VINDOLANDA AND ITS TEXTILES:
GAVVO AND HIS TOSSEAE
John Peter Wild*
T
filling of the western ditch outside the first timber
fort waterlogged, but so were two successive timber
phases of the commandant’s house, and later a barrack block erected on the same spot (Birley 2002,
col.pl. 4). To avoid wet feet, the residents repeatedly
laid down carpets of bracken, straw and moss – and
dropped their rubbish. Everything survives.
Archaeological excavation was a challenge, but
the finds are without parallel. The most striking are
over 1000 pieces of leaf tablets, thin sheets of alder
wood used like papyrus for writing in ink. Reading
them is helped by infra-red photography. They include both official records and private correspondence, and give us a vivid glimpse of the social and
economic activities in and around the fort (Bowman
2003; Bowman 2006; Birley 2002). One tablet, for
example – it was originally folded in half – seems to
be an inventory of special textiles for banqueting
used by Flavius Cerialis, the fort prefect of AD 101105 (Bowman and Thomas 1994, 166-170, no. 196).
It is headed cubitoria, ‘for banqueting’, and the last
line reads tunicas ce[natorias], ‘dinner tunics’.
Along with the tablets and old leather shoes were
at least 700 fragments of textile, rags discarded on the
floor after long use and repeated recycling. They are
all of wool, as one would expect in bog conditions;
no processed vegetable fibre survives, although linen
bandaging soaked in honey is mentioned on a tablet
listing medical materials (Bowman and Thomas
2003, 44-46, no. 591.b.10). How far this accident of
survival skews our view of what soldiers wore is a
debatable point. In Egypt, where animal and veg-
theme of this colloquium is textiles and dyes
in the ancient city. Vindolanda, however, is not a
city, but a Roman auxiliary fort situated just behind
Hadrian’s Wall, the northern frontier of Roman
Britain (Birley 1977; Bidwell 1985) (Fig. 1). Nevertheless, Roman forts shared several characteristics
with Graeco-Roman cities (Goldsworthy and Haynes
1999). They had dense populations, a clearly defined
social hierarchy and many of the hallmarks of Moses
Finley’s ‘consumer city’ (Finley 1985, 123-149, 191196; Mattingly 1997). Soldiers grew some of their
own food, often made their own pottery and were
supplied with certain goods and services by their
camp-followers in the civil settlements which grew
up around the forts (Swan 2008; Dannell and Wild
1987, 66-70; Wild 1993, 85-86). But for their clothing and textiles they were reliant on supplies from the
hinterland of the province – and sometimes much further afield (Wild 2002, 31-32).
The first fort at Vindolanda was built about AD
87 by the First Cohort of Tungrians on a plateau; but
when it was doubled in size to c. 2.8 ha in AD 90/92
to house the next garrison, it was extended westwards
over poorly drained terrain. The consequent waterlogging of floors and buildings proved to be a major
bonus for archaeologists; for it preserved a range of
organic materials which do not survive on most archaeological sites (Birley 1994, 12-14).
The organic deposits were first revealed in 1973
during drainage work to facilitate excavation of the
stone buildings of the civil settlement outside the later stone fort. It became clear that not only was the
HE
* University of Manchester.
PURPUREAE VESTES. III Symposium Internacional sobre Textiles y Tintes del Mediterráneo en el mundo antiguo (C. Alfaro, J.-P. Brun,
Ph. Borgard, R. Pierobon Benoit, eds.), pp. 69-73.
69
70
Fig. 1. Location map of Vindolanda.
etable fibres are both preserved, wool seems to be the
preferred fibre in early military contexts.
The initial discovery in 1973 led to three seasons
of work on the deep deposits. I published in 1977 a
catalogue and brief discussion of the 60 textiles
which had been found (Wild 1977; Wild 1979). Ten
John Peter Wild
years later in 1985 digging resumed for a further five
years, and the textile corpus multiplied tenfold. A
more precise chronology for the first four textilebearing periods of occupation was established, thanks
to dates on the tablets and the dendrochronology of
building timbers, and it became theoretically possible
to assign the deposition of a textile to a specific tenyear time-span, or less (for latest position on dating:
Birley 2002, passim).
I started the cataloguing at once, but on a parttime basis. However, in 1992 the Leverhulme Trust
gave my colleague Bill Cooke a three-year grant to
enable two research assistants, Lucy Campbell and
Colin Cork, to evaluate the textile corpus from a new
angle – treating it as if it were modern material, and
applying the latest digital technology, particularly image analysis, to answering specific questions, such as:
for how long were the garments worn? How many
different spinners supplied the yarn? Positive answers
were obtained, and published (Cooke, Wild, Cork,
Fang-Lu 1995; Cork, Cooke, Wild 1996; Cork, Wild,
Cooke, Fang-Lu 1997); but this research was a distraction from the everyday data recording, and I am
still doing that. The basic measurements have all
been made, but I am still recording structural features
such as selvedges and sewing.
The general character of the textile corpus is
clear, though no definitive statistics are available yet
(Wild 1992; Wild 1993). Nearly two-thirds of the fabrics are in 2/2 diamond twill, with a wide spectrum of
weights – and presumably function (Fig. 2). Lucy
Campbell has drafted the repeat patterns of nearly
300 pieces – though others are concealed under a
thick nap. This diagram shows the relative proportions of different diamond centres present. Eventual-
Fig. 2. Close-up photograph of 2/2
diamond twill TT667 (photo by L.
Fang-Lu).
Vindolanda and its Textiles: Gavvo and his «Tosseae»
71
Fig. 3. Textile vocabulary at Vindolanda.
ly, the three main variables – number of warp and
weft in a pattern unit, shape of the diamond centres
and overall weight of the fabric – will be reviewed
comparatively and the results interpreted.
Plain 2/2 twill is scarce (at 5%); its weavers use
only Z-spun yarns, in contrast to the standard Z/S
spin of diamond twills. 2/1 twill is even rarer.
The remaining 30% of the corpus is divided evenly between plain tabby fabrics of many types and finer basket weaves and half-basket weaves. Traces of
decoration are few: a handful of tapestry-woven
bands on half-basket weaves, spin-patterning on
some diamond twills – and hints from dyestuff analysis that red, purple and blue were among the colours
favoured by the garrison (Taylor 1983). There were
scarlet, green and purple curtains, too, in one of the
principal buildings (Bowman and Thomas 2003, 5358, no. 596.19-22).
Where does the research go from here? At no other site in the Roman Empire, even in Egypt, can one
juxtapose two such rich veins of textile information,
documentary and archaeological, from a closely-dated context. The first exercise must be to discover how
far they can be related to one another in detail.
I have extracted a list of textiles from the tablets,
and divided it into body garments, accessories, outer
garments and textile furnishings (Fig. 3). Some entries are familiar – sagum and tunica for example –
while others are virtually unknown, like tossea. Gavvo, who features in our title, supplied three tosseae
around AD 100 to the fort prefect (Bowman and
Thomas 1994, 159-161, no. 192.6). He was probably
a Briton, a trader, and tosseae are only attested in
Britain and northern Gaul, in one case as a valuable
leaving-present (Wild 2002, 25; Bowman and
Thomas 1994, 160). They seem to have been rugs of
some sort, and the best counterpart in the textile collection is a heavy rug with blue pile from the first
fort. I have interpreted it as a sleeping-mat, but its
Ghiordes-knotted pile is so far unique in the northwest provinces (Wild 2007) (Fig. 4).
A child’s sock in diamond twill from the Period
III prefect’s house looks home-made, so one should
not perhaps be surprised to find no reference to socks
in the tablets (Wild 1993, pl.XII). A rather strange
cap is also not mentioned, probably for the same reason; it is really basketry, skilfully plaited from the
stripped stems of hair moss, polytrichum commune,
which grows everywhere round Vindolanda (Wild
1994). The site team believe that it was worn by a lady as a fly-repellent – but it is more likely to have
been worn by a soldier against the Vindolanda rain.
Listed in a Period III account are 15 saga
corticia, literally ‘bark cloaks’. They are said to be
about 5 feet (1.5m) long, and at 15 denarii each are
comparatively expensive (Bowman and Thomas
2003, 53-58, no. 596. 11-12). Can they really be
sheets of bark? Or, thinking of the hair-moss cap,
could they be woven of tree-bast? Another suggestion
is that they were of bark-tanned leather. There are objections to all of those hypotheses. The research on
the lead tags from Siscia by Ivan Radman-Livaja (in
this volume) promises a new perspective.
Another Period III account records that Lucanus
John Peter Wild
72
Fig. 4. Close-up photograph of knots
on mat TT639.
paid for his ventralis, ‘waist band’ to be repaired and
sewn together (Bowman and Thomas 2003, 68-69,
no. 609.b.2). Contemporary sculptural evidence – a
torso from Casacco in North Italy – indicates that the
ventralis was worn below the military belt to help
support the weight of the weapons suspended from it
(Ubl 1989, Abb.6, 7). Re-enactors have found that the
ventralis also cushions the back against chafing from
armour (Graham Sumner, pers.comm.). A heavyweight Vindolanda diamond twill would serve the
purpose nicely.
Chasing minutiae must not distract us from addressing some of the wider questions posed by the
Vindolanda material, such as: where did the textiles
at Vindolanda come from?
Gavvo and his fellow traders supplied some of the
clothing, we know. One account lists 3 saga, 7 sagaciae, 6 tunics and 7 palliola ‘from Gavvo’ (Bowman
and Thomas 1994, 178-180, no. 207). He will have
relied for the bulk of his goods on weavers in Britain,
but exotica for the prefect’s entourage probably came
from across the Channel. How far the military authorities controlled as well as stimulated the trade is an
open question.
Given the frequent changes of garrison at Vindolanda, the extent of personal import of clothing by
incoming troops should not be underestimated. The
Tungrian and Batavian cohorts who manned the fort
in pre-Hadrianic times came from the Lower
Rhineland where one might expect to find a textile
repertoire similar to that of Britain. If the dominant
diamond twills at Vindolanda reflect Iron Age traditions both in Britain and the Rhineland, the challenge is to identify nuances within an apparently
uniform corpus. Lucy Campbell’s work on diamond
pattern units is a start, as is Colin Cork’s demonstration that the fingerprint of an individual spinner can
be recognised by image analysis (Wild, Cooke,
Cork, Fang-Lu 1998). As to other weaves: Hero
Granger-Taylor has shown that in the eastern
provinces 2/1 twill was often adopted for cloaks and
capes, half-basket weaves for tunics and light cloaks
(Granger-Taylor 2007; Granger-Taylor 2008, 15).
Neither weave is likely to have been in the standard
repertoire of the northern British weaver so soon after the conquest.
There was an old-fashioned alternative not available to the ordinary soldier. The prefect’s wife maintained the honoured tradition of lanificium in her own
household, it seems, spinning and weaving wool. On
one occasion Gavvo sold her 38lbs of it, presumably
ready sorted and washed (Bowman and Thomas
1994, 159-161, no. 192.4-5). Her slaves could have
converted that into 10 tunics or more.
In sum: the finds from Vindolanda have given us
some rich food for thought on broad technological,
social and economic issues. The problem is to avoid
indigestion.
Bibliography
BIDWELL (1985): P.T. Bidwell, The Roman Fort of Vindolanda at Chesterholm, Northumberland, HBMCE Archaeological Report 1, London.
BIRLEY (2002): A.R. Birley, Garrison Life at Vindolanda:
A Band of Brothers, Stroud.
BIRLEY (1977): R. Birley, Vindolanda: A Roman Frontier
Post on Hadrian’s Wall, London.
BIRLEY (1994): R. Birley, The Early Wooden Forts, Vindolanda Research Reports NS I, Bardon Mill.
BOWMAN (2003): A.K. Bowman, Life and Letters on the
Roman Frontier: Vindolanda and its People, revised
edition, London.
BOWMAN (2006): A.K. Bowman, «Outposts of empire: Vin-
Vindolanda and its Textiles: Gavvo and his «Tosseae»
73
dolanda, Egypt and the empire of Rome», JRA 19, 7593.
BOWMAN, THOMAS (1994): A.K. Bowman, J.D. Thomas,
The Vindolanda Writing Tablets (Tabulae Vindolandenses II), London.
BOWMAN, THOMAS (2003): A.K. Bowman, J.D. Thomas,
The Vindolanda Writing Tablets (Tabulae Vindolandenses III), London.
COOKE, WILD, CORK, FANG-LU (1995): W.D. Cooke, J.P.
Wild, C.R. Cork, L. Fang-Lu, «The characterisation of
archaeological textile fibres using Advanced Image
Analysis software», ATN 18-19, 16-17.
CORK, COOKE, WILD (1996): C.R. Cork, W.D. Cooke, J.P.
Wild, «The use of image analysis to determine yarn
twist level in archaeological textiles», Archaeometry 38
(2), 337-345.
CORK, WILD, COOKE, FANG-LU (1997): C.R. Cork, J.P.
Wild, W.D. Cooke, L. Fang-Lu, «Analysis and evaluation of a group of early Roman textiles from Vindolanda, Northumberland», JAS 24, 19-32.
DANNELL and WILD (1987): G.B. Dannell, J.P. Wild, Longthorpe II: The Military Works-Depot: An Episode in Landscape History, Britannia Monographs 8, London.
FINLEY (1985): M.I. Finley, The Ancient Economy, 52, London.
GOLDSWORTHY and HAYNES (1999): A.K. Goldsworthy and
I. Haynes (eds.), The Roman Army as a Community: including Papers of a Conference held at Birkbeck College, University of London, on 11-12 January 1997,
JRA Supplement 34, Portsmouth RI.
GRANGER-TAYLOR (2007): H. Granger-Taylor, «‘Weaving
clothes to shape in the ancient world’ 25 years on; corrections and further details with particular reference to
the cloaks from Lahun», ATN 45, 26-35
GRANGER-TAYLOR (2008): H. Granger-Taylor, «A fragmentary Roman cloak probably of the 1st c. CE and offcuts
from other semicircular cloaks», in ATN 46, 6-16.
MATTINGLY (1997): D.J. Mattingly, «Drawing a line beneath
the consumer city», in H.M. Parkins (ed), Roman Urbanism: Beyond the Consumer City, London, 210-218.
PEARCE (2002): J. Pearce, «Food as substance and symbol
in the Roman army: a case study from Vindolanda», in
P. Freeman et al. (eds.), Proceedings of the XVIIIth International Congress of Roman Frontier Studies held
in Amman (Jordan) 2000, BAR International Series
S1084, Oxford, 931-944.
TAYLOR (1983): G.W. Taylor, «Detection and identification
of dyes on pre-Hadrianic textiles from Vindolanda», TH
14 (2), 115-124.
SWAN (2008): V.G. Swan, «Builders, suppliers and supplies
in the Tyne-Salway region and beyond», in P.Bidwell
(ed.), Understanding Hadrian’s Wall: Papers from a
Conference held at South Shields 3rd – 5th November
2006 to mark the Publication of the 14th Edition of the
Handbook to the Roman Wall, Newcastle, 49-82.
UBL (1989): H. Ubl, «Was trug der römische Soldat unter
dem Cingulum?», in C. van Driel-Murray (ed.), Roman
Military Equipment: The Sources of Evidence: Proceedings of the Fifth Roman Military Equipment Conference, BAR International Series S476, Oxford, 61-74.
WILD (1977): J.P. Wild, The Textiles from Vindolanda 19731975, Vindolanda III, Vindolanda Trust, Bardon Mill.
WILD (1979): J.P. Wild, «Roman textiles from Vindolanda,
Hexham, England», TMJ (Washington) 18, 19-24.
WILD (1992): J.P. Wild, «Vindolanda 1985-1989: first
thoughts on new finds», in L. Bender Jørgensen, E.
Munksgaard (eds.), Archaeological Textiles in Northern
Europe: Report from the 4th NESAT Symposium 1.-5.May
1990 in Copenhagen, Tidens Tand 5, Aarhus, 66-74.
WILD (1993): J.P. Wild, «Vindolanda 1985-1988: The textiles», in C. van Driel-Murray et al., Vindolanda Research Reports NS III: the Early Wooden Forts: Preliminary Reports on the Leather, Textiles, Environmental Evidence and Dendrochronology, Bardon Mill, 76-90.
WILD (1994): J.P. Wild, «A hairmoss cap from Vindolanda», in G. Jaacks, K. Tidow (eds.), Archäologische Textilfunde – Archaeological Textiles: Textilsymposium
Neumünster 4.-7.5.1993, NESAT V, Neumünster, 61-68.
WILD (2002): J.P. Wild, «The textile industries of Roman
Britain», Britannia 33, 1-42.
WILD, COOKE, CORK, FANG-LU (1998): J.P. Wild, W.D.
Cooke, C.R. Cork L. Fang-Lu, «Vindolanda: some results of the Leverhulme Trust programme», in L. Bender Jørgensen, C. Rinaldo (eds.), Textiles in European
Archaeology: Report of the 6th NESAT Symposium 711th May 1996 in Borås, Göteborg, 85-95.
WILD and WALTON ROGERS (2007): J.P. Wild and P. Walton
Rogers, «A knotted-pile mat from the Roman fort at
Vindolanda near Hadrian’s Wall», in A. Rast-Eicher, R.
Windler (eds.), NESAT IX: Archäologische Textilfunde
– Archaeological Textiles: Braunwald, 18.-21.Mai
2005, Glarus, 71-78.