d reason to be co nv ince d,
as the hill -for t stood dir ect ly in th e path of th e Roman inv asion, an d hi s team
uncovered th e remains of wh at he th ou ght were war graves .
In AD 43 the Rom an II (Augusta) Legion advan ced rapid ly th rough southern
Eng land, led by its com ma nde r, the futur e Roman empero r Vespasian. According
to h is biograp her, Vespasian subdued 'two very formidab le tribes and over
20 towns' (or rath er hill-fort s), one of which was probab ly Maiden Castle; th e
tribes were probably th e Belgae and th e Durotri gii. By th e time th e Romans
reach ed Maiden Castle the defenders had prepared them selves as best they could;
48
archaeologists have found th e rem ains of substantial cache s of stone shot for
slings, th e ma jo rity of whic h appear to have bee n gat hered fro m nearby Chesil
Beach. Sir Mortimer Wheeler argu ed that Vespasian would have crossed th e River
Fro me wh ere Dorchester now stands, and ha ving seen how formidable th e
western defen ces wer e, o rdered his legion to concentrate in front of the east ern
gate. The hill -for t, d escribed by th e h isto rian Leonard Cott rell in 1958 as the
wo rk of a 'Vauban of th e Iro n Age', wou ld indeed have looked form idable. In
h is rep ort o n the excavations published in 1943, Wheeler described what he
though t occurred next:
First th e regime n t o f a rtillery wh ich usuall y acco m pa n ied a legion was
ordered into action a nd put down a barrage of ballista arrows. The arrows
have bee n found about th e site, and buried amo ngst the outworks, as was
a man wit h an arrowhead sti ll embedded in one of h is ver teb rae (to be see n
in th e Dorch est er Mu seum). Follo win g th e barr age, th e Roman infantry
adva nce d up th e slo pe, cutting its way fro m rampart to rampart until it
reached t he innermost bay, where some circu lar hu ts had recently been
built . These were set alig h t, and u n der th e rising clo uds of smoke th e ga tes
were sto rme d an d th e positi on carr ied . But resistan ce had been obsti na te
an d th e fur y o f th e legionaries was aroused . For a spa ce, co n fus ion, and
massacre do mi na ted th e sce ne . Men an d wo men, yo ung and o ld, we re
savagely cut dow n be fore th e t roo ps were ca lled to heel.
A systematic slight ing of t he defen ces follo wed, whereafter th e legion was
wit hdrawn, doubtless taking hostages with it, and th e dazed inhabitants
were left to bury their dead amo ngs t th e ashes of th e huts bes ide th e gates .
The tas k was carried ou t anxiously and wit ho ut order, but, even so, from few
graves were o m itted th ose tr ibutes of food and d rink which were the proper
requisites of th e dead . Wit h th eir cu ps an d food -vessels and trinkets,
the bones, ofte n two or more skeleto ns huddled into a sing le grave and
many of the skulls deeply score d with swor d cuts , ma de a sad and drama tic
showing - the ea rliest British war-cemetery known to us.
Wh ile today o ne mi ght applaud th e flair wit h wh ich a rchaeo log ists wro te
repo rts in th ose days, su bseq ue n t excavatio ns have revealed th at seve ral
assu m ptio ns Sir Mo rti mer Wheeler made to develop his ' Battl e of t he East Gate'
t heo ry ca n no t now be sustained. His assa u lt theory rests o n th e prese nce of h is
'war-cem etery', but h e failed to show that of the 52 bodies di sco vered there,
o n ly 14 had died by vio lent means . Today a rchaeo log ists co ns ide r it more likely
th at the site was indeed a cemetery, but o ne wh ich developed ove r ti me , and
to wh ich bodies we re broug h t for burial. They were therefore no t bur ied where
th ey had fallen in defen ce of th e easte rn ga teway. Mod ern fore nsic st udies ha ve
even sho wn that so me o f th ose that sustaine d wounds did n ot die from t hem ,
but ins tea d lived o n for so me time afte r receiving th em .
Wheeler based his acco unt of th e burning of th e fort's guard h uts on his
d iscovery of a charcoal layer just o utsi de the eastern gateway. However,
archaeo logists now believe th is was produ ced by iron- working, th e evide nce
of which in th e form of Celtic arrow hea ds now suppo rts thi s th eor y rather
than provides us with evide nce of an assault. After all, bows were used in
insign ifican t numbers by th e Celts at this time, and thei r prese nce does not
necessarily suggest th at a unit of Celtic archers made a last sta nd on th e spo t. The
o n ly piece of evide nce that ha s successfully stood th e test of time is the slighting
of the gateway - as revealed by th e collapsed sides o f th e sto ne-clad gateway. The
remains were over laid by early Rom an pottery, suggesting the collapse occu rred
prior to Rom an settleme n t in th e area - a date co nsistent with the Rom an
in vasio n . Th is is co nsisten t with th e idea of Vespasian's progress th rou gh southe rn
England, and may repr esent a deliberate policy of destroyin g th e gateways of
Celtic fortificatio ns as a mean s of guara nteeing the subj uga tion of the in ha bita n ts.
49
Maiden Castle was not the only hill-fort attacked by the Romans, or even
by the Celts. Around 100 BC the hill-fort of Danebury was destroyed, or at least
its gateway was burned down . Tools and horse trappings were abandoned,
suggesting a hurried departure of the inhabitants, or a violent end. The remains
of 21 mutilated bodi es were fou nd in two grave pits close to the gate , of both
sexes, and ranging in age from four to 45. The pits were never properly cove red,
suggesting o pen graves int o wh ich th e bodies were th rown . All this poin ts
towards a violent end to th e occupation of Danebur y, but once again archaeology
sto ps sho rt of explain ing wha t exactly happen ed. Professor Barry Cun liffe, th e
di rector of th e Dane bury excavat ion, suggested th at th e end of th e fort was a
result of ten sion created by a popu lat ion expansion in southern Englan d, but th e
full sto ry may never be known .
At Hod Hill, Dorset th ere is evide nce of th e hurried repair an d improveme n t
of th e defens ive works, possibly undertaken as a response to news of th e Rom an
invasion. The last-m inute im provemen ts did littl e to help th e defe nde rs, as the re
is evide nce th at Hod Hill was attacked by th e Rom an ar my in AD 43, almos t
certain ly the work of Vespasian and h is II (Augusta) Legion as it marched west
through modern-day Dorset. The site was excavated d uring th e 1950s by Sir Ian
Richmond, who was parti cularly keen to find evidence of a Rom an assault. Wha t
he d id discover was even more intriguin g. One of th e roundhouses withi n th e
enclosure was larger tha n th ose aro und it, which might suggest it was th e hom e
of a chiefta in or an imp ortant admin istrative buildin g. Archaeologists un covered
11 Rom an ballista bolt s amid its ruins, bur ied nose-first as if fired from th e same
locat ion somewhe re outside th e south-east corne r of th e fort. The accuracy of
the fire was parti cularly impressive, as th e bo lts were co nce ntrated aro und th at
one target. It has been suggested th at whe n th e defe nde rs of Hod Hill refused
to sur rende r, the Rom ans dem on str ated the efficiency of th eir siege train by
destroying th is one hut - thus prompting th e ga rrison to open their gates. As
th ere is no othe r evide nce of battle thi s remai ns a plausible explana tion of
what happen ed, but as usual th e eviden ce is open to int erpretation . Certa in ly
Vespasian considere d the site to be impo rta nt: he orde red the build ing of a
Rom an aux iliary cam p in the nort h-east co rner of the Iron Age defen ces.
Ano the r site worth noting is th at of Burnswark in Dumfr ies, a hill-fort flanked
by th e remai ns of two Rom an siege camps. Wh en the site was first excavated
in 1898 it was assumed th at Burnswark had been besieged by the Rom an s, whose
camps were built with in siege-engine range of th e ramparts. However, more recen t
excavations conducted in th e 1970s have proved that th e Rom an siegewor ks were
50
built after th e fort fell int o disuse, almost certai nly providin g a training groun d for
the Roma n troops statio ned in southern Scotland during th e late 1st century and
early 2nd century AD. In effect th e hill-fort had becom e a Rom an firing rang e. Th is
suggests th at th e Rom an s took th e reduction of hill-forts seriously, possibly as a
means of preparing for campaig ns against the un-pacified Celtic tribes to the
north . These works are in stark con trast to the lack of Roman siegeworks in
the rest of Britain, whic h suggests th at if th e Rom an s ever enco untered a fortress
that defied th em , th ey would lay siege to it in accordance with th eir milita ry
doctrine, establishing secure camps from wh ich to bombard th e defences.
Once the defenders were driven from th e ramparts th e Rom an s would probably
send in auxiliar y troops to secure the fort, holding their veteran legiona ries back
as a reserve.
This is how th e Rom an s fough t at Mon s Graupius (AD 84) agai ns t th e
Caledon ii, or wh en th ey pacified Gaul, judaea and Dacia. The lack of fort ified
camps in th e British Isles suggests th at instances of resistan ce agai nst th e
Romans were rare. It is more likely tha t the methods suggested at Hod Hill - a
demonstration of Roman mil itar y might - were sufficie n t to force the sur render
of mos t of Brita in 's Celtic fortresses. Wh ile it is appealing to imagi ne th e
defen ce of sites suc h as Maide n Castle as roma ntic last stan ds by th e Celtic
in habitan ts in Britain , th e truth was probabl y mu ch more mundan e. Faced
with the futili ty of resistan ce, th e defenders made peace with th e in vad ers, so
bringing the era of their political and military independence to a close .
51
Aftermath
The Rom an invasion of Britain in AD 43 sou nded th e dea th kne ll for Celtic
culture in southern Britain . We have already seen how h ill-forts like Maide n
Castle continued to be used for two or even th ree deca des afte rwa rds, befo re
th ey were repl aced by a Rom an provin cial town a few mil es away. In the case
of Maid en Castle th ere is ev ide nce that th e pop ulat ion began to drift awa y
from th e old fortified settlem ent durin g this period, so that in the space of one
o r two gene rations th e population dwindled awa y until on ly a few inhabitants
rem ain ed . The Rom an s brou ght peace in th eir wake (if we ign ore th e lceni
revo lt of AD 6 1- 63), and th erefor e th e British populati on had less need to
pro tec t th em selves and th eir co m m u n ities. Th e Rom an policy of sligh ting th e
defen ces of th e hill -forts th at submitte d to th em or were cap tured mad e these
places ind efen sible, and speeded th e drift towards n ew settleme n ts. However,
th e proc ess of pacification was not com pletely smooth . Archaeologists have
sho wn th at th e gateway of th e hill -fort of South Cadbury in Som erset was
dest royed aro un d AD 70- 80, wh ich sugges ts th e fort rema in ed occupied after
AD 43, but th at it was prob ably attac ked and destroyed by th e Romans in what
might well have been a punitive and ret aliator y action . It appears th at th e fort
was th en co m plete ly abandoned until th e Rom an occupation of southe rn
Britain ca me to an end around AD 410 .
The pattern appears fairly clear. In areas where th e Rom an s bro ugh t th e
popul ati on directly under th eir co ntrol the o ld fortifications fell into disuse - if
not im me diate ly, th en with in less than three generations . In th e north an d
west of wh at is n ow Eng land, wh ere th e Roman izat ion of th e populati on was
less th orou gh , th e occ upa tion of some forts co n tin ued for man y decad es, and
in so me cases th rou gh out th e Rom an period . Un fortu nate ly we know all too
littl e about thi s period of occ upancy, including wh ether these sites retained
th eir olde r levels of pop ulatio n and sen se of com m un ity, o r if their po pulation
declined almost as d ramatica lly as th at of th e hill -fort dwe llers of southe rn
Eng lan d. Some hill-forts were certain ly eithe r abandoned or else changed th eir
role fro m fortified sett leme n ts into fortifi ed farmstea ds, with a co nsequent drop
in populati on. The process of Rom an ex pansio n was a gradual one, as Agricola
on ly pu sh ed into wha t is now southe rn Scotl and in AD 82-83. By th en the
stro ngho lds of th e Brigant es of toda y's northern England had fallen un der
Roman co ntrol, so the Celts of Scotland and Wales rem ained th e o n ly trul y
in dependent in d igen ou s tribes in th e British Isles.
52
53
LEFT
cam ps, where the Roman garri son in Scotland could practise
siege techniques.They would th en employ these skills du ring
o ne of several punitive Roman exped itions into northern
Scotland , or when called upo n to quell a revo lt further
south. The hill-fort defences were des igned around the
slingshot - the standa rd defensive missile we apo n of
Ancient Britain.At Burnswark caches of clay slingshot stones
we re found - akin to mo dern dummy bullets.This suggests
th e Romans used auxiliaries to represent Celtic defende rs
during the ir moc k attac ks on Burnsw ark fort.Also shown
are two types of Roman ballista (to p left and right).
exam ple of Hod Hill, wh ere a Rom an fort was built in to one corne r of th e old Iron
Age earthwor ks. Whil e thi s was probably a matt er of prim e locat ion and eco no my
of effort, it certain ly appea rs th at th e Rom an s were aware of th e th reat posed by
the occupancy of h ill-forts by hostile tribesmen or rebels, and train ed th em selves
to deal with just such a situation.
Fina lly we have already seen how some hill -fort s such as Dane bury in
Ham pshire served as religiou s as well as politi cal cen tres. It is th erefore not
co m pletely sur prising th at after its fall Daneb ury rem ained in use during the
Roman period , as evidenc ed by cartwheel tracks and Romanized fin ds discovered
on top of th e olde r layers of occupancy and defenc e. It ha s been suggested that
the fort housed a sma ll religiou s com mun ity, wh ich remained in residenc e to
maintain th e sanctity of a sacred site. A sim ilar site may well have existed in
Maiden Castle, as th e older Rom an o-British Ch ristian tem ple built th ere might
well have been placed on th e site of an older place of wo rsh ip.
These Iron Age hill -forts came in to th eir own agai n after th e Rom an s left
Britain. Duri ng th e 5th and 6th cen tur ies AD several were re-occu pied, as the
defenceless towns of Sub-Roman Britain were vu lne rable to attack by a grow ing
number of in vaders. However , it appears th at the peop le of Sub-Roman Britain
appeared well aware of the defen sive possibilit ies of sites such as South Cadbur y
and Castle Dare, bot h of whic h were re-fortified duri ng wha t is some times
known as th e 'Arth urian' period. In less Rom ani zed areas, fort s such as Dinas
Em rys in Wales and Dun Eidi n (Edinburg h) or Traprain Law in Scotland becam e
im portant political cen tres once again - which mu st be left fo r ano the r study.
55
Scotland
Fortified sites
The Brach of Mousa, Shetland
The finest surv iving broch struc ture, sta nding ove r 13m high . Owne d by
Scottis h Herit age. Located on th e island of Mou sa, accessible by ferry boat
from San dwick, 14 mil es south of Lerwick, She tland. See th e Historic Scotl and
website (www.h istoric-scotkmd.gov.ulc for ferry inf ormati on an d o pen ing times.
Altern ati vely call the Histor ic Scotl and office in Skara Brae, Orkney for up-to-date
informati on: (01856) 84181 5.
Click h im in Brach, Shetland
Broch tower and associated settleme nt and oute r defen ces. Owned by Scottish
Heritage. Located one mile south-west of Lerwick, Shetland. See th e Visit Shetland
website (www.visitsnetland.comi for open ing tim es and contact inform ation .
56
An Iron Age hill-fort th at was used by th e Rom an s as a military train ing gro und .
In private hands but accessible to th e public. See the following websites for mor e
57
England
For t ified sites
Ma iden Castl e, n ear Do rc heste r, Do rset
Th e largest Iron Age h ill-for t in Britain , the imposin g fortific ations of Maid en
Castle were excavated during the 1930s and 1980 s. The site is now maintain ed by
Eng lish Herita ge. Maid en Castle is o pen th rou gh out the year, and a self-guided
trail is provided . Website: 1V1V1V.englisll-lleritage.org.uk/server/sllOlV/ CoIIProperty.279.
Danebury, n ear Stockb ridge, Hampshire
Th e Iron Age hill -for t of Dan ebury was exte ns ively excava ted over some
20 yea rs, making it th e most closely stud ied hill -fort site in Britain . Danebury
is now main tained by Ham pshire County Co uncil.
The site is open th roughout th e yea r, and a selfguided trail is provided, Website: 1VIVIV.l1l1I1ts.gol'.uk
/countryside /da nebury/in dex.html.
Stanwick, n ear Forcett, No rt h Yorks hire
Excava ted by Sir Mortimer Wheeler in th e early
1950s, Sta n wic k wa s t h e Late Iron age ca pita l o f
the Brigantes, the most important tr ibe in preRom an no rthern Brita in . The oldes t sectio n o f th e
3 1O-hectare site is n ow main tained by Eng lish
Heritage. The site is o pe n th rou gh out th e yea r, and
a self-gu ide d trail is provid ed . Web site:
Il'l vlV.ellgl isll-l leritage.org.llk!selv er/sllOlV/ oIIProperty.384.
58
site of a Norma n castle. The site is now maint ain ed by English Heritage. Old
Saru m is open th rou ghout th e year, and a self-guide d trail is provided. Website:
IV\I'l I'.el lglisll-l leritage.org .llk!sen 'er/ sl lOw/ CoI1Property.293 .
59
Museums
60
W a le s
Fo rti fied sites
Tre' r Ceiri, Llyn Peninsu la , Gwynedd
A specta cular sto ne -built hill-fort. Full public access via footpaths. Websit e:
www.penl lyn.com /I /sIlery/il i thfaen/S ,II tm l.
Din as Em rys , Ffest in iog, Gwynedd
A small hill-fort, wit h sto ng post-Roman links. Privat e ow nersh ip, but access
available by perm ission from local farm er. Websi te :
IV IV IV. vort igen 15 tudies.org.uk /a rtci t/ di na s.litm .
Caer-y-Tw r, n ea r Ho lyh ead, Anglesey
A sm all but im po rtan t hill- fo rt, in pr ivat e ow ne rsh ip but access ible to th e
public via footpaths leading up Holyh ead m ountain from th e town. Websit e:
1V1V1V.lIl egaI i til ic.co. uk/article.phpr sid -ci 955.
Pen Dinas, Aberystwyt h , Cered ig
A hill- fort ove rloo king the m od ern town of Aberys twy t h. Public access.
Website: www.lValespast.colll/articfe.sll tlll l?id= 4 1.
Castell He nllys, Meline, Crymych , Pem b rokesh ir e
A pa rtially reconstructed Iron Age hill- fort wit h reconstructed round ho uses.
Own ed by th e Pembroke Coast Nationa l Park . Shop and sma ll interp retation
cen tre on site . Call (0 1239) 89131 9 for details. Websi te: www.castell henllys.com ,
Museu m s
Th e Na tional Muse um of Wales, Ca t h a ys Pa rk, Ca rd iff
The mu seum displays cover the Iron Age in Wales, and th e collection incl udes
artefacts recover ed from hill- fort s. Ope n daily (except Mondays) from lOam to
Spm. For further information visit th e mu seum website:
www.m useum wales.ac.uk/en/ho me.
61
Bibliography
Arm it, Ian Towers in tile North : Tile Brochs ofScotland
(Stroud, Glouceste rshi re: Tempus Publis hi ng, 2003)
Arrnit, Ian Celtic Scotland (London : B.T. Batsfo rd for
Histor ic Scotl and, 2005)
Barr ett, [ .C. et al., Cadbury Castle, Somerset (Lo ndo n:
B.T. Bat sford for Eng lish Heritage, 2001)
Bradl ey, R. and Ellison, A. Rallis Hill - British
A rchaeological Reports No . 19 (Ox for d:
Arch aeo press, 1975)
Ch ad wick, No ra and Cun liffe, Barry Th e Celts:
A Penguin History (Londo n: Penguin, 199 7)
Cu n liffe, Barry and Miles, David Aspects of the Iron
Age in Central Sout h Britain (Oxfo rd : Oxford
University School of Archa eo logy, 1984 )
Cun liffe, Barry Iron Age Communities in Britain
(Londo n : Routledge, 1991; 3rd edition)
Cun liffe, Barry Danebury (Lon do n: B.T. Bat sford for
Eng lish Heritage, 1993)
Cun liffe, Barry Iron Age Britain (Londo n: B.T. Batsford,
1995 )
Cun liffe, Barr y The Ancien t Celts (Oxford : Ox ford
Un iversity Press, 1997)
Cun liffe, Barry Facing the Ocean: Tile Atlantic and its
Peoples, 8000 BC to AD 1500 (Oxford: Ox ford
Un iversity Press, 200 4)
Cun liffe, Barry Iron Age Com m un ities in Britain: An
Account of England, Scotland and Wal es [rom the
Seventh Century BC III I til the Rom an Conquest
(Londo n: Routledge , 2004 )
Dyer, James Penguin Guide to Prehistoric England and
Wal es (London: Penguin , 1981 )
Dyer, James Hillforts of England and Wales (Risbo rough,
Buckinghamshire: Shire Publications, 200 3)
Foj ut , Nopel The Brochs of Gurness and Mid howe
(Ed in bur gh : Historic Scot land, 2001 )
Forde-Iohnston, James Hillfort s of the Iron Age in
England and Wa les: A Survey of the Surfa ce Evidence
(Live rpo ol: Un iversity of Liverp ool, 1976)
Guilbe rt, G. (ed .) Hill- fort Studies (Leiceste r: Leicester
Unive rsity Press, 1981)
Harding, D.W. The Iron Age in the Upper Thames Basin
(Oxford : Oxford Un ive rsity Press, 19 72)
62
Glossary
An ext ension to the earth wo rks of an Iron Age
fort. ofte n built as a later addit ion to the
fort ificat ion system.
Bank
In terms of hill-forts. these are often associ ated
with ramparts. although more accurat ely the
latter rep resents the final bank before the inne r
enclosure. Banks were usually but not always
built behind a ditch. from which the so il for
the bank was excavated .
Berm
A flat space between the foo t of a bank and
the start of a ditch .
Bivallat e
An Iro n Age fortification system where t he
central enclosure is surrounded by two sets
of banks and associated ditches.
Bronze Ag e The pe riod from around 2100 BC unt il
700 BC when bronze was produced by the
indige no us peoples of Britain.
C om pact
A bivallate, tr ivallate o r multivallate hill-fort where
the systems of banks and ditches are close
toge ther - usually within 10m of each ot her.
C ont our for t The technical term for a hill-fort built to
take advantage of the contours of a hill.
Invariably the shape of the fort follows the
contou r line. producing an irreg ular shape to
the fortification.
C ount e r scar p The exterior slo pe or wall of a ditch. which
in t he case of hill-forts was sometimes revet ted
using stone or timber.
D isper sed A bivallate, tr ivallate or multivallate hill-fo rt where
the systems of banks and ditches are well spaced
out - usually mo re than 10m from each other.
Ear t hw ork An earthen em bankment. part of a fortification.
In most cases a bank or ram part is classified as
an earthwork.
G lacis
The slo pe extending down from the outer
works of a fortifi cation over which an attacker
would have to move as he approached the fort.
H ill -for t
A defensive earthwork o r stone-built Iron Age
structu re built on an easily defens ible pos ition.
usually the plateau or summ it of a hill.
Iron Age
The perio d from around 700 BC until t he
Roman co nquest of Brita in in 43 AD whe n the
inhabitants of Brita in pro duce d and use d iron .
M ult iple enclosu re fort A form of earthwork or
fortification where the defences form a
netwo rk of banks and sometimes ditches. In
terms of Iron Age fortifications it is generally
presumed that these sites were non-military in
nature. and the mult iple e nclosures co ntaine d
Annex
63
Index
Figures in b o ld refer to illustrations
agriculture 4 1, 60
ballistas 54
banks 8,9, 16,25
Bar ry Hill 12
Beacon Hill 16, 19
Blac kbury Camp 59
Brat ton Camp 59
Bred o n Hill 44
bro chs
definition 4. 12
design 20-2
e ntrances 23,40.43-4,50
. interiors 43.45.47
lo cati on I I , 13
see also individual brochs by name
Brown Caterth un 57
Burghead 8,8
Bur nswa rk 50-1 ,53 ,54.57-8
Butse r Ancient Fa rm 60
Cabu rn 14-15
Caer-y- Twr 53, 61
Caesar. Julius 4-5, 47- 8
Cast ell He nllys 61
Celts: defi nition 4-6
chevaux de frise 45
Cissb ury 43
Cl ickh imin 50,56
Clovelly Dykes 9-12
cr annogs 8, 58
Cun liffe , Profes sor Bar ry 24,25, 50
cup a nd ring mar ks 55
daily life 29
Dan -y-Coed 12
Daneb ury 27
history of 24-9.42- 3.48.50,55
no wa days 58
de fe nce principles 42-5
Dinas Emrys 61
Dinorben 42.44,46
ditch es 17, 25
d ruid ic prac tic es 16
Du n Ca rlow ay (C har labha igh) 48 ,5 7
Du n Telve 5 1.57
duns 13
Du nsap ie C rag 37
Du nsinane Hill 28
Dyke Hills 9
Eildo n Hill No rth 21 ,57
Cabu rn 15
co nst r uctio n 17
Danebu r y 26-8
defending 43-5
Dinorben 44.46
Lo rd e nshaws 29
Maiden Castle 33, 33, 38- 9,44, 52
O ld O swest ry 18
St. Cat he rine's Hill 17. 45
Uffi ngt o n Castl e 7
Glastonbu r y lake village 8,61
Glen Lyo n ring forts 12
gra na r ies 28
G urn ess 20.21, 22,42-3 ,57
Hallstatt.Au st ria 5
Hambledo n Hill 56
Hardy.T hom as 3 I
hea dland forts 8-9
Hembu ry 44
hill-forts
co nst ruct io n 15-20
defending 45-5 1
definit io n 8-9
form and func t ion 14-15
lo catio n an d site 10. 12- 13, 16
see also individual hill-forts by name
Hod Hill 42. 45. 50. 55. 58
Iron Age
A an d B Cultu re s 32
C Cultu re 33
definition 5-6
jave lin head s 32
Ladle Hill 16-20
Lhuyd, Edward 5
locatio n 10-11 ,1 2-13
Lo rd en shaws 29,35-6
Maide n Castl e
co nstructio n 20
as econom ic and pol itic al ce nt re 4 1
fo rtifi cations 30-3 ,3 I. 33 , 38-9. 44.
52-3
nowadays 58
and the Ro mans 33, 48-9, 52. 55
as se ttl em e nt 35-41
manufactu ring 41
Midhowe 21. 22. 40, 57
Mit her Tap 13
Mons G raupius, Battle of (AD 84) 13,5 1
Mou sa 20. 21.44-5 ,47,56
muse um s 58, 60-1
Flag Fe n 60
64
palisades 17. 42
Peat Moors Centre 61
Pe n Dinas 61, 61
Piggo t t. Professo r I6
plateau forts 9
promonto ry forts 8-9
Rainsborough 9
Rame Head 8
ram par ts
Cabu rn 15
co nstruction 42-3
Dan eb ury 25- 6. 27
defin ition 8
Dino rb e n 46
Duns inane Hill 28
Maide n Castle 53
Mith e r Tap 13
Uffington Castle 5-6
Wh ite Cat e r t hu n 25. 28
W od e n Law 9
Richmond. Sir Ian 50
ring fo rts/raths 12
Ro man s in Br ita in 13.33.48-56
rou ndhouses 37-40,37-9.60.6 1
St. Ca theri ne 's Hill 17.22,45
settle me nts 28-9. 35-4 1,38- 9. 60
Sha rples. Niall 40
shrines 28-9
sieges 24.45-51 ,54
slingshot st ones 53
Somerset Levels 61
South Cadb ury 44,52
Stan wic k 58
sto ne shot 25
La Ten e, Switze rland 5
Trap rain Law 30.53,57
Tre'r Cei ri 34 , 43.53.6 1
Uffington Castle 4-7,60
valley fo rts 9
Vespasian . Rom an em pe ror 33.48-9.
50
vitrifie d forts 13,43
W andl e bury 43
W arham 60
wea po ns 32,32. 45-7
W he ele r, Sir Mortimer 30, 31, 4 1. 48-9.
53
Wh ite Cate r t hu n 25.28,5 7
Wi nkleb ury 59
W oden Law 9,20,57
Woo ds ide 12
Wo rleb ury 43
Yarn bury 57
Related Titles
ISBN
SERI ES
No. TITLE
1 841 764302
Fortress
1 841 768324
Fortress
31
1 84176 7824
Elite
126
1 8417 6 3926
42
085045606 I
Men-at-Arms
158
1 841 7648 7 6
Men-at-Arms
374
1 84 176559 7
Me n -at -Arms
390
1 841 76 6054
New Vangua rd
78
1 841 76 6348
New Vangua rd
89
18553241 99
Warrior
1 85 5325 675
Warrior
IS
1 841 76 14 3 5
Warrior
30
Ce lt ic Warrio r
1 84 1763462
Warrior
50
1 84 1766003
Warri or
71
1 841 76 6011
Warrior
72
www.ospreyp ublishing.co m
To order an y of these titles, or for more informati on o n Osp rey Publishing, con tac t:
Osprey Direct (North America) Toll free: 1-866-620 -6941 Fax: 1-800-659-2436 E-mail: info@ospreydirect.com
Osprey Direct (UK) Tel: +44 (0) 1933 443863 Fax: +44 (0) 1933 443849 E-mail: info@ospreydirect.co .uk
www.o spreypublishing.com
The Forts of
Photographs
Full colour artwork
1--'
. --- 1
T
SCOTlAND
-=-0::-.-- -
NOR T H
SEA
IRELAND
__ '1'- -
Unrivalled detail
Co lour maps
OSPREY
PU BL IS H I NG
www.ospreypublishing.com
78 18 46 030642