Download file in PDF format: TRAC 1992: Gender in Question (pp. 3–21)
GENDER IN QUESTION
Carol van Driel-Murray
At the first TRAC meeting, several of the contributors felt it necessary to
stress their ‘gender awareness’, untrammelled by modern preconceptions –
though the effect was somewhat marred by the ensuing unhesitating use
of beads and bangles as indicators of female presence. Yet a glance at their
own audience should have dispelled any illusions as to beads, bangles, long
hair, or ear-rings being exclusive to females – in fact the converse was true.
So, if in contemporary terms there is such a gap between the perceived and
the apparent, how are we to fare with an ancient society where we know
even less of the social conventions?
For the identification of women and children in archaeological contexts
we must clearly turn to the external variables, as Binford’s middle range
theory exhorts us to do. Biological differences naturally lead to the
cemeteries, but quite apart from the problems caused by cremation and
small numbers, the trouble with burials is that the people contained were
dead and don’t tell us all that much about people in life: who was doing
what, where, when, with whom, and how often. And if we are trying to
establish the context of female existence, this is the information we need.
One biological variable that may be indirectly traceable in the detritus of
a living population is body size. Since the female human tends in general to
be smaller than the male, and this sexual dimorphism seems to be more
marked in the ancient than the modern population (Wells 1982: 140), size 4
c. VAN DRlEL-MURRAY
differences reflected in discarded clothing might be one way of approaching the problem. Conveniently, virtually the only kind of Roman clothing
to survive intact – footwear – is also a relatively sensitive exponent of
sexual dimorphism. TIlls allows us to exploit a number of very large
footwear complexes from a variety of Roman sites in north-western Europe
as a source of demographic information.
The foot is the first part of the human anatomy to reach adult dimensions. After growing steadily throughout childhood, boys’ and girls’ sizes
begin to diverge from about the age of 10, with girls achieving their
full adult size between 11 and 13. Boys continue to grow till they reach
their larger, adult male sizes at about 15-16. The longer growth results in
a difference of about 2cm between the average foot size of adult females
and adult males. Naturally this is no absolute rule, but exceptions are only
a problem on the individual level, not in the context of the aggregate
groups which will be examined here. Slight differences between modem
statistics and those from the 19th century suggest that though the course of
growth is unchanged, children now are more advanced, so in antiquity the
divergence between girls and boys may have begun a little later and boys
may not all have reached their adult size till about 16-17. The size difference between males and females remains constant (Martin and Saller 1958:
97lff; Groenman-van Waateringe 1978; Grew and de Neergaard 1988:
The normal size distribution of adult shoes illustrates both the differences between females and males (Fig. 1.1, 1-2) and the area of overlap
between the larger females and smaller males (Fig. 1.1, 3).2 In archaeological footwear contexts, however, the groups are not neatly separated and
what we find is a continuum from the smallest babies’ shoes (from llcm)
to the largest male feet. Thus account has to be taken not only of the
overlap of females and males, but also of the growth of children into
adults, and in particular, the growth of boys into young men, since at this
time they occupy positions in the normal female range. In reality, this is
proportionally not a major source of error since boys’ shoes in the range
35-40 only account for about 5% of the total (van den Burg 1948: 392),
though on an individual level this remains the most problematical category.
If the modem distribution is translated into archaeological terms, a normal
popUlation would appear as Figure 1.1, 4, with the female range of sizes
exaggerated by a) growing youths and b) overlap with the tail end of the
That this is indeed the pattern found in most large medieval urban com GENDER IN QUESTION
2 0 ) 0 .0
Figure 1.1. Modem shoe sizes (after Groenman-van Waateringe 1978: fig. 2).
1. women, 2. men, 3. combined showing overlap, 4. as found.
c. VAN DRlEL-MURRAY
Figure 1.2. Siu distributions from Roman sites (relative scale). GENDER IN QUESTION
plexes (Groenman-van Waateringe 1978; Grew and de N eergaard 1988:
104) as well as on Roman sites (Fig. 1.2, Saalburg, Zwammerdam) suggests
that foot size as reflected by footwear is a valid criterion for distinguishing
men, women and children as components of a living population. Though
children’s shoes can to some extent be correlated to age, there is no differentiation beyond these three categories. At this point it may justifiably be
questioned whether such an obvious finding could even be relevant. However, in the context of Roman military sites, the proportions between the
categories do become interesting because it is generally accepted that there
was a legal ban on the marriage of soldiers (Campbell 1978; Saller and
Shaw 1984: 143-44). Even by contemporaries this ban was regarded as an
extra hardship to be endured and the awkward legal consequences were
mitigated by successive emperors until the restriction was eventually lifted
by Septimius Severns at the end of the second century. When an authority
as eminent as Calvin Wells invokes the unmarried veterans, still dependent
on the ‘pleasant’ (though apparently restricted) amenities of Cirencester’s
prostitutes to explain the imbalance in sex-ratio in a cemetery founded
over one century after this ban had been lifted (Wells 1982: 135), one
realises just how deeply this typically 19th century notion of segregated
military communities pervades thought on Roman military life and, indeed,
the entire subject of interaction between soldiers and civilians.
Looking at the distribution of shoe sizes from the first century military
sites (Fig. 1.2: Valkenburg, Vindonissa) we fmd, not a normal population,
but a single curve, corresponding to the modem graph (Fig. 1.1, 2) for
Thus the expectations of unmarried soldiers, living in a
closed camp are fulfilled. Though the analysis is not yet complete, the footwear from the early first century fort at Velsen shows a similar pattern. In
all three cases, the footwear represents the population of the camp interior
(van Driel-Murray 1985: 49-53) and the possibility of camp villages outside
is therefore not excluded. These forts date to the unstable, campaigning
phase of Roman expansion, when family life would indeed be difficult to
In the second century, however, the picture changes (Fig. 1.2: Saalburg,
Zwammerdam). The graphs for the mixed vicus and fort refuse dumps at
Zwammerdam and the Saalburg reveal relatively balanced populations with
large numbers of children and only a slight preponderance of males, in no
way resembling the male dominated communities envisaged by Wells, Saller
and Shaw or others. As might have been expected, permanent forts, surrounded by permanent vici were evidently more conducive to family life. 8
c. VAN DRIF.lrMuRRAY
VINDOLANDA: A CAsE STUDY — –
The site of Vindolanda on Hadrian’s Wall (Birley 1977 and in press) provides considerable refinement due to the large amount of footwear distributed over four well-dated phases (periods I-IV) between c. AD 90-120
and the availability of two control groups from the later ditches. In addition, written documents give a unique insight into the individuals and the
units occupying the earlier phases, thus providing an interpretative framework for the analysis of finds from the site. The finds from the earlier
phases come from demolition levels which seal rubbish left by the previous
occupants. Here, as at Valkenburg (van Driel-Murray 1985: 49-53) old
buildings were flattened and refuse was levelled off to raise the level of the
wet and subsiding site for the next building phase. Concentrations of particular kinds of refuse indicate that find distribution does actually reflect the
activities being carried out in the buildings in the final period of their
occupation and is not refuse brought in for the purpose of levelling. Period
V is excluded as it is an ill-defined secondary deposit.
The distribution of shoe sizes from the two ditches (Fig. 1.4, top two
graphs) is pretty well identical to that from the Saalburg and Zwammerdam, which are comparable in context and date, with the expected overrepresentation of female sizes characteristic of a normal balanced population. Incidentally, in medieval contexts, the number of children’s shoes
fonus an index of generally rising living standards: if this criterion is applicable to Roman times, then the vici of Vindolanda seem to have been relatively prosperous.
The quantity of footwear from period I is insufficient for conclusions to
be warranted, but by period II numbers have increased. The footwear
comes from deposits covering the praetorium, the quarters of the commanding officer and his staff, in this case either Vettius Severus or Flavius
Genialis, prefects of the Cohors VIIII Batavorum (Birley 1990: 18 and in
press). The distribution is much as Valkenburg, mainly male sizes at 34 and
above. Making allowances for the tailing off of male sizes, this probably
leaves one woman (or youth) and a child for the prefect’s household, a not
unexpected picture. One of the prefects possessed a natty pair of openwork shoes in size 39 (Fig. 1.5).
Period III ended in AD 103. Preparations for departure appear to have
been quite hurried and a lot of material was abandoned more or less in
situ. Concentrations of characteristic debris suggest that tents were being
checked and recycled in the praetorium courtyard and horse gear, including GENDER IN QUFSTION 9
chamfrons, was being refurbished in one of the side rooms. The praetorium
may, therefore have functioned as a collection and sorting point in the last
weeks of occupation in addition to being the residence of the commander.
The footwear shows a different pattern to the other earlier military sites,
due to the presence of a number of distinct individuals, each represented
by several shoes sometimes in pairs and often of different styles (Fig. 1.3,
period III; Fig. 1.7). Despite the disapprobation of the rigidly theoretical,
the abundant documentation makes it tempting to link these distinctive
feet to the inhabitants of the praetorium. F1avius Cerialis, the commander,
with exceptionally elaborate openwork shoes (which for full effect must
have been worn with coloured socks); his wife, Sulpicia Lepidina with a
narrow, extremely elegant foot, size 33 (Fig. 1.3, individual 6; Fig. 1.7, no.
862). She had sensible closed shoes as well as fashionable sandals stamped
with the maker’s name: Lucius Aebutius Thales, surely the flrst designer
label in history (Fig. 1.6). Her foot shape is virtually unique, of all the
hundreds of soles from the site, only one resembles it. This is a smaller
version (Fig. 1.3, individual 5b with shoes and sandals; Fig. 1.7, no. 313 and
1541) and from the same level: a daughter or a servant? There is a whole
group of children, four of the shoes possibly from the same individual
growing from a baby to a child of 5/6 (Fig. 1.3, individuals 1-4; Fig. 1.7 nos
398,431), then an older boy (Fig. 1.3, individual 5a; Fig. 1.7 no. 1520) with
expensive openwork shoes who oddly and characteristically scuffed the outside of his left foot. Are these the ‘pueros tuos’ of the letter Inv. 412?
(Birley in press; van Driel-Murray, in press). This is an exceptional case of
individuals identifiable through a combination of circumstances and the
conclusions are unlikely to be generally applicable. Apart from these individuals and a few isolated soles in the range 27-34 which may belong to the
household, the pattern is the expected male, military model.
On withdrawal of the Batavian cohort, the structures of period III were
demolished, covered with turf and a barrack block was erected on the site
in AD 104 (dendro date; Birley in press). The garrison now seems to have
been the Cohors I Tungrorum, with evidence for legionary detachments
appearing in some of the texts. The actual room deposits of course reflect
the flnal occupation and probably date to c. 115/120 after which the whole
site was cleared for the fabrica of period V. Gone are the individuals of
period III: a few pairs can be identified, but there are only three possible
‘individuals’. The m<U0rity of soles are large, with 34 still marking the
bottom end of the male range. But in addition, there are a fair number of
women and children here, actually inside the barrack block (Fig. 1.3, 10
N :. 11 2
C. VAN DRlELrMURRAY
r- rn I
)0 34 40
Figure 1.3. Vindolanda shoe sizes by period. Arrow points to male/female overlap,
numerals in Period III indicate discrete individuals. (Period II c. AD 90-95; period
III c. 95-102/ 3; period IV 104 – c. 120.) GENDER IN QUESTION
Period V I
An tonine Diten
Figure 1.4. Vindolanda shoe sizes by period. Arrow points to male/female overlap.
(Period V c. 120 – mid-2nd cent.; period VI late 2nd cent.; Inner Ditch 3rd cent.)
c. VAN DRlEL-MURRAY
period IV; Fig. 1.9).
While we may have been prepared for some form of family life in the
settlements outside the forts, cohabitation inside the fort at this date is unexpected, to say the least.
The unconscious model for Roman camp organisation has always been
drawn from British military practice with garrisons regarded as military
preserves, occupied with noble and dedicated young men, strictly segregated from the surrounding settlements. Brothels under military supervision
cater for physical needs and that is the limit of ‘native interaction’. This
view lies behind the bleak and loveless picture of military life accepted by
Wells (1982), Saller and Shaw (1984: 142) and Gilliam (1978; quoted in
Roxan 1991: 465). But the Dutch military authorities tended to be more
pragmatic and they also tolerated a far greater degree of mixing between
Europeans and natives. To me the situation in the Vindolanda barrack is
recognisable as the archaeological exponent of the system of camp concubinage, as practised by the Dutch forces stationed in the Indonesian
colonies in tlle recent past. Without going into too many details here, it
may be profitable to look at the workings of this system as an alternative
form of military organisation in which unofficial unions were officially
tolerated (de Braconier 1913; van der Wurff-Bodt 1989).
Until well into the 20th century, common soldiers in Dutch military
service were not permitted to marry at all, while officers had to meet exorbitant property qualifications before an application for marriage would
even be considered. The reasons were unashamedly financial. Pay for the
lower ranks was poor and it was readily admitted that if soldiers were to be
allowed to marry, wages would have to be substantially increased. The military authorities were reluctant to bear the costs of providing transport and
accommodation for families, and above all, the cost of the widow’s
pension. to which officially married couples were entitled. Essentially, these
are the reasons for the Roman ban. but then clothed in finer sounding
terms (Campbell 1978: 154).
Despite the official ban, and with full connivance of the authorities,
soldiers in Indonesia did have wives: concubines taken from the
native population associated with the camp (in Roman terms, the view).
This population (the vicani) came to constitute almost a separate cast,
having little contact with the surrounding native population, iliough in
some cases still linked to it through ties of kin. With their disregard for
constraining native traditions, their military orientation, and their language
laced with pidgin terms, these villagers were totally dependent on the army GENDER IN QUESTION 13
Figure 1.5. Openwork shoe, Vindolanda Period II.
presence and could, like their Roman equivalents, hardly be termed
‘civilian’ in the conventional sense. A1:, late as 1905, when marriage conditions for officers had been considerably eased, half of the officers and c.
16% of the common soldiers, most of those serving more than the standard
6-year term, still had a native concubine or ‘housekeeper’. The native
Indonesian forces, which made up 2/3 of army strength, were almost all
married under their own, Islamic, law, but as this was not recognised, these
women were also regarded as concubines by the military authorities. Not
that it made any difference to their rights under local law as regards dowry,
inheritance or property. The women had to be registered by the camp
authorities as official concubines, with a ration entitlement and with
specific rights and duties. Until the early years of the 20th century only the
officers had separate accommodation: the rest, married and single alike,
slept in the communal dormitories. Children slept in either a hammock
slung above the bed or on the floor under it, the meagre possessions
stored in a chest or on a shelf running the length of the barrack block. To
modern sensibilities, the lack of privacy is astonishing, but conditions in
native kampongs and, indeed, the wretched homes from which most
European recruits were drawn were hardly better. With moral pressure
growing in the Netherlands towards the end of the 19th century, the
military authorities became remarkably active in defense of the system of
camp concubinage. Novels appeared describing the life of concubines and
the issue was discussed in leading liberal magazines (Si-Tamoe-Larna, 1913;
de Braconier, 1913). The military authorities had little time for moral
arguments about the holy state of matrimony: they recognised concubinage 14 c. VAN DRIEL-MURRAY
Figure 1. 6. Lepidina ‘s Period Ill.
as one of the few mitigating features in the comfortless, ill paid life of the
soldiers. It was seen as a stabilising factor, reducing crime and especially
drunkenness, encouraging men to sign up for a further 6 year term of
duty. Maintaining a concubine encouraged thrift – indeed, wages and
rations were paid direct to her.5 Criticism was countered by placing
curtains between the beds in dormitories and allowing more space for cohabiting men.
The women left the barracks in the morning, spending the day in a
specially built women’s hall washing, cooking and doing odd jobs to earn
money. Concubinage may have been advantageous for men and authorities, but for the women it was simply a case of survival. If a soldier died or
was killed on duty, she had no rights and she and her children were turned
out immediately, their only hope being to find someone else to take them
on. As most men eventually returned to Europe, abandonment was almost
inevitable and women attempted to make some provision for the eventuality. The lack of provision for women in soldiers’ wills, noted by Roxan
(1991: 465) is perhaps suggestive of similar relationships, though the
strength of local customary law should not be underestimated. Despite the
uncertainties potential concubines seem to have preferred European
soldiers (they earned more). A form of upward mobility: one wonders
whether the legionary detachments at Vindolanda had the same advantage
over the Tungrian auxiliaries.
The real problem was formed by the children. Boys were expected to GENDER IN QUESTION
o /\ __ ) 0
O iQ o \,
Figure 1.7. Selection of period III individuals (scale c. 2:7): individual 2,
431; individual 4, 398; individual 5a, 1520; individual 5b, 313, 1541;
individual 6, 862.
join the native forces, as their Roman counterparts did, but for girls there
was little hope other than concubinage or prostitution: ‘it is a melancholy
fact that many, if not most, daughters of the European common soldiers
are, in a sense, compelled to prostitute themselves, as they are destitute of
any provision’ (Ducimus 1902: 196, original emphasis). If the same applied 16 c. VAN DRlEL-MuRRAY
to the Roman vicus inhabitants, we may have one more reason why so few
daughters appear on diplomas (Roxan 1991: 465-66). Despite the glory,
fine uniforms and regular, if low, pay, the soldiers and the camp population were despised by both the native Indonesians and the Europeans of
the administrative levels: mixed race children were unacceptable to either
group. Thus between the native population and the governing elite was a
third element, the military population with its own culture, its own symbolism of status, centred on itself and essentially self-perpetuating. With the
lack of evidence for Roman interaction on native settlements in northern
Britain, this perhaps encapsulates the social isolation of the forts along
Hadrian’s Wall. Enclave theory seems more promising than frontier theory
in this situation.
The size distribution of the footwear from Vindolanda period IV fulfllis
all expectations of camp concubinage. There are further implications: the
women spend most of the day outside the fort (in one of the vicus buildings of uncertain function?6) so most female activities and their attendant
losses of personal ornaments and tools will also be found outside. Indonesian women returned to their families or to a female relative with a home
in the camp village to give birth to their children, so we would not even
expect to find dead new-borns under barrack floors as proof of female
presence. These women would have been largely invisible in the material
culture of the fort itself.
Even by ancient standards, it might be awkward to accommodate men,
wives and children in the familiar bunks of the 8 man contubernium, so if
concubinage were a regular practice there were evidently fewer men to a
barrack block than usually envisaged at this date. It is perhaps unlikely that
all men would have a concubine and it may be that only one or two rooms
were set aside for such men. The distribution of footwear belonging to
women and children in the Vindolanda barrack is suggestive of this (Fig.
The other small finds from the barrack block are scrappy and rather
sparse, the sort of oddments people would discard when sorting out their
possessions.7 Demolition and levelling in preparation for the next construction phase has tended to spread material, so find distribution need not
exactly represent actual room activity, but that it does bear some resemblance is suggested by the spread of soles fitting together, forming a pair or
clearly belonging to the same individual. The quite marked concentration
of women’s and children’s shoes in rooms III and IV (Fig. 1.8) suggests
these may be the ‘married quarters’. Although the officers’ quarters at the GENDER IN QUESTION 17
end have been only partially excavated, there is a curious absence of female
Sewing needles, broken knives, bronze spatulae and bone or glass gaming
counters are appropriate soldierly attributes, along with a single lonca
plate, a loop junction and quite a number of bolt heads. Considering the
obvious literacy of the soldiers, the number of styli (8) is hardly unexpected. The only remotely female items – in a conventional sense – are a
pin, a ring, a penannular brooch, two long hair pins and a fair number of
wooden combs. None of this is especially conclusive: I presume that men
also occasionally combed their hair. Other than their shoes, the women
and children have left remarkably little trace of their presence. What is
interesting is the amount of broken furniture – chair legs and boxes –
which hints at some standard of comfort and perhaps also at rather more
space in rooms than if 8 men lived in each. Also notable amongst the finds
are the number of broken wooden barrels: for food storage and interior
toilets perhaps? There appears to have been a toilet pit in room XV, so
sanitary arrangements for ladies seem to have been catered for. It seems
unlikely that the conventional 80 man unit was squeezed into this particular barrack block.
The Indonesian concubines considered themselves to be properly
married and they were regarded as respectable members of camp society.
It is clear that there was a widespread aspiration to the formation of stable,
nuclear relationships (contubernium) between men and women who were,
for a variety of reasons excluded from full Roman citizen marriage in its
strictly legalistic interpretation (Rawson 1974; Treggiari 1981). Contubernium applied to a range of such conditions, including what was probably full marriage under native law and custom and would be equally
appropriate to the status of Indonesian camp concubines, even down to
the multiple contubernia (Treggiari 1981: 61). That the word also applied to
the soldier’s messmates perhaps helped to obscure de facto marriages in
camp records. It is surely no coincidence that a document from the period
IV barrack itself should refer to the contibernalis (sic) of Tagamatos the
standard bearer (Birley 1990: 29-30, T944). If this person was female, she
clearly controlled finances in much the same way as the fictitious Saridjem
The evidence from period IV for women and children may not seem very
strong, but it does not stand alone. Exactly contemporary is the legionary
fabrica on the Bonner Berg (van Driel-Murray and Gechter 1983). The
footwear distribution is remarkably similar to that of Vindolanda period IV 18 C. VAN DRlEL-MURRAY
•  c
+ °O<±>+ XVII 0
• 0 o II!
0 0 CO 0
• 0 0 IV
• ® (f)(±JO ®O 0+0 0 V xv •
• 0 VI 0
• VI! c
XlV Xll … —….
+ children < 19 em
® w/e 19/20em
o women 20.5 . 22.5 em
FiguTe 1.8. Vindolarula Period IV barrack (restored), location of
smaller soles: children <19cm (to size 38); juvenile/small adult
19-20 cm (28-30); small adult 20-22.5 em (30-34). GENDER IN QUFSTION 19
(compare Fig. 1.2. with Fig. 1.3). Writing in 1980, and faced with what
seemed to be an impossible situation, I made ridiculous attempts to explain
away the small sizes as very young recruits (op. cit.: 23). They must have
been amazingly young to still be in the juvenile stages of growth evidenced
by 20 individuals below size 34, which, as at Vindolanda, marks the tail end
of the adult male distribution. Thus are we blinded by our preconceptions:
it is now perfectly obvious that these are the concubines and children of
legionaries, helping their menfolk out with tasks outside the camp, just as
their Indonesian counterparts were to do 1700 years later.
GENDER IN QUESTION —-
Though with its relatively low-key involvement and its tolerance of native
systems, the Dutch colonial experience may be a more useful source of enlightenment than the British, a more serious criticism which may be
levelled against the present construct is that it is patently founded on
personal preconceptions. Here, I have assumed that men and women
aspire to permanent unions. But this is the question posed in the title. I
consciously prefer to see women and children living in concubinage in the
barrack, but with a different life-style I might be tempted in another direction.
It has already been mentioned that the female size distribution is distorted by the presence of boys growing through these sizes to reach their
ultimate adult size. It is, therefore, possible that all sizes below 34 (the
standard tail end of adult males throughout all Vindolanda periods) belong
to boys and juveniles (Figs 1.3 and 1.4). Because it is not a consistent curve,
but bi-modal, like that at Bonn (Fig. 1.2), some of the older, but still
growing boys are absent. In other words, we would be left with the
beardless youths so beloved of Latin poets: the predilections of Julius
Caesar and, indeed, Hadrian himself need no further elaboration. Under
this model, the concentration in rooms III and IV look, I am sorry to say,
more like a male brothel than anything else. I am culturally adverse to the
idea of children abused in the barracks at Vindolanda, but given a different
society, and especially a slave society, it might not have been seen that way.
As yet it is impossible to separate male and female footwear except for a
few highly distinctive and fashionable styles. Lepidina’s ‘daughter’ had a
very serviceable pair of boots masking the extreme elegance of the sole and
the foot impression within. So we come to a full circle again. The
impossibility of indisputably defining female presence leaves two 20 C. VANDRIEL-MURRAY
alternatives to choose from. Both have repercussions on the population
density in the camps and raise questions as to the actual strength of the
garrisons. Both improve the lot of the hapless soldier in bleak frontier
1. The contribution presented at TRAC 2 and printed here is a preliminary version of a
paper due to be more fully expanded elsewhere.
2. British children’s sizes 1-13, adults 1-12; continental sizes are foot length in cm x 1.5.
3. Figures from production of footwear for issue on ration in the Netherlands in 1947, hence
not affected by the vagaries of consumer choice.
4. I use ‘European’ since only about 50% actually came from the Netherlands. The rest was
composed of Germans, Belgians and Danes – about as ethnically mixed as the Roman forces.
5. Prostitution and the attendant VD infections were seen as a far greater threat to the morals
of the force: comparisons unfavourable to the British practices were drawn. There, the military
brothels, medically controlled or not, resulted in an annual VD infection of 45%. as against 35%
for the Dutch forces, concentrated mainly in the single term group, and only 18% amongst the
native forces, who were almost all married (Ducimus 1902: 189).
6. Building Xl for example, which lies just outside the military annex of the ‘Vicus’ I might be
a later candidate.
7. For the find distribution I am indebted to R. Birley, who made the lists available to me
prior to publication (Birley in press).
Birley, R 1977. Vi1UUilanda, a Roman Frontier Post on Hadrian’s WalL
Birley, R 1990. The Roman Documents from Vi1UUilanda. Greenhead.
Birley, R n.d. Vi1UUilanda Guide. Greenhead.
Birley, R in press. The Early Timber Forts at Vindolanda.
de Braconier, A. 1913. Het kazerne-concubinaat in Nederlandsch Indie,
Vragen van den Dag 28:974-995.
van den Burg, G. 1948. Handhoek voor den schoenhandel 1948. Nijmegen.
Campbell, B. 1978. The Marriage of Soldiers under the Empire. Journal of
Roman Studies 68:153-166.
van Driel-Murray, Carol 1985. The Production and Supply of Military
Leather Work in the First and Second Centuries A.D.: a Review of the
Archaeological Evidence. In M. C. Bishop (ed.), The Production and
Distribution of Roman Military Equipment, 43-81. Oxford, British Archaeological Reports (Int. Ser. 275).
van Driel-Murray, Carol in press. The Footwear. In Birley.
“.an Driel-Murray, Carol and M. Gechter 1983. Funde aus der fabrica der
legio I Minervia im Bonner Berg. Rheinische Ausgrabungen 23, Beitriige GENDER IN QUESTION
zur Archiiologie des romische Rheinlands 4:1-83.
‘Ducimus’ 1901. Het kazerne-concubinaat. Indisch Militair Tijdschrift
‘Ducimus’ 1902. Het prostitutie vraagstuk in het Indische leger. Indisch
Militair Tijdschrift 33: 188-212.
Grew, F. and M. de Neergaard, 1988. Shoes and Patterns. Medieval Finds
from Excavations in London 2, London.
Groenman-van Waateringe, W. 1978. Shoe Sizes and Paleo demography?
Helinium 18: 184-189.
Martin R. and K. Saller 1958. Lehrbuch der Anthropologie. 3rd ed. Stuttgart.
Rawson, B. 1974. Roman Concubinage and Other De Facto Marriages.
Transactions of the American Philological Association 104:279-305.
Roxan, M. M. 1991. Women on the Frontiers. In V. A. Maxfield and M. J.
Dobson (eds), Roman Frontier Studies 1989, 462-467. (Proceedings of the
XVth International Congress of Roman Frontier Studies, Exeter 1991).
Saller, R. P. and B. D. Shaw, 1984. Tombstones and Roman Family Relationships in the Principate: Civilians, Soldiers and Slaves. Journal of Roman
Si-Tamoe-Lama, 1913. Saridjem – de kazerru! huishoudster. Utrecht.
Treggiari, S. 1981. Contubernales in CIL 6. Phoenix 35:42-69.
Wells, C. 1982. The Human Burials. In A. McWhirr, L. Viner and C. Wells
(eds), Romano-British Cemeteries at Cirencester, 135-202. (Cirencester
Excavations II, Cirencester 1982).
van der Wurf-Bodt, C. 1989. Het kazerneconcubinaat in Indie. Spiegel