people and spaces in roman military bases || site selection and data processing
TRANSCRIPT
Cambridge Books Online
http://ebooks.cambridge.org/
People and Spaces in Roman Military Bases
Penelope M. Allison
Book DOI: http://dx.doi.org/10.1017/CBO9781139600248
Online ISBN: 9781139600248
Hardback ISBN: 9781107039360
Chapter
4 - Site selection and data processing pp. 50-64
Chapter DOI: http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge University Press
4 Site selection and data processing
INTRODUCTION
Fundamental to the objectives of this study is the suitability of artefact dis-
tribution inside Roman military bases for analysing socio-spatial behaviour
within and between sites. There are a number of reasons why Roman mili-
tary sites are particularly appropriate for such spatial analyses. This chapter
presents the logic behind the choice of military sites, per se, as well as the
factors that are important for selecting sites, and for selecting these specific
case studies in Germany.
THE SUITABILITY OF MILITARY SITES FORARTEFACT DISTRIBUTION ANALYSES
One of the main reasons why military sites are suitable for intra- and inter-
site analyses of socio-spatial behaviour is their apparently formulaic layout.
Not only are they, by archaeological definition, physically bounded by a
fortification wall, but they also tend to have a certain amount of structural
conformity in their internal layout. The consistency of this structural con-
formity has sometimes been exaggerated in the presentation of excavated
sites, by reconstructing its assumed existence in poorly preserved or unex-
cavated areas of military sites (e.g. at Oberstimm: see Schonberger 1978:
figs. 65–6; and Rottweil: see Franke 2003: plan 14). Nevertheless, each known
military base has at least some of the recognisable components. As discussed
in Chapter 2, there is a general perception that this systematic layout dic-
tated a systematic use of space within these sites. As well as this bounded
internal space, a military base can include an external settlement – outside
the fortification walls – whose function is widely assumed to have been that
of a civilian settlement, attached to the military base but with contrasting
functions to those inside the fort proper. This relative conformity of layout
and dichotomised approach to the use of space makes these military bases
ideal for this type of analysis. In his concern for the suitability of GIS tech-
niques for analyses of archaeological sites Stanton Green asked (1990: 4)50
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
Suitability for artefact distribution analyses 51
‘Can one partition human behaviour into spatial packages?’ Current per-
ceptions of the organisation of Roman military bases indeed imply that they
constitute a series of in-built ‘spatial packages’, within a complete ‘spatial
package’. Testing for the veracity of such perceptions through GIS-assisted
analyses would seem appropriate. This study concentrates on sites dating
to the first and second centuries, a period when these military bases were
considered to be most systematic (see Gardner 2007a: esp. 107), segregated
and dichotomised. Of vital significance for the identification of meaningful
artefact distribution patterns at such sites are also good understanding of the
depositional processes and general taphonomic conditions. Ideal sites are
those which were short-lived and rapidly abandoned, and which have been
extensively and well excavated with specific attention paid to taphonomy
and artefact context.
Abandonment and depositional processes
Many Roman military bases were long-lived, continuously occupied sites
which often developed into towns and even cities (e.g. Colchester, London
and Cologne). Other military bases, which did not become towns, were also
frequently long-lived as military establishments before their final abandon-
ment. Many of these, which may originally have been built in the first and
second centuries, continued in use into the third and fourth centuries. An
example is the site of Vindolanda where a military presence was recorded
from the late first until the end of the fourth century ce (Birley 2009).
However, because of military aggression and disillusionment, expan-
sionary policies, and changing imperial strategies during the early empire
(see Maxfield 1987b; Sommer 1999a), there are also numerous first- and
second-centuries military bases that were short-lived installations in fron-
tier regions which experienced rapid abandonment and which were never
built over again and which potentially provide useful sites for this study.
Even so, many such sites had several building phases, or phases of occu-
pation, and complex depositional processes, prior to and during the final
abandonment. As intimated in the previous chapter, the main such depo-
sitional processes that impact on how artefacts were left behind on such
Roman military bases comprise loss, reuse, discard, and abandonment of
such items and the decommissioning of a fort by the departing troops so
that nothing of use was left for the enemy.
Many of the key artefacts used for this study (e.g. jewellery, hairpins,
brooches and items for personal hygiene) are the types of items that were
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
52 Site selection and data processing
likely to have been lost during daily activities. Because most buildings and
spaces within military bases, particularly during the early empire, were often
not paved, one might expect a higher probability for lost items to be left in
situ in this context than in more urban sites. While such lost items might have
been systematically cleaned out, at least from dwellings and paved buildings,
and possibly dumped in designated refuse areas as well as in street and open
areas and related ditches, comparisons with other archaeological contexts
suggest that they could often have been located within the vicinity of the
structure. For example, at the aboriginal mission station at Wybalenna, on
Flinders Island (Bass Strait, Australia), Judy Birmingham discussed evidence
for sweeping detritus to immediately outside the cottages (Birmingham
1992: esp. 57 and 183–4); and, under the verandahs that surrounded the
nineteenth-century Old Kinchega Homestead, in outback New South Wales,
Australia, were found many artefacts associated both with these verandahs
and their adjoining structures (see Allison 2003: esp. 182–4).
Important among the depositional processes that affect an artefact’s loca-
tion of end use, is the general disposal of rubbish. Excavations of many mil-
itary bases have indicated that the rubbish of routine occupation was often
dumped in the fortification ditches and ramparts. Some such rubbish might
also be dumped in disused areas of the fort or used to provide levels for
new buildings, as has been argued for the fort at Ellingen (see Chapter 10,
p. 235), and so could be relocated within the fort. Rubbish could also be
disposed of in pits and wells within the fort, often in close proximity to
location of use (e.g. pits between barrack buildings at Dangstetten [Finger-
lin, pers. comm., July 2003; see Fingerlin 1970: fig. 28; 1986: 13]; see also
Shaft 6 at Ellingen [Chapter 10, p. 241]). Gardner has argued that small, and
seemingly valuable, items such as coins might be found in such accumulated
refuse (2007a: 71–9). This apparent unintentional dumping of these types
of artefacts away from their place of use, and especially in the fortifications
and disused parts of the fort, needs to be taken into account in this study.
It is also important to note that artefacts, particularly of metal, could be
kept as scrap for reuse. Bishop (1986) emphasised the deliberate discarding
of material, most notably military equipment, which was not therefore lost
or abandoned, but rather was found in its place of end use as scrap. Such
scrapped items can sometimes be associated with specific buildings, and so
can provide either general or more specific contextual information on the
activities (e.g. metalworking) carried out within the associated structure
or area. Again many of the types of small, often complete, items used for
this study are unlikely to have been recycled in this manner. However, it
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
Suitability for artefact distribution analyses 53
is conceivable that some of the larger fragments of metal (such as combat
equipment) could have been discarded material ready for reuse.
When a military base is rapidly abandoned, or destroyed, one might
expect much more material to have been left in its location of use. How-
ever, there is also likely to have been more inadvertent dropping of items in
streets and open areas than during more normal occupation. At sites with
less rapid abandonment, the systematic evacuation and decommissioning
of a military base can result in the relocation of artefacts away from place
of use. For example, at Inchtuthil in Scotland, the complete removal and
subsequent deposition of building material in a 1.8 metre-deep pit, perhaps
for later retrieval, was reported (see Pitts and St Joseph 1985: esp. 279–80;
for other examples: Hanson 1978: 302–5; see also Bishop 1986: esp. 721).
Conversely, sites where metal remains and particularly building materials
(e.g. nails) are reported are likely to have been more rapidly abandoned, less
methodically dismantled and therefore less likely to have experienced the
type of relocation of artefacts that results from more systematic decommis-
sioning. Even at sites which were systematically demolished, though, one
might expect rubbish to be left in situ and for buildings to be pushed over
in place, rather than for such rubbish to be moved to other parts of the site.
For example, at sites where military units come and go, such decommis-
sioning can include the relaying of earth floors, and the dumping of rubbish
between these floors by departing soldiers, potentially sealing a deposit (e.g.
at Vindolanda – see van Driel-Murray 1995: 8 and 1997: 57; Birley 2009:
esp. 45–6 and 53). Such occupation surfaces, both inside and outside build-
ings at such sites, have the potential to include many lost, discarded and
abandoned items left at or near their location of use.
As well as the usual depositional processes which can impact on the spatial
distribution of artefacts inside Roman military bases, some artefacts could
potentially have been ‘ritually’ deposited and this ‘symbolic value need not
reflect [their] practical use’ (Clarke and Jones 1994: esp. 119), although
this is probably the least likely scenario. Nevertheless, it can be found in
settlement sites, notably in wells and pits (see Allason-Jones 1996a), as well
as in burial contexts. An important factor in the analyses of the sites included
in this study was the recording of the contents of wells and pits.
There are other taphonomic conditions that can impact on the distri-
bution of artefacts, especially at abandoned sites that are not continuously
occupied. The most significant of these is erosion, which can remove occu-
pation layers, but which can also move artefacts around the site, as is evident
at Hesselbach, discussed below.
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
54 Site selection and data processing
While all these depositional processes and taphonomic conditions poten-
tially impact all Roman military sites, sites were chosen for this study where
their impact was likely to be minimal and where these factors could be
accounted for. However, it is important to note that it is extremely difficult
to identify any diurnal or seasonal activities through this study of artefact
assemblages.
Excavation procedures
Careful excavation using modern procedures can effectively identify these
depositional processes and taphonomic conditions, and so trace the end use
of artefacts as related to their location of use, and phases of occupation, even
at military sites that became largely civilian settlements, were methodically
decommissioned or suffered from erosion and other environmental impact.
However, most modern excavation is generally development-led, piecemeal
excavation of former Roman military bases within modern towns and cities
(e.g. Colchester: see Crummy 1992) and restricted by limited funding and
more stringent approaches to site preservation and excavation strategy. Such
excavations almost never involve the open-plan excavation of an entire
military base, as is desirable for more comprehensive spatial analyses of
artefact distribution. This means that few well-excavated and well-recorded
sites, excavated over the last half century, are suitable for the types of analyses
in this study. Potential exceptions are Elginhaugh (Hanson 2007) and also
Vindolanda, with more extensive excavation than most and with good
conditions for artefact preservation (e.g. of leather and wood) that have
yielded a great range of artefactual material from across both the inside
and the outside of the military base. However, by no means all of the
excavations at Vindolanda have paid rigorous attention to the recording of
precise artefact context (see Birley 2009), resulting in doubt being cast on
van Driel-Murray’s findings.
Many of the military sites that were relatively rapidly abandoned and
which did not later become towns have long been easily identifiable in
the landscape, and were therefore excavated in the nineteenth and ear-
lier part of the twentieth centuries by scholars, often with more classical
than archaeological training, and without careful attention to all deposi-
tional conditions and to precise artefact contexts, although often exposing
extensive areas of the site (e.g. South Shields on Hadrian’s Wall: see Dore
and Gillam 1979). This applies particularly to military bases with substan-
tial stone structures. It is less true for more ephemeral, frequently early
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
Suitability for artefact distribution analyses 55
imperial, wood and earth-built camps which have often only more recently
been identified, largely through aerial photography, and so have not gen-
erally been excavated and whose identifiable structural features are usually
limited to the fortifications (see Jones 2011, 2012).
Despite more modern standards of excavation, however, some of the
easily identifiable sites that were more extensively excavated in the early to
mid-twentieth century are indeed the most suitable for this study. This is
largely because they were often short-lived and rapidly abandoned sites. So,
irrespective of the less than ideal quality of the excavation procedures at
such sites, they often provide comprehensive coverage of the site, with at
least some attention to the depositional processes and with artefact data sets
which run into the thousands,1 thus allowing spatial analyses of artefact
assemblages across different components of these sites.
That said, though, there is, sadly, a lack of early imperial sites where
both the extramural settlement and the military base itself have been sub-
stantially excavated with adequate attention to artefact deposition. The
early, but extensive, excavations of military bases have generally paid little
attention to the extramural settlements of these sites. There are excep-
tions, however, such as Saalburg and Zugmantel in Germany, but these are
unsuitable for the current study as discussed below. Since the 1970s there
has been more concern for the excavation and study of such settlements
(e.g. Vindolanda: Birley 1977: 29–30; for further examples in Germany:
Sommer 1997). At Vindolanda, though, no first- and second-century extra-
mural settlement to compare with the contemporary fort has yet been
excavated, only in the third- and fourth-century occupations which are
outside the concerns of this study.2 This lack of attention to settlements
outside the fort walls during periods of more extensive excavation mean
that few sites which include both an extramural settlement and an intra-
mural site have been comprehensively excavated. Also few early imperial
sites excavated with modern methods have comprehensive coverage that can
be used for comparative analyses (compare Housesteads: Rushworth et al.
2009).
1 Tens of thousands would be preferable but such data sets are not currently available in a useable
form.2 Andrew Birley (Director of Vindolanda Excavations) has confirmed that: ‘[t]here is very little
evidence from the second-century extramural settlements at Vindolanda as most of the
excavations in these levels have focused on the forts themselves thus far’ (A. Birley, email,
4 April 2012). His investigatons of some of the artefact patterning between the third- and
fourth-century forts and extramural settlements at Vindolanda have, however, produced
significant patterns (Birley n.d).
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
56 Site selection and data processing
For many of these reasons – notably re-occupation, early excavation, and
lack of attention to taphonomy and systematic recording of finds – the
Italian peninsula, Cisalpine Gaul and the Iberian Peninsula are not partic-
ularly suitable for this study. Also in the eastern empire and North Africa
there are few suitable early imperial sites with appropriate depositional and
excavation histories (see Kennedy 1996: Pollard 2000: esp. 11–15). In the
north-west frontier, however, early imperial sites have tended to be better
preserved, have appropriate depositional and occupational conditions and
have been more comprehensively excavated and published. Most notable are
those in Germany (see Tac. Hist. 4.21–5). Good potential examples are the
forts and fortresses that resulted from Augustus’ expansion of military force
into the Lippe Valley and Tiberius’ subsequently withdrawal, but there are
also other military bases in the Rhine and Danube regions which are useful
for this study. As discussed below, though, none are ideal. Most notably
the techniques of excavation in Germany (i.e. in levels rather than strati-
graphically) mean that, while these sites have been extensively excavated, the
precise depositional processes are not always noted and carefully recorded.
AVAILABILITY OF USEABLE DOCUMENTATION
Because the best sites for this study are those that were extensively excavated,
those from older excavations often have the potential to be more useful
than sites that have been more recently excavated, despite the less rigorous
excavation procedures and recording methods. That said, the objectives of
these past excavations, as of some more recent investigations, have been
principally to isolate the structural remains to develop an understanding of
the chronology of these sites and of the types of units stationed at them. Most
have only published material pertinent to these aims. This means that, in
terms of artefact recording, there has been a propensity to concentrate only
on the site-wide recording of diagnostic pottery and military equipment,
or the documentation of exceptional finds (e.g. silvered horse harness from
Vetera I – Jenkins 1985). For few such excavations have the types of precise
contextual information on artefacts and artefact distribution, desired for
this study, been comprehensively recorded. Crucial requirements in the
selection of extensively excavated, short-lived sites for this synthetic study,
then, are the availability of appropriate documentation of the excavation
processes, depositional processes, and artefact contexts, as well as relevant
chronological information.
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
Availability of useable documentation 57
My analyses of artefact distribution in Pompeian households required
time-consuming archival research of unpublished, hand-written, excava-
tion reports (see Allison 2004a, 2004b, 2006a). For this study’s synthetic
approach more accessible excavation reports with useable plans of the
site and its excavations, and comprehensive and contextual artefact cat-
alogues, with good artefactual illustrations, were needed. For this reason,
the remains of the first- and second-century forts at Vindolanda are not
suitable as much post-excavation artefact analysis is still being carried out
to characterise this material and to re-contextualise the assemblages from
this site. For this reason, sites in Germany were chosen for this study because
there is a tradition in Germany of comprehensive publication of archaeo-
logical excavations. Each regional archaeological monuments’ authority, or
‘Bodendenkmalpflege’, has its own series in which such excavations have been
published (e.g. Archaologie im Rhineland; Forschungen und Berichte zur Vor-
und in Fruhgeschichte in Baden-Wurttemberg). Particularly important for
this study is the Limesforchungen series of the Romisch-Germanische Kom-
mission des Deutschen Archaologischen Instituts. These publications are
generally in a traditional format (see Allison 1997). They tend to involve a
comprehensive interpretation of the structural remains and, as mentioned,
use the ceramics and other significant artefacts to date the various phases
of the site. Other artefacts are recorded in a catalogue but rarely used in the
interpretation of the site. This type of relatively extensive, and often lavish,
publication has received criticism (e.g. Gechter 1998). Given its usefulness
for this type of study, though, such criticism seems unwarranted.
Indeed, the extensive excavation and publication of these sites in mod-
ern Germany, and limited interpretative presentation of the artefacts, make
them the most suitable case studies for this study.3 More recent piecemeal
excavations in Germany can also be useful, because of the comprehensive
publications of artefact catalogues giving detailed contextual information
on artefacts and artefact assemblages. However, as noted above, these exca-
vations and excavation reports are not without their problems. Until quite
recently there has been a lack of stratigraphical excavation of Roman sites
in Germany, with many excavations having been carried out in a series of
spits or levels (Planum or Flache) that can cut through specific depositional
3 Putting aside the later dates for many Roman military sites in Britain, a lack of comparable
funding for their excavations and publications and a need to focus limited resources on
developing research agenda around development-funded archaeology (James and Millett 2001:
esp. ix) are among some of the reasons why Romano-British sites have not usually been so
extensively excavated and published (although see Hanson 2007).
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
58 Site selection and data processing
features and occupation layers (see e.g. Zanier 1992: esp. 69 nn. 249–53; see
also discussion in Chapter 10, p. 236).
As outlined above, ideal sites for this study would have been military bases
that comprise both a fort, or fortress, and an accompanying and contem-
porary extramural settlement, a vicus or a canabae. Sebastian Sommer has
carried out a detailed study of such sites in Germany where military bases
with extramural settlement have been excavated (Sommer 1988). However,
none of these sites have appropriate and accessible artefact reports. While
Saalburg and Zugmantel might seem appropriate for this study, with exten-
sive excavation of both fort and extramural settlement, the provenance
information for these forts is not currently available in a useable form.4
Besides the reasons given above for the selection of the particular sites in
this study, the data in the published reports needed to be in a systematic
format that could be easily collated and analysed spatially, using GIS-assisted
techniques. As has been discussed elsewhere (Allison 2008b; see also Witcher
2008), and was apparent for my previous Pompeian studies (e.g. Allison
2004a, 2006a), these printed ‘legacy data’ from more extensive open-plan
excavation than is often feasible today proved more useful for the aims of this
study than more recent, piecemeal, digitised or unpublished excavations.
SELECTION OF INDIVIDUAL SITES
Five sites were selected for this study on the basis of all these criteria – the use
of the site, the depositional processes, the extensiveness and rigorousness
of the excavation and recording procedures and the quality and format of
its publication. These sites are, in chronological order: the double legionary
fortress of Vetera I, near the town of Xanten in the Lower Rhine region; the
legionary fortress and cohort fort, Forts I and II, at Rottweil on the Neckar
River (a major tributary of the Upper Rhine); the supply fort at Oberstimm
in the Upper Danube Region; the numerus or auxiliary fort at Hesselbach,
4 When I selected sites for this study, during the initial stages of this project in 2003, the records
of artefacts excavated at Saalburg, held at the Saalburg Museum, consisted of hand-written
cards which lacked easily identifiable contextual information. Miriam Etti (Saalburg Museum)
has since commenced collation of the old excavation reports of the fort and extramural
settlement at Zugmantel (Etti n.d.). Investigation of the recording processes used for other
extensively excavated sites in Germany (e.g. Dangstetten, Vindonissa, Oberaden and Anreppen)
revealed that, at least in 2003, appropriate cross-referenced digital and geo-referenced data were
not easily accessible and neither did such sites have printed artefact and provenance data that
could be easily prepared for digital analyses by procedures used in this study.
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
Selection of individual sites 59
on the limes in the Oldenwald, 40 km north-east of Heidelburg; and the fort
at Ellingen, also in the Upper Danube Region (Figure 1.1).
The four main sites – Vetera I, Rottweil, Oberstimm and Ellingen – were
short-lived, seemingly fairly rapidly abandoned, lacking systematic demo-
lition or erosion and had relatively well-preserved and well-documented
deposits. The fort at Hesselbach has a similar quality of publication but was
not rapidly abandoned, would appear to have been systematically decom-
missioned, has evidence of a later non-military settlement and suffered from
erosion. This fort therefore serves as a control for the quality of the data and
for the character of the artefact assemblages at the other four sites.
Four of these five forts have been extensively excavated by open-plan
excavation, while Forts I and II at Rottweil were excavated by a series of
more recent, development-led, piecemeal excavations. Four of the sites were
mainly excavated since the mid-1960s. The other, Vetera I, was excavated
in the early twentieth century but has recently been reinvestigated and its
excavated artefacts and their provenances comprehensively published.
For each selected site, the published artefact catalogues are illustrated
and have provenance information that locates the find spots of the artefacts
on the published site plans. Many excavation reports, where the artefact
catalogues are organised typologically, do not include easily identifiable
provenance information for each artefact. That said, not all the data from
all the selected sites are of equal quality. A brief description of these sites,
and their excavation and recording procedures is given here, with further
discussion on the histories of these sites and on the excavation procedures
and resulting documentation in the relevant chapters.
Vetera I
Vetera I is predominantly a first-century double legionary fortress, covering
an area of c. 58 hectares (Figure 6.2). It was relatively short-lived, lasting
from c. 13 bce until c. 70 ce. The first legions at Vetera I were probably the
Legio V Alaudae and the Legio XXI Rapax (Lehner 1930: 7, after Tac. Ann.
1.45), the latter being replaced by the Legio XV Primigenia in 43 ce (see
Hanel 1995: 5–7).
The central area of this fort was extensively excavated in the first decades
of the twentieth century by Hans Lehner, in a series of 1 m wide trenches
(e.g. Lehner 1907, 1912, 1930). These trenches were ostensibly to trace the
walls of the last Claudian–Neronian fortress but Lehner also identified the
remains of the fortifications of at least three earlier fortresses and recorded,
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
60 Site selection and data processing
particularly in his earlier excavations, numerous pits and their association
with different phases of the fortress. The find spots of the artefacts from
these excavations were relatively well recorded, often including particular
depositional characteristics (e.g. ancient surface finds, or finds from pits),
although these were not initially published. Catalogues of these artefacts and
the site plans were comprehensively published more recently by Norbert
Hanel (1995). Hanel’s catalogues include illustrations of these artefacts and
provenance information that can be co-ordinated with that in the site plans.
Because the excavations focused on the central area of this large fortress,
little is known of the barrack buildings, however, and evidence for any
external settlement is minimal (see Hanel and Song 2007). Despite its early
excavation, these excellent, provenanced artefact catalogues published by
Hanel (1995), which comprise over 12,000 artefacts, including many non-
combatant remains and some potentially associated with women, make
Vetera I suitable for such artefact distribution analyses.
Rottweil
On the Neckar River at Rottweil there were five Roman forts and fortresses
and a civilian settlement, identified as the Arae Flaviae (see e.g. Planck
1975) (Figure 7.2). Most were on the east bank of the Neckar and have been
excavated throughout the twentieth century. The two used in this study,
Forts I and II, are on the west bank. This fort and fortress were the last of a
series of short-lived military bases constructed in the Rottweil area during
the first and early second centuries. The earlier of the two, Fort I, covered
c. 17 hectares, is identified as a legionary fortress dating from the beginning
of Vespasian’s reign, and probably housed the Legio XI Claudia (Franke 2003:
143, 158). Fort II was constructed inside Fort I, probably for a cohor milliaria
or a legionary vexillatio when Legio XI Claudia was recalled to Vindonissa
in c. 85 ce (Franke 2003: 158). With the exception of the excavation of
a bath complex, first excavated in the nineteenth century and then more
extensively in 1967 (Klee 1988), the excavations of these forts, Forts I and
II, comprised urban rescue excavations, carried out between the early 1980s
and the mid-1990s and compiled and published by Regina Franke (2003),
showing close attention to depositional processes.
Oberstimm
The fort at Oberstimm covered an area of c. 1.75 hectares, inside the for-
tification ditches (Figures 8.1–8.4). The main central building was first
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
Selection of individual sites 61
excavated by D. H. Witz between 1909 and 1926 (Witz 1911, 1923–9). Much
of the rest of the fort was subsequently excavated by Hans Schonberger
between 1968 and 1971, using open-plan excavation, and published in
1978 (Schonberger 1978). These excavations, combined with those of Witz,
exposed some 75 per cent of the fort which had been occupied in two differ-
ent phases between c. 40 ce and c. 120 ce. Some precise artefact find spots
and depositional processes are recorded, while many other artefacts can
usually be ascribed to a particular area of the fort and to particular building
phases. Schonberger has suggested (1978: 150) this fort had functioned as a
supply station for troops further to the east.
Ellingen
The fort at Ellingen covered c. 0.9 hectares (Figure 10.1). Limited exca-
vations were carried out here in the late nineteenth century, but it was
comprehensively excavated by open-plan excavation in the 1980s and pub-
lished by Werner Zanier (1992). The fort had two main building phases,
between c. 120 ce and the end of the second or beginning of the third cen-
tury. The discovery of much structural material, especially nails, suggests
that this fort’s abandonment had been rapid and unmethodical. Again some
precise artefact provenances and depositional characteristics are recorded
and other artefacts can be ascribed more generally to particular areas and
building phases. Not all pits and wells are clearly documented but their
depositional processes can often be reconstructed. It is unclear what units
were stationed here. Zanier (1992: 164–70) deduced that in the earlier phase
of occupation the fort housed either a service vexillatio, as a detachment
of either the Ala Auriana or Ninth Batavian Cohort at Weissenburg. In the
second phase he thought it probably housed a Pedites singulares who were a
half-cohort-strong part of an auxiliary infantry unit.
Hesselbach
The fort at Hesselbach was extensively excavated between 1964 and 1966,
using open-plan excavation (Baatz 1968, 1973). It covered an area of
c. 0.56 hectares and dated between c. 95 ce–165 ce (Baatz 1973: 66–7)
(Figure 9.2). Dietwulf Baatz identified it as a numerus, or auxiliary fort,
to secure the border (1973: 11), possibly a Numerus Brittonum (1973: 70).
In contrast to the other sites in this study, this early second-century fort
appears not to have been rapidly abandoned. The almost complete absence
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
62 Site selection and data processing
of any metal remains is comparable to the evidence at Inchtuthil in Scotland
and suggests that this base was decommissioned (Baatz, pers. comm., 13
July 2005). This contrasts with the extensive metal assemblage excavated at
Ellingen (Zanier 1992: 195–7). In addition, erosion was reported at the fort
at Hesselbach, from the north-eastern part of the fort to the south-west, but
not at any of the other forts in this study. Thus, this fort’s abandonment pro-
cedures and recorded preservation do not meet the criteria required for this
study but are useful as a control to compare the comprehensiveness of the
assemblages and the spatial representativeness of the artefact distribution
at the other four sites in the study.
DATA PRESENTATION AND ANALYTICALPROCEDURES
The specific data used in this study are, therefore, published reports of
excavations carried out from the early to late twentieth century at military
sites in Germany, in particular the artefact catalogues, artefact illustrations,
and site plans in these publications. The criteria for site selection were: that
the forts had evidence for a short occupancy and relatively rapid abandon-
ment without substantial reoccupation; that they were relatively extensively
excavated with at least some attention to the depositional processes; that
the excavation reports have site maps and plans, clearly showing excavated
areas; and that they have relatively comprehensive artefact catalogues with
provenance information. Each entry in each artefact catalogue needed to
have at least some information about the find spot, or provenance of that
particular artefact.
Sites with digitised and geo-referencing data would have been desirable,
but this was not essential (see Allison 2008a). Specific procedures were
developed so that GIS-assisted techniques could be used to spatially analyse
the distribution patterns of the artefacts and their activity and identity asso-
ciations, represented by these non-geo-referenced ‘legacy data’. Preparing
these data for these analyses involved capturing and digitally formatting the
printed artefact catalogues; capturing the numeric and graphic information
from the printed plans and digitally redrawing them; importing these dig-
ital plan data into a GIS environment; capturing the co-ordinate system of
the relevant features produced by the GIS environment; and merging these
data with the provenance information from the digitised artefact catalogues
so that the artefact distribution patterns could be plotted onto a GIS map.
While relatively straightforward, these procedures were time-consuming
(for details of procedures: Allison et al. 2008).
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
Data presentation and analytical procedures 63
None of the four main selected sites is perfect for these analyses but their
significance is their relative suitability, the accessibility of their documenta-
tion, and also their inclusion of appropriate artefacts that can be categorised
through the processes discussed in Chapter 5. As outlined above, a number
of taphonomic factors at each site needed to be taken into account. Also,
as indicated above, the provenance information in the artefact catalogues
in the printed excavation reports for these sites is of varying precision.
While these variations in depositional processes and recording methods
were accounted for, as best as possible, in the analyses, the impact of each
specific item on the main identifiable patterns is relatively minimal (see
Eckardt 2005: 142). This study is not essentially concerned with the precise
location in which each artefact was found, or the verifiable relationship of its
find spot to the artefact’s end use, or indeed the true geo-referencing of this
material. An advantage of spatial analyses of artefact distribution patterns
across these less-than-perfect sites is that the specific provenances of each
deposited, or redeposited, artefact become less significant. More significant
are consistent, or habitual, patterns within and between these sites. Thus,
GIS-type environments are useful for data of this nature and quality and
the aims of this study.
Basic statistical analyses were also used to test the robustness of the
data (see Appendices B–F). These involved analyses of the percentages and
densities of all artefacts within each space (e.g. building or street area)
at each site, and of the more prolific types of artefacts not included in
the main analyses (i.e. coins and ceramics). A relatively uniform overall
artefact density suggested that the data are fairly robust in that specific
depositional processes (e.g. systematic discard or post-depositional erosion)
or excavation strategies in particular areas of the fort have not had a major
impact on the overall spatial distribution of artefacts across the site. Within
such a uniform artefact density, any variation in density of the selected
types of artefacts suggested that their spatial distribution was likely to be
related to the location of use and to the socio-spatial practices within the
military base. This is by no means a precise statistical analysis, and densities
were noticeably low, but these analyses gave a general impression of the
suitability of spatial analyses of these data for interpreting social practices.
Correspondence analysis was used for Vetera I and demonstrated that the
artefact distribution patterns identified through GIS-assisted analyses were
also apparent in multivariate analysis.
These statistical analyses required specific decisions to define boundaries
between specific spaces, whereas GIS-assisted analyses do not demand such
strict spatial separation. GIS techniques are, therefore, not being used here
to enhance precision. Rather, they are used to aggregate these data into a
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013
64 Site selection and data processing
range of assemblages, and to visually represent them and their different
groupings – rapidly and from a number of different perspectives – for
largely intuitive analyses of their distribution patterns. In this, as outlined
in Chapter 3, these techniques are used somewhat atypically – for small-
scale intra-site analyses and as a relational tool to conceptualise a number of
different scenarios and to analyse the fuzziness of socio-spatial relationships
as represented by these artefact distribution patterns. Guidelines for viewing
these data and the distribution maps compiled for the following analyses
are provided in Appendix A.
As also outlined in Chapter 3, this study constitutes a synthesis of data
from previous excavations. It is therefore dependent on well-published
reports. As such it demonstrates how full publication of excavation material,
and the application of digital methods, can lead to more comprehensive
analyses and to more available data for inter-site comparisons. One of the
major issues for this current study is that, to date, these types of analyses
have not been done at other military sites that might be usefully used
as comparisons to support the conclusions. It is hoped that this study
can provide a useful starting point for this type of research towards more
informed understandings of socio-spatial practices inside Roman military
forts.
Downloaded from Cambridge Books Online by IP 216.165.126.139 on Tue Nov 26 07:49:12 WET 2013.http://dx.doi.org/10.1017/CBO9781139600248.005
Cambridge Books Online © Cambridge University Press, 2013