Sie sind auf Seite 1von 21

A Survey of the Roman Fort and Settlement at Birdoswald, Cumbria

Author(s): J. Alan Biggins, David J. A. Taylor, B. Coxon, B. Esselmont, A. Frank, C. Hudson,


P. McCloy, E. Montgomery, J. Robinson
Source: Britannia, Vol. 30 (1999), pp. 91-110
Published by: Society for the Promotion of Roman Studies
Stable URL: http://www.jstor.org/stable/526674 .
Accessed: 08/10/2011 12:18
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Society for the Promotion of Roman Studies is collaborating with JSTOR to digitize, preserve and extend
access to Britannia.

http://www.jstor.org

and

Survey

of

the

Settlement

at

Birdoswald, Cumbria

Fort

Roman

By J. ALAN BIGGINS and DAVID J.A. TAYLOR'


with contributions by B. Coxon, B. Esselmont, A. Frank,C. Hudson, P. McCloy,
E. Montgomery, and J. Robinson
INTRODUCTION

he fortat Birdoswaldis situatedonthe lineof Hadrian'sWall,some25 km(15/? miles) east

of Carlisle (FIG.1). The site lies between the forts of Great Chesters 10 km (61/3miles) to
the east, and Castlesteads 12 km (71/3 miles) to the west.
The strategic importanceof the site, which shows evidence of possible Neolithic activity2on the
promontoryto the east, was recognised by the Roman army.A signal-tower to the south-east of the

Birrens

"l

Bei.a

CO

e'

Netherby
.SS

~e5chester

s~

Lanhe

C26

IQ"

FIG.1. Mapshowingthe locationof the site.

The surveywas undertaken


by the authorsas a joint academicresearchproject.J.A.B.was readingfor an MA in
ArchaeologicalSurveyat the Universityof Durham,andD.J.A.T.was investigatingthe stonebuildingsin the fortson
Hadrian'sWallfor a PhD,also at the Universityof Durham.
2 Wilmottforthcoming(a).
I

92

J. ALAN BIGGINS and DAVID J.A. TAYLOR

present fort probably formed part of the Trajanic Stanegate communication system,3 and there is
evidence of at least one earlier fort beneath the extant stone fort.4 Milecastle 49 is situated one
third of a mile (530 m) to the east on the edge of the escarpment, and overlooks the important
bridging point over the river Irthingat Willowford.5The Maiden Way ran northfrom the fort to the
outpost fort at Bewcastle.
The site is centred upon NY 616 664, at OD 165 m, and comprises the fort itself with evidence of
settlement to the south, east, and west. The remains of the curtain-wall surroundingthe fort are
upstanding,with the exception of the greaterpart of the north wall which is overlain by later farm
buildings and the modern road. The north-west corner of the site is occupied by a farmhouse and
ancillarybuildings, some of which are now used as a visitor centre. The consolidated remains of the
portae principales, porta decumana, horrea, fabrica, and part of a basilica6 can be seen. The
surroundingfields do not contain any visible excavated buildings, although field boundariesdefine
the fort to the east and west, and there is a small wood to the north-west.Several mounds can be seen
in the field to the east, some of which may in partcomprisethe spoil from earlierexcavations. Those
runningeast-west, however, are probablythe result of laterfield clearancealong the line of Roman
features. The southernboundaryof the site is defined by the top of a steep escarpmentand there is
evidence of considerable landslip in the past, as a result of which some of the formersettlement has
been lost.7 More recent landslip has occurred because of the instability of the ground due to the
waterlogged ditches of archaeological features. To the north of the fort and Wall lies Midgeholme
Moss, and a Roman cemetery is sited some 350 m to the south-west, south of the Military Way.8
The Wall, in its original form, comprised a stone barrierto the east of the river Irthingand a turf
barrierto the west with milecastles; the intervening miles each contained two turrets. Turret49a
was sited in the centre of the fort, and was constructed of stone, as were all the turretsto the Turf
Wall. When the decision was taken to position forts on its line, a fort with turf rampartswas built
astride the Wall. It was positioned so that the Turf Wall abutted the south guard-chambersto the
portae principales, and built over the Wall ditch. At about this time the vallum was constructed,
almost certainly by cohors I Aelia Dacorum (RIB 1365), and re-filled within a very short period.9
Some time after this the further decision was taken to enlarge the fort and rebuild it in stone. It
would seem that, once this work had started, it was resolved to realign the Stone Wall on the north
side of the promontory, which would have greatly increased the area available for the military
annexe and vicus. The timetable for these events as suggested by Wilmott1o(taking into account
Breeze and Dobson, and Stevens) indicates that the Turf Wall was completed by 123, by which
time the primaryforts and vallum were underconstruction, with the latter being completed by 125.
Occupation of the fort occurredaroundthis time, and is thought to have been by a cohors milliaria.
The construction date of the stone fort is not known, but it concludes with the blocking of the
portae quintanae, made redundantwhen the Stone Wall joined up with both northerncorners of
the fort. Shortly after construction commenced there was a period of abandonment11during the
second century with no evidence of destruction. The consolidated horrea were built early in the

3 Woolliscroft 1989; Daniels 1978, 206; Shotter 1996, 41-54; although Breeze and Dobson 1991, 19-26, do not
consider the evidence satisfactory.
4 Wilmott 1997, 41-5.
5 Bidwell and Holbrook 1989, 50-98.
6 Wilmott 1997, 97-8; the basilica at Birdoswald is as yet unique in an auxiliary fort.
7 Simpson and Richmond 1932, 144; Wilmott forthcoming (b).
8
Wilmott 1993.
9 Wilmott 1997, 44-5.
10 Wilmott 1997, 40-4.
11 Wilmott 1997, 73-9.

AT BIRDOSWALD
A SURVEYOF THEROMANFORTAND SETTLEMENT

93

third century (RIB 1909). Considerable rebuilding occurred in the late third or early fourth century
(RIB 1912), probably undertaken by cohors I Aelia Dacorum.12 By the mid-fourth century the
patternof activity in the fort had changed, which is illustratedby the collapse of one of the horrea,
whilst the other was put to a different use. Evidence of sub-Roman activity on the site can be seen
in the reuse of an horreum as a hall and this period of activity could have a seventh-century
terminal date. At no other site in northern Britain has such tangible evidence for continuity of
settlement been identified.13Documentary evidence for the post-Conquest period is summarised
by Wilmott,14and shows that a tower house and a bastle were erected on the site in respectively the
medieval and post-medieval periods. The present farmhouse dating from the seventeenth century
was extended and modified in the eighteenth century.
The archaeological history of the site has been comprehensively summarised by Birley,15
Daniels,16 and lately by Wilmott.17The first importantexcavation took place on the site between
1850 and 1852 when H. Norman, the owner of the site, and H.G. and W.S. Potter18excavated the
portae quintanae. During four seasons of excavation F. Haverfield19identified the Turf Wall and
the vallum, and, importantly, that the fort was built across the line of the Turf Wall. Seven
consecutive seasons of work by F.G. Simpson and I.A. Richmond20took place between 1927 and
1933, during which extensive excavations to the south of the fort identified both prehistoric and
Roman ditches, and defined the line of the vallum. Work inside the rampartstook place in the
retentura and in the east of the praetentura where a possible barracks was identified. During
consolidation of the gates and rampartsbetween 1949 and 1952, including the porta principalis
dextra, some limited excavation was conducted by J. Gillam.21 Major excavations took place
between 1987 and 1992 by T. Wilmott22for English Heritage in the south-western sector of the
praetentura.

The porta principalis

sinistra and a section of curtain wall to the north were

excavated, together with the two horrea, part of the basilica, and adjoining buildings to the south.
Work in the final year exposed the porta quintana dextra and a section of the adjacentcurtainwall.
Excavations to the south of the fort by T. Wilmott in 199623revealed a putative Neolithic burial,
and re-examined the results of Simpson and Richmond's campaign. This work also established the
extent of the erosion threat on the spur end.
THESURVEY

The survey was carriedout between May and October 1997. The original intention was to carryout
a topographical and geophysical survey limited to the fort and its immediate defences, but, in view
of the significance of the results, the scope of the survey was extended. Limited resources only
permitted magnetometry survey outside the fort.

12 Daniels 1978, 205.


13 Wilmott 1997, 408-9.
14

Wilmott 1997, 410.

15
Birley 1961, 196-203.
16 Daniels 1978, 198-206.
17 Wilmott 1997, 1-14.

18 Potter 1855, 63-75.


19 Haverfield 1897a, 185-97; Haverfield 1897b, 413-33.
20 Richmond 1929; Richmond and Birley 1930; Richmond 193 1;
Simpson and Richmond 1932; Simpson and Richmond
1933; Simpson and Richmond 1934.
21 Gillam 1950.
22 Wilmott 1997.
23 Wilmott forthcoming (a) and (b).

J. ALAN BIGGINS and DAVID J.A. TAYLOR

94

15

f
Outbuildings

155

156

\
156

Farmhouse
156

5157
156
Granaries
156

157

156

158

157

157

15915

160

158
16

162
159

16.161

161

50m0

50m
FIG. 2.

Topographicsurvey.

160

A SURVEY OF THE ROMAN FORT AND SETTLEMENT AT BIRDOSWALD

95

A close contour topographicalsurvey was carriedout within the areaof the fort (FIG.2).24 Whilst
the general contour configurationis likely to apply from Roman times, the present-daysurface may
differ substantiallyin detail. Significant till wash has occurredfrom the formerlyterracedcentre of
the site, as is apparentby the presentexposure of the primarygate sills to the porta decumanaand the
south wall of the principia, which is recorded as standing thirteen courses high,25and now stands
just above present ground level. The advantageof close-contour survey is that certain elements of
micro-topography,not readily visible from the earthworksurvey (FIG.3) and often concealed by
vegetation, become apparent.It became evident that the ridge and furrowwas largely confined to the
southern face of the fort, i.e. the retentura, but perhapsthe most strikingremains were elements of
the principia. This is situated in a slight saddle formedby the ruinedmasonryin the low-lying centre
of the fort from which laterterracingranto the northand south. The lowest partof the fort was in the
north-eastcorner where the latrineswere probablysituated.
A resistivity survey was carried out only within the area of the fort (FIG.4).26 Because of the
slower application speed it was restrictedto the interiorof the fort. In many respects the resistivity
survey was disappointing and did not produce the response expected in the presence of so much
masonry; this is probably due to the depth of the overlying rubble especially in the retentura.
Significantly the more prominent earthwork features gave the better response, no doubt due to
their improved drainage. The survey did, however, provide valuable additional evidence and was
particularlyuseful in determiningthe extent of a probable Romandrainagesystem and the location
of a number of kilns.
The magnetometry survey (FIG.5) included the fort, together with an area 80 m to the west and
120 m to the east.27 The survey of the fort, subjected to so many building phases, was never
expected to prove a simple prospect. The net effect was a palimpsest of features, particularlyin the
southern part of the fort, which seems to have undergone the most intensive reconstruction. The
superimposition of later ridge and furrow complicated the analysis.
INTERPRETATION(FIG. 6)
THE TURF WALL

The Turf Wall (1) was built from a point west of the river Irthingto Bowness-on-Solway. It has
been demonstratedthat the Wall was c. 6.10 m wide at its base, and, close to the fort, was protected

24 The survey used a Wild TC1010 total station from a


single station point by the porta decumana. The walls and
consolidated masonry were also surveyed but not included as contourable points. Data was downloaded onto a CAD
programme (Liscad) and after data manipulation was transferredto EasyCad as an industry standarddxf file, which was
then smoothed using Cubic B Spline transformation.The high density survey (c. 3000 readings per ha) was designed to
emphasise micro-topographical anomalies. The north-western part of the survey area, including the farm buildings and
conjectural structures, was appended from other sources, including the RCHME 1987 survey and the most recent site
excavations.
25 Richmond 1931, 127.
26 The
survey was conducted at 0.5 m sample and 1 m zigzag traverseintervals using a Geoscan RM15 resistivity meter
which had a twin electrode configuration and a probe separation of 0.5 m. Data was treated as magnetometry, except that
desloping was performed.
27 This survey was conducted using a Geoscan FM36 fluxgate gradiometerusing 0.5 m sample and I m paralleleast-west
traverses, with the application of drift correction. Data was downloaded and analysed using both Geoplot 2 and Insite
software. Inherentprogrammelimitations with Geoplot indicated the use of Insite for larger survey areas. Data processing
used the stages; edge matching, despiking at a tolerance of + or - 2 S.D.'s, after which it was smoothed by interpolation
using a factor of 0.125. It was found necessary to utilise a numberof different compression and contrastvalues in order to
visualise features of different relative intensities.

J. ALAN BIGGINS and DAVID J.A. TAYLOR

96

.-

----

i !1i

-q

--

"...
. ...
.... ?..--F

"'

" .

i
-

Ram

"-

'

rr-

Th
-"

,"
., .:

.--

rr
:!-

. . .
. . ..
.. . . .. .

"----_
--'....
-,,

I"

lllil

Ei

,I--:

50mr
-(
:

".

....I'-Jo
.

' r-

""0m

50m

FIG.

3.

Earthworksurvey.

A SURVEY OF THE ROMAN FORT AND SETTLEMENT AT BIRDOSWALD

Outbuildings

Farmhouse

Granaries

50m

-ww

FIG.4.

Resistivity survey.

97

J. ALAN BIGGINS and DAVID J.A. TAYLOR

98

br

A.

r
7IO

FIG.

5. Magnetometersurvey.

o-5

A SURVEY OF THE ROMAN FORT AND SETTLEMENT AT BIRDOSWALD

99

to the north by a ditch 8.23 m wide at the top and 2.74 m deep.28The ditch at Birdoswald was found
to be 7.70 m wide and a projected 3.80 m deep.29The original, likely intention of the fort builders
was that the Stone Wall would follow the same line, and abut the Hadrianic fort at the enlarged
southern guard-chambersto the portae principales. These two guard-chambersare both 5.48 m
long, while both northern guard-chambers have contracted lengths (4.69 m and 4.64 m). At
Chesters there is a slight increase in the width of the guard-chambersto the gates of the portae
principales when comparedto the other two main gates. The probable explanation for the increase
in size at Birdoswald is that the Turf Wall was in situ at the time the construction of the stone fort
commenced.30
Gillam31 found that the north portal and north guard-chamberof the porta principalis dextra
were built precisely over the infilled ditch of the Turf Wall. Consequently, the line of the via
principalis and the eastern exit road appearsto have been close to the inner edge of the Wall Ditch.
This is supportedby the discovery of the southern edge of the Turf Wall 6.70 m south of the south
portal of the porta principalis sinistra,32 additionally conforming with a typical berm width of c.
1.80 m. Evidence of the ditch can be seen as an anomaly in the east of the fort north of the via
principalis. The route of the roads leaving the fort to both the east and west does not quite follow a
straight line. The road leaving the fort to the east is at three degrees less than a right angle to the
curtain wall, whilst the road leaving to the west is two degrees less than a right angle. It is
probable, therefore, that there was a realignment of the Turf Wall at the west gate as indicated by
Wilmott,33but not to the extent shown.
THE DEFENCES

Two clearly defined ditches (2) can be seen to the west, c. 13 m apart and with the centre of the
inner one c. 9 m from the face of the stone fort wall. There is some evidence of later recutting as a
further section of ditch can be seen to the south-west of the outer ditch. Three ditches were found
on the west of the south-west comer during the 1996 excavation;34all were primaryand the outer
two cut the vallum backfill. The ditches to the east (3) are much more complex. Simpson and
Richmond35found two ditches to the north of the Turf Wall ditch, the outer one of which, c. 3.30 m
wide with its inner edge c. 15.20 m from the face of the curtain-wall,ran into the Turf Wall ditch as
if contemporarywith it. A furtherditch was found to the south of the Turf Wall only c. 1 m wide,
with its inner edge c. 13.70 m from the curtain-wall. This stopped c. 15 m short of the centre of the
porta principalis dextra, and it was considered by Simpson and Richmond to have been cut in
association with the Turf Wall ditch, so as to leave a gap for the construction of the Turf Wall. The
ditches immediately to the south of the gate and the line of the Turf Wall are confused, and seem to
be made up of a single ditch close to the curtain-wall, with a group of three ditches furtherto the
east, one of which must be that described by Simpson and Richmond above; none of these appear
to run parallel to each other. To the south of these, two ditches are clearly defined. The ditches on
the east are seen to run up close to the south of the Stone Wall.

28 Daniels 1978, 20.


29 Wilmott 1997, 47.
30 Wilmott 1997, 89-90.
31 Gillam 1950, 65.
32 Wilmott 1997, 90.
33 Wilmott 1997, 90 fig. 59.
34 Wilmott forthcoming (b).
35 Simpson and Richmond 1934, 127-9.

31

.U...
..16

_i

7I1

18

14
1 7
1

Li

9.........
L :

..

rr

2
50 m

24.

FIG.6.

the drawingincludesevidence derivedfrom the surveyandnot othersignificantfeaturesmentionedby previo


Interpretation;
althoughthese might sometimesbe mentionedin the text.

A SURVEY OF THE ROMAN FORT AND SETTLEMENT AT BIRDOSWALD

101

Comparison with other forts per lineam valli is limited. Petch36cut a section through the eastern
defences at Benwell and found two ditches 6 m wide separatedby a 1.50 m space, with a berm of 3
m. Simpson and Richmond37found a similar arrangementto the western defences of the same fort,
with the exception that the inner ditch was 7.60 m and the outer ditch 3 m. At Halton Chesters it
was found that the double ditches to the north and north-west comprised an outer ditch c. 3 m wide
with an inner ditch c. 7.50 m separated by a c. 6 m space.38There was a c. 6 m berm between the
inner ditch and the fort rampart.It has been shown that the sequence of ditches as fort defences is
extremely complex.39 At both Birdoswald and South Shields it was found that cutting and
recutting had taken place on many occasions, for a ditch section cut through clay has only a limited
life.
Causeways can clearly be seen at the porta principalis dextra where the road crosses the ditches,
although no causeways can be seen outside the portae quintanae. These gates would have become
redundantwhen the Stone Wall was moved up to the line of the north fort curtain-wall. It has been
shown that the gates were blocked, and it is assumed that ditches had been cut across the causeway
shortly after the completion of the building of the stone fort.40
The porta quintana dextra does not show up on the survey, although this was excavated recently
by Wilmott.4' Evidence of considerable later activity can be seen in the rampartto the fort wall by
the great many magnetic anomalies, probably reflecting the greater number of people occupying
the fort in its later periods. Simpson and Richmond42found that the rampartto the southern fort
wall was made up partly with rubbish from an earlier fort. An Hadrianicoven was also seen close
to the south-west comer, whilst a successive series of ovens (34) were found to the east of the
porta decumana,43and the south-west sector of the retentura.
INTERIOROF THE FORT

The street pattern


The four sides of the fort are not set out parallel to each other, although the south and west
curtain-walls are set out close to a right angle. The portae principales are not quite directly
opposite one another,which probably accounts for the alignment of the via principalis and the east
and west roads.
The original street patternwithin the fort followed the normal patternfor forts astridethe Wall.44
Later activity within the fort obscured this street pattern, and this reflects the final periods of
occupation. The via principalis (4) is clearly defined, and the line of the Turf Wall ditch can be
made out to the north along its length. Some intrusion into the northernpart of this road can be
seen adjacent to the porta principalis dextra. This probably reflects the blocking of the north
portal which Potter noted was at a higher level than the original level of the floor to the gate.45

36 Petch 1927.
37 Simpson and Richmond 1941.
38 Berry and
Taylor 1997.
39 Bidwell and Speak 1994, 127-44; Wilmott 1997, 86.
40 Wilmott 1997, 100.
41 Wilmott 1997, 65-9.
42 Simpson and Richmond 1932, 142-3.
43 Simpson and Richmond 1933, 252-4 and fig. 10.
44 Johnson 1983, 70.
45 Potter 1855, 142.

102

J. ALAN BIGGINS and DAVID J.A. TAYLOR

Replacement pivot blocks had been provided to both portals. The interpretativeand other plans
show the portals with the later blocking.
The via quintana was not identified. However, Richmond46observed that, as built, it was 20 ft (6
m) wide and made up of gravel 300 mm thick; duringthe period of Severan occupation the line of
the road was covered by buildings.47The via decumana was not detected; but was observed by
Richmond who found it to be 20 ft (6 m) wide and in a badly rotted and disintegrating condition.
This width is narrow, and could reflect a late phase, as it was usual for the width of the road to be
approximately equal to that of the gate portals and spina c. 7.70 m. A very small area of the via
praetoria can be seen in the survey close to its junction with the via principalis at which point the
road would seem to be obstructed. A well defined street can be seen passing between the principia
and the praetorium about 6 m wide. The survey does show an apparent small size of the
intervallum space. It is not clear whether this was a result of later buildings encroaching into this
space, or if it was built to a narrowwidth.
The praetentura
The buildings within the eastern sector are clearly defined, although it was not possible to survey
the northernportion due to the line of the modem road. The entire barrack-block(5) can be seen c.
49 m long by c. 10 m wide. Across the street to the south (c. 6 m wide) are sited a further two
barrack-blocks (6) each with eight contubernia and officers' quarters sited adjacent to the
intervallum road. Both buildings are c. 49 m long, but whilst the northernone is c. 10 m wide, the
one to its south appearsto be only c. 8 m wide.48The contubernia are c. 3-3.50 m wide, and some
internal division walls can be seen in both buildings, which could represent later alterations. The
two buildings were set very close together, and are similar in that respect to the two blocks to the
south (7), which are separatedby an alley c. 900 mm wide. Horsley49describing his visit to the site,
records that the 'houses' within the fort were visible, and measured their wall thickness as 28
inches (710 mm). He stated that the breadth of the passages between the rows of 'houses' or
barrackswas no more than 32 inches (813 mm). The barrackswere laid out per scamna.
In the south of this sector, adjoining the via principalis, Richmond and Birley found two narrow
parallel buildings50 (7), which underwent a series of later alterations and different uses. The
northern building was identified as a barrack-block. From scaling the plan of the excavated
buildings5" it can be seen that the width of the contubernia identified fall within the range c.
3.80-4.80 m, with the depth of the building over the external walls being c. 8.50 m. As the overall
length of the building is c. 49 m it is likely that the numberof contubernia would have been eight,
as the other identified barracks.52The building to the south, fronting the via principalis was
probably used as stables or stores. The northerly of the two buildings faced a barrack-block (6)
across a street c. 5 m wide. Daniels53 re-interpretsthe excavation as perhaps incorporating two
large chalets, suggesting that barrack-blocks generally underwent significant alterations in the

46 Richmond 1931, 128.


47 Richmond 1931, 128.
48 Barrack-blocksof similar varyingwidths can be seen in the mid-Antonine stone fort at South Shields, Dore and Gillam
1979, 39-41; Bidwell and Speak 1994, 14, 18.
49 Horsley 1733, 152.
50 Richmond and Birley 1930.
51 Richmond and Birley 1930, fig. 1.
52 A barrack-blockwas partially excavated in the north-west of the praetentura, by the Central Excavation Unit of
English Heritage underthe direction of Tony Wilmott, in December 1997. The stone-built barrackwas seen to be Hadrianic
and the eight contubernia varied in width from 3.78-4.00 m (pers. comm. T. Wilmott).
53 Daniels 1980, 173-93.

A SURVEY OF THE ROMAN FORT AND SETTLEMENT AT BIRDOSWALD

103

fourth century, when these chalet-type blocks were commonly constructed. It is suggested that
much of the alterations to this building and other barracks,represents the construction of chalets,
where each building is a separatestructure.Evidence of chalets was found in a barrack-blockin the
north-west of the praetentura excavated in December 1997.54Examples of these have been seen
on many sites in the northern military zone including Housesteads, Wallsend, South Shields,
Ebchester, High Rochester, and Newstead. Restating the excavated evidence (with suggested
reconstructions),55Daniels posited considerable alterations to barracksduring the fourth century.
Supportive evidence comes from Newstead which showed that chalet-type blocks were being used
as early as c. 160.56The number of contubernia in each of the barrack-blocksat eight is unusual
per lineam valli, and is only matched at Halton Chesters where sixteen contubernia are provided in
each double block.
To the south-west, the corner of a building (8) could representpart of the eastern section of the
basilica. This building was excavated by Wilmott57and part of the south aisle was found to overlie
the Wall ditch. The basilica was seen to consist of a nave and two aisles, it was estimated that its
length was 42.78 m,58 equal to the distance between the via praetoria and the intervallum road.
The building was found, in this survey, to be set back from the via praetoria, possibly indicating
that the length of the building could be less than 40 m. The setting back of this building from the
road could reflect its importance.
The latera praetorii
South of the junction between the via principalis and the via praetoria lies the principia (9),
althoughthe centreof the building does not lie on the centralaxis of the via praetoria. The cross-hall
and rear range of five rooms can clearly be distinguished, together with the central courtyard
surroundedby a series of rooms which are almost certainly later additions. Some later subdivision
appearsto have takenplace in the cross-hallto formthreemain units. The overall size of the building is
c. 28 m by c. 32 m, with the cross-hallbeing c. 9 m wide overall,with no aisle visible to the south of the
courtyard.An ambulatorywas formedto the north,east and west sides of the courtyardc. 4 m wide.
The projectionof the rearrange is c. 6 m. The northerlysection of the building would seem to have
undergone later alteration with an extension being built onto the north wall, together with a
reconstructionof the northernambulatory,partially extending into the via principalis. No certain
identificationcould be made of a strong-room,although it could be representedby a strong positive
anomaly in the room at the south-westcornerof the rearrange.
Richmond excavated part of the principia59 and estimated the back wall to be fifteen courses
high adding that, 'the back rooms of the principia must be amongst the most remarkablein Britain;
in one room alone we found the voussoirs of the arched entrance lying as they had fallen'. The
nearest comparative size of a principia, per lineam valli, is that at Carrawburghwhich measures
26.20 m by 28 m, although that at Halton Chesters is c. 30 m by 39 m and Chesters 27.54 m by
39.19 m.
The praetorium (10) was situated to the east of the principia, separatedfrom it by a street c. 6 m
wide and set back from the assumed line of the via principalis, calculated at 7.70 m wide, by some

54
55
56
57

T. Wilmott pers. comm.


Daniels 1980, 173-93.
Clarke 1996, 5.
Wilmott 1997, 79-82.
58 Wilmott 1997, 81.
59 Richmond 1931, 127.

J. ALAN BIGGINS and DAVID J.A. TAYLOR

104

3 m. The building measured c. 33 m by 26 m, and comprised a series of rooms set arounda central
courtyard,which may have had a width of c. 10 m. The building shows evidence of considerablelater
alterations and additions, and it is probablethat the range of rooms to the west was c. 6 m wide.
Richmond60noted material from a hypocaustwhich he thoughthad come from the building. Whilst
excavating 'not far' from the east gate, Potter61found a room with a hypocaust together with a
similar room leading from it. These rooms measured 10 ft (3.05 m) by 8 ft 9 inches (2.70 m) and 10
ft (3.05 m ) by 9 ft 6 inches (2.90 m). A furtherroom was seen adjacent,measuring9 ft 8 inches (2.95
m) by 9 ft 6 inches (2.90 m). The walls to the hypocaust room were described as being covered in
'red coloured stucco', presumablyopus signinum. From their portrayalit is almost certain that the
description refers to a suite of baths attached to the praetorium, and situated adjacent to the
intervallumroad. High negative anomalies in this area could indicate the position of the furnaces.
A substantial building is sited to the west of the principia (12) some 6 m south of the two
third-centurygranaries. The building, c. 27 m by 9 m, is divided internallyinto three by two walls,
and could be the one described by Richmond62as having a north wall at least eight feet (2.40 m)
high. Unfortunately this is the only description given of this building, but it is shown on
Richmond's site plan being c. 93 ft (28.30 m) by 25 ft (7.60 m). The walls to this building are
somewhat thicker than those of the buildings previously described, being in the region of 1 m
thick. Taking this into account with the width it is possible that the structuremight be a granary.
The absence of any buttresses seen by Richmond could reflect the superficial nature of the
excavation, which was almost certainly by narrow trenches, with a trench possibly being sited
between the buttresses. The overall width of both of the third-centurygranaries at Birdoswald is
8.26 m. A pathway can be seen leading from the north-eastcorner of the building to the south-east
corner of the principia, and from thence, passing by a group of buildings to the south, to the
north-west corner of the possible farmstead(13). A significant later structure(11) was constructed
adjacent to the south east corner of the praetorium.
The retentura
A positive, linear feature, with a roundedcorner, runningnorth and west adjacentto the south-east
corner may suggest that this represented a defensive feature of an earlier fort. A similar linear
feature can be seen to the west just inside the intervallumroad.
The dominant building in this sector of the fort is a clearly defined farmsteadand enclosure (13),
which can be seen to overlie earlier buildings, possibly barracks.The enclosure measures c. 22 m
by 18 m, and the farmhouse c. 11 m by 11 m seems to cut into the south-east corner of the
enclosure, and in this respect it is almost identical to a similar complex at Sundayburn in the
Tynedale Liberty, adjoining the river Irthing.63This is a type of farmstead which is common to
north-east Cumbria,and is typified by a house and one or more enclosures, though rarelymore than
one building. It could date from the eighteenth century,although there is no documentaryevidence
to support this assumption. The absence of documentaryevidence could suggest that the building
is much earlier, perhapsrepresentativeof a phase of occupation on the site between the sub-Roman
period and the farmhouse in the north-west corner. The large cleared area to the west of the
complex could reflect the desire to have a cleared area free from upstanding masonry around it.
The linear feature (32) to the south running to the south-east is a probable prehistoric defensive

60
61
62
63

Richmond 1931, 128.


Potter 1855, 67-9.
Richmond 1931, 127.
Ramm, McDowall and Mercer 1970, 49, 52.

A SURVEY OF THE ROMAN FORT AND SETTLEMENT AT BIRDOSWALD

105

feature which can also be seen outside the fort to the east. This can be identified as a bank on the
topographical survey.
It is likely that there were two pairs of barrack-blockseach side of the via decumanaadjacentto the
south intervallumroad (14), each with eight contubernia,as the lengths of the buildings are similar
to those in the praetentura. Furtherbarrack-blockscould have been built to the north of these.
Simpson and Richmond64show the south wall of a possible barrack-blocksome 11.90 m northof the
east guard-chamberto the south gate and 7 m to the east of its eastern wall. This narrow trench
apparently found no evidence of walling to the south. It is probable that the buildings in the
south-east sector (14) represent a pair of barrack-blocksseparatedby a narrow alley, with some
considerable lateralterationto form chalet-typedwellings, the separateunits of which can be clearly
distinguished. The overall size of each block is c. 47 m by 8 m. There is a suggestion that there may
be one or more large buildings, probablybarrack-blocks,in the south-westernsector. The buildings,
possibly barracks,sited on the line of the via quintana, could be those mentioned by Richmond,65
and thought by him to relate to early third-centuryoccupation.
Part of a semi-circular feature (15) can be seen to the north of the south gate, possibly forming
part of a post-Roman enclosure wall. The position of the south-west angle tower is clearly seen as
an intense feature, and could reflect occupation over a prolonged period.
DRAINS AND WATER SUPPLY

A main drain can be identified (16) cuttingdiagonallyacross the south-westernpartof the retentura
before passing under the principia in an approximatelynorth-south direction, and then proceeding
diagonally across the praetenturato a point at the rampartsjust north of the intervaltower. Several
subsidiarybranchesjoin the main drain. This was probablythe lowest point of the original ground
level, and it is possible that the latrineswere sited to the inside of the curtain-wallat this point. This
featureis shown very precisely on the resistivity survey, as is the modem drainrunningnorth-eastin
the south-east corner of the fort. As the response intensity of both drains is similar, it cannot be
discounted that feature (16) forms part of a later drainagesystem. The strong linear feature in the
south-westof the retenturarunningnorth-west/south-east,shown in FIG.5, is a moderndrain;a similar
featurecan be seen in the vicus to the west.
Bruce66 states that there was a spring to the west side of the station, and that he saw water
flowing in a water course about three hundred yards from it. The water course, made up of flat
stones, fed a cistern near the centre of the station, presumablyclose to the principia.
THE VICUS

Bruce67observed that, 'The field on the east side of the station contains the foundationsof several
suburbanbuildings'. More recently Shotter68contendedthat, 'the continuing strengthof the vicus at
Nether Denton (which survivedthe abandonmentof the fort)has been adducedto explain the apparent
absence of a vicus at the neighbouringWall Fortof Birdoswald'.

64 Simpson and Richmond 1933,


fig. 10 opp. 254.
65 Richmond 1931, 128-30.
66 Bruce 1867, 260-1.
67 Bruce 1867, 261.
68 Shotter 1996, 50.

J. ALAN BIGGINS and DAVID J.A. TAYLOR

106

At Birdoswald there is a marked difference between the buildings to the west of the fort and
those to the east. The former are much slighter in their appearanceand are probably more lightly
constructed, with the use of timber-framedconstruction. These buildings are generally much more
linear in their layout, with concentrations on each side of the road leading out of the fort. The
majority of the buildings to the east, however, appearto be of more substantial construction, more
closely aligned, and are almost certainly stone-built. Some of the buildings to the east may
represent a form of military annexe and it is possible that the blurred linear feature (31) could
define its original extent. The military annexe usually contained the bath-house, and other military
buildings which could not fit comfortably within the ramparts.At Housesteads the exceptional
masonry of Buildings V and VII is thought to representmilitary use,69and it is even suggested that
troops might have been billeted there.
West of the fort
The survey was continued to the west of the fort for a distance of 80 m. Buildings fronted the road,
c. 9 m wide, leadingwest out of the fort. These were not only of the traditionalrectangularform with a
narrow gable fronting the street (17),70 but also of squarishproportionon plan (18). Some of the
former, possibly tabernae, can be seen fronting the street to the south, whilst the latter are more
generally spaced. Some slighterstructuresindicatedby a thin darkline (19) could representbuildings
using timberconstruction,and could relateto those seen by Simpsonand Richmondto the south of the
fort.71The excavators suggested that these buildings, possibly used for storage, were of a temporary
naturedating from the constructionof the stone fort. One timberbuilding (20) has a drainrunningto
the south on its west side. Circularpositive features behind the buildings fronting the road could
representrubbishpits.
The cemetery is sited some 350 m to the south-west of the fort, and several isolated buildings
have been identified in the field to the west of the fort72which could be associated with it. Birley
and Keeney73point out that the position of temples and cemeteries has an importantbearing on the
plan of a settlement, and state that at several forts the temples are set some distance from the main
roads. This proximal relationship of cemeteries can be seen at both Housesteads and Saalburg,
where in general the cemetery lies on a road leading away from the settlement.
East of the fort
The survey extended 120 m to the east from the curtain-wall. Simpson and Richmond74found that
the large mounds outside the east gate covered the vicus (or fort annexe), which was seen to have at
least three, well defined occupation levels. The heap aligned north-south, north of the porta
principalis dextra,is predominatelyexcavationspoil, presumablythat producedby GlasfordPotterin
1850. The presenceof these moundsinterferedwith the magneticreadingsof the anomaliesbelow them.
The road leading out of the fort past the outer ditches was fronted by much larger stone buildings
than those to the west. Buildings on each side of the road seem to have impinged on its width and
reduced it to c. 5 m. Similar stone buildings are closely grouped within the linear feature (31).

69
70
71
72
73
74

Birley and Keeney 1935; Crow 1995, 73.


Salway 1958, 236.
Simpson and Richmond 1934, 125-9.
Walker 1986, unpublished resistivity survey, pers. comm.
Birley and Keeney 1935, 237.
Simpson and Richmond 1934, 130.

A SURVEY OF THE ROMAN FORT AND SETTLEMENT AT BIRDOSWALD

107

Another road (21) runs south-east at an angle of c. 60 degrees from the north-east interval-tower,
incorporatinga kink close to some substantialbuildings (22) sited close to the south of the Wall. It
is possible that one of these might have been a bath-house. A furtherroad (23) runs from the latter
road towards the north-east.
A ditch (24) runs from the south of the Wall to cut the vallum (25). This ditch does not cut any of
the roads, and, as it does not follow a straight line, it is implied that it was excavated in sections
thereby respecting the existing roads; as it cuts the vallum it would have a terminuspost quem of c.
125. The ditch clearly cuts through Building 28 together with several buildings to either side of the
east-west road. It is probable that it is a late defensive feature, and is evidence of the abandonment
of these buildings outside the fort. A similar set of circumstances has been seen to have occurred in
the vicus at Malton,75where a large ditch (3.60 m wide) with a rampartwas cut through buildings
close to the south-east gate, and a fifth-centurydate has been adduced.
The truncated ends of two parallel ditches (26) can be identified where landslip has occurred to
the south-east of the survey, and these could relate to the north-south ditch (24). The curving end
of a furtherditch (27) can be seen to the north of the vallum. This could be the end of the ditch seen
by Richmond,76which ran from a point north of the porta quintana dextra stopping short of the
vallum in a curving end, implying that the vallum was there when it was dug. He considered this
ditch cut off an area defined on the south by the vallum, and on the east by the ditch.
A well defined square feature (28) can be seen on the edge of the survey on a continuationof the
line of the via quintana. The building measures c. 9 m on each axis, and a ditch was seen on three
sides. There can be little doubt thatthis is the building seen by Richmond77to the east of the fort and
described as, 'A good building, twenty feet square,with three-foot walls standing thirteen courses
high....'. A drain was also seen to run away from the feature towards the escarpment. If the
dimensions stated by Richmond were the internalones, then the size almost exactly matches that
shown on the survey. It is probablethat this building is a signal-tower connected with the Stanegate
limes,78 local examples of which are known at Pike Hill, Walltown, and Mains Rigg. This latter
tower is 6.40 m (21 ft) square, and surroundedon four sides by a ditch. During the course of the
recent excavations on the site, a photographictower was placed approximatelyon the site of this
building. It was found that it was possible to see over the fort and observe the west and north
horizons.79The building has been extended in its full width to the east, as can be seen by broken
joints in the masonry, for a distance of c. 20 m. In addition some nine cellular type units have been
built onto the tower and its extension, to the south, forming one block of three and one block of six.
The widths of the blocks are c. 7 m and 14 m respectively, and the length of each c. 18 m. A wider
unit was provided in the westernmost block adjacentto the tower. A gap of some 3 m was formed
between each block. The extension to the signal-tower has been cut by the north-south ditch (24).
An ear-shaped sub-circular enclosure (29) can be seen just outside the defensive ditches
measuring c. 27 m by 15 m. This is constructed in stone and overlies earlier buildings. It may relate
to a much later farmstead similar in form to that at Mount Hulie,80but this is only a provisional
interpretation.
A drain (30) can be seen running north-south from the point on the Wall where a drainage outlet
can be seen discharging to the north. This was constructed as an integral part of the Wall implying
a connection to an existing drain for an existing settlement.

75 Wenham and Heywood 1997, 29-30, 38.


76 Richmond 1931, 122-4.
77 Richmond 1931, 130.
78

See note 1.

79 T. Wilmott pers. comm.


80 Ramm et al. 1970, 47, 49 fig. 13, 187.

108

J. ALAN BIGGINS and DAVID J.A. TAYLOR

A short distance to the south of feature (28), a possible linear prehistoric earthwork(32) can be
seen running east-west. It is evident in the southernpartof the retentura and runs to the east of the
curtain-wall, where it subsequently cuts the vallum, to the edge of the survey. Evidence of a ditch
(33) can be seen to the south of the Wall which probably formed part of the defences to the
promontory fort on its north side. The position of this defensive feature may have influenced the
siting of the realigned Stone Wall. To the east of the late ditch (24) there is little evidence of any
stone buildings; there are however many slighter features which probably represent timber
buildings.
Several anomalies can be seen associated with the defensive ditches to the east of the fort (35).
These features can be seen to cut across the ditches in many instances and could possibly be
interpretedas buildings. This could be considered doubtful as it is unlikely that buildings would be
erected in this position during the occupation of the vicus. It is, however, conceivable that they
may have been erected at a much later date. It is possible that the anomalies were caused by the
material making up the fill of the ditches.
CONCLUSION

The survey has quickly and economically revealed the extent of the archaeology of most of the
fort, together with significant areas of a possible military annexe and vicus to the east and west.
The interpretation is complex, for it is probable that some form of continuous occupation took
place within the fort and its settlement from prehistoric times until the present. The geophysical
method of survey cannot distinguish between different phases of construction, and only provides a
composite plan of all the features within the instrument's depth range. It is only possible to
produce an interpretationof limited accuracy, and hence only a tentative reconstruction is offered;
excavation is the only sure method of verification. It is not considered that this paper is an
appropriate place to enter into the wider discussion on the implications of the number of
contubernia, the potential number of barrack-blocks,and the size of the garrison in the fort.
The size of the vicus is unknown, and no significant work has yet been carriedout on vici in the
northernmilitary zone. A geophysical survey has been carriedout on part of the vicus at Chesters
which has been seen to extend for over 150 m beyond the fort to the south. Extensive settlements are
known at Benwell, Housesteads, and Chesterholm. At Newstead the mid-second-century civilian
settlement in the southern annexe is thought to have comprised 80-100 buildings, although this
interpretationis not wholly accepted, whilst the eastern annexe had buildings on either side of the
road for a distance of c. 80 m.81 It is probable that the civilian population has been greatly
underestimatedper lineam valli. The Romanarmyhas been shown to have provideda major input of
wealth into the native population,82and acted as a honey-pot for camp-followers and traders.
The scale of the prehistoric features seen was unexpected, and it is now known that more lie to
all sides of the fort. This is supported by the recent finding of part of a lost resistivity survey
carried out to the north, east, and west of the fort.83This survey shows evidence of ditches and
earthworksto the east as far as Milecastle 49, and to the west for a distance in excess of 250 m. A
great many features were also seen to the north. It is hoped that the survey will be resumed to
determine the full extent of the occupation of the site in both its Roman and prehistoric phases.
This paper should therefore be considered as an interim report.

81 Clarke 1996, 4.
82 Millett 1994, 57-60.
83 See note 69.

A SURVEYOF THEROMANFORTAND SETTLEMENT


AT BIRDOSWALD

109

ACKNOWLEDGEMENTS
The assistance of all the staff on the site at Birdoswald is greatly appreciated together with that of Bruce
Bennison of CumbriaCounty Council Heritage Services. Thanks are also given to Professor MartinMillett of
the Departmentof Archaeology, University of Durhamfor his helpful comments on a draft of this paper, and
to Tony Wilmott, of the Central Archaeology Service, English Heritage for his help and advice during the
survey and the preparationof this paper.

Department ofArchaeology, University of Durham


Thispaper is published with the aid of a grantfrom Cumbria County Council

BIBLIOGRAPHY

Berry, J., and Taylor, D.J.A. 1997: 'The Roman fort at Halton Chesters: a geophysical survey',
Archaeologia Aeliana5 25, 51-60
Bidwell, P.T., and Holbrook, N. 1989: Hadrian's Wall Bridges, English Heritage Archaeological
Report 9, London
Bidwell, P.T., and Speak, S. 1994: Excavations at South Shields Roman Fort. 1, Soc Antiqs
Newcastle upon Tyne Monograph4, Newcastle upon Tyne
Birley, E. 1961: Research On Hadrian's Wall, Kendal
Birley, E., and Keeney, G.S. 1935: 'Fourth report on excavations at Housesteads', Archaeologia
Aeliana4 12, 204-58
RCHME 1987: Birdoswald (Banna) Roman Fort and its Environs, unpub. survey
Breeze, D.J., and Dobson, B. 1991: Hadrian's Wall, Harmondsworth
Bruce, J.C. 1867: The Roman Wall,Newcastle upon Tyne
Clarke, S. 1996: Trimontium:A Roman Frontier Post and its Phases, University of Bradford
Crow, J. 1995: Housesteads, London
Daniels, C.M. 1978: The Roman Wall, Newcastle upon Tyne
Daniels, C.M. 1980: 'Excavation at Wallsend and the fourth century barrackson Hadrian's Wall',
in W.S. Hanson and L.J.F. Keppie (eds), RomanFrontier Studies 1979, pt 1, BAR int. ser. 71 (i),
173-200
Dore, J.N., and Gillam, J.P. 1979: The Roman Fort at South Shields, Soc Antiqs Newcastle upon
Tyne Monograph 1, Newcastle upon Tyne
Gillam, J.P. 1950: 'Recent excavations at Birdoswald', Trans Cumberland& WestmorlandAntiq
& Archaeol Soc2 50, 63-8
Haverfield, F. 1897a: 'Report of the CumberlandExcavation Committee 1895', TransCumberland
& WestmorlandAntiq & Archaeol Soc1 14, 185-97
Haverfield, F. 1897b: 'Report of the CumberlandExcavation Committee 1896', Trans Cumberland
& WestmorlandAntiq & Archaeol Soc1 14, 413-33
Horsley, J. 1733: Britannia Romana (Grahamedn reprinted 1974), Newcastle upon Tyne
Johnson, A. 1983: Roman Forts, London
Millett, M. 1994: The Romanization of Britain, Cambridge
Petch, J.A. 1927: 'Excavations at Benwell (Condercum),first interimreport(1926)', Archaeologia
Aeliana4 4, 135-92
Potter, H.G. 1855: 'Amboglana', Archaeologia Aelianal 4, 63-75, 141-9
Ramm, H.G., McDowall, R.W., and Mercer, E. 1970: Shielings and Bastles, RCHME London

110

J. ALAN BIGGINS and DAVID J.A. TAYLOR

Richmond, I.A. 1929: 'Excavations on Hadrian's Wall in the Gilsland-Birdoswald-Pike Hill


Sector, 1928', Trans Cumberland& WestmorlandAntiq & Archaeol Soc2 29, 303-15
Richmond, I.A. 1931: 'Excavations on Hadrian's Wall in the Birdoswald-Pike Hill Sector, 1930',
Trans Cumberland& WestmorlandAntiq & Archaeol Soc2 31, 122-34
Richmond, I.A., and Birley, E.B. 1930: 'Excavations on Hadrian's Wall in the Birdoswald-Pike
Hill Sector, 1929', Trans Cumberland& WestmorlandAntiq & Archaeol Soc2 30, 169-205
Salway, P. 1958: 'Civilians in the Roman frontierregion', Archaeologia Aeliana4 36, 227-44
Shotter, D. 1996: The Roman Frontier in Britain, Preston
Simpson, F.G., and Richmond, I.A. 1932: 'Report of the CumberlandExcavation Committee for
193 1. Excavations on Hadrian's Wall', Trans Cumberland & WestmorlandAntiq & Archaeol
Soc2 32, 141-5
Simpson, F.G., and Richmond, I.A. 1933: 'Report of the CumberlandExcavation Committee for
1932. Excavations on Hadrian's Wall', Trans Cumberland & WestmorlandAntiq & Archaeol
Soc2 33, 246-62
Simpson, F.G., and Richmond, I.A. 1934: 'Report of the CumberlandExcavation Committee for
1933. Excavations on Hadrian's Wall', Trans Cumberland & WestmorlandAntiq & Archaeol
Soc2 34, 120-30
Simpson, F.G., and Richmond, I.A. 1941: 'The Roman fort on Hadrian's Wall at Benwell',
Archaeologia Aeliana4 19, 1-43
Walker, R. 1986: Resistivity Survey, unpub. Kendal
Wenham, L.P., and Heywood, B. 1997: The 1968 to 1970 Excavations in the Vicus at Malton,
North Yorkshire,Yorkshire Archaeological Report 3, Leeds
Wilmott, T. 1993: 'The Roman cremation cemetery in New Field, Birdoswald', Trans
Cumberland & WestmorlandAntiq & Archaeol Soc2 93, 79-85
Wilmott, T. 1997: Birdoswald, English Heritage Archaeological Report 14, London
Wilmott, T. forthcoming (a): 'A possible Neolithic burial from Birdoswald, Cumbria', Proc
Prehist Soc
Wilmott, T. forthcoming (b): 'Excavations on the Birdoswald Spur 1997', Trans Cumberland &
WestmorlandAntiq & Archaeol Soc
Woolliscroft, D.J. 1989: 'Signalling and the design of Hadrian's Wall', Archaeologia Aeliana5,
17, 5-20

Das könnte Ihnen auch gefallen