You are on page 1of 11

EDWARD BIDDULPH

FORM AND FUNCTION: THE EXPERIMENTAL USE OF


ROMAN SAMIAN WARE CUPS

Summary. Wear-patterns inside Roman samian ware vessels provide a clue


as to how the pots were used. The wear repeatedly seen in the cups,
Dragendorff 27 and Dragendorff 33, is particularly distinctive. This paper
reports the results of using reproduction cups to replicate the patterns in order
to discover how these may have been formed. The results suggest that
Dragendorff 27 was used in the kitchen as a mortar, while Dragendorff 33 was
a wine-drinking vessel. Evidence from historical sources and graffiti supports
this view, and suggests that the inhabitants of Roman Britain were conversant
with Roman ways of cooking and dining.

introduction: the evidence


Samian ware red-pink fineware coated with a glossy red slip was a highly distinctive
tableware of the Roman world. It was made at various centres in Gaul during the first, second and
third centuries AD and exported across the western provinces of the empire. A wide range of
forms was produced, although the range available in Roman Britain was generally smaller. Still,
a number of types were standard across all areas of exportation, and included vessels described
today as plates (e.g. Dragendorff types 15/17 and 18), dishes (e.g. Dragendorff 18/31 and 31),
bowls (e.g. Dragendorff 37 and 38), and cups (e.g. Dragendorff 27 and 33).
The body of samian ware was made from a very fine clay, purified through a succession
of filters. The distinctive red surface was a slip of refined liquid clay into which the pots were
dipped. The pots were manufactured on an industrial scale. Hartley (2005, 114) estimates that
each factory produced 1260 pots a day, resulting in an industry-wide total of over 15 million
vessels per year. These figures notwithstanding, the scale of the industrys output is indicated by
the large, almost house-like kilns that could take massive loads, up to 30,000 vessels in a single
firing (Webster 1996). The oxidizing kilns reached temperatures of around 1,060/80C, resulting
in a robust, hard-fired vessel, smooth to the touch, and at its best, incredibly glossy, almost
modern in appearance.
Even when recovered as small fragments on archaeological sites, samian stands out
compared with the more usual dull, coarse, greywares of Roman Britain. It must have seemed
equally fantastic to an inhabitant of Roman Britain; indeed, compared with coarseware vessels,
samian was more likely to be repaired if broken, or have an owners name scratched on it,
suggesting that samian pottery was valued as a prestige ware in the province. The limited data on
prices suggest that pots were at the high end of the pottery market; a decorated vessel cost a
OXFORD JOURNAL OF ARCHAEOLOGY 27(1) 91100 2008
2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK
and 350 Main Street Malden, MA 02148, USA.

91

FORM AND FUNCTION

Figure 1
Dragendorff 33 from Heybridge, Essex, showing typical wear-pattern. Photo: Jeff Hobson.

legionary a day-and-a-halfs pay, while a plain dish was cheaper at half-a-days pay (Hartley
2005, 116). Given these observations, we could naturally conclude that, for a Briton, samian
ware was expensive, high-status tableware; the best china which contained the exotic food of
the Roman Empire.
Yet for prestige pottery, it appears to have been incredibly well used. Excavations at
the Roman settlement at Heybridge, Essex (Atkinson and Preston 1998) produced a massive
samian assemblage of well over 3000 vessels. While recording and analysing this material
with my colleague, Joyce Compton, I noticed that many of the samian vessels were worn
internally, as if subjected to heavy or long-term use (Biddulph et al. forthcoming). The wearpatterns were particularly noticeable on cups, which contributed around 20 percent of the
samian assemblage. The wear in conical cups (Dragendorff 33) appeared as a distinctive
double-ring pattern. One ring formed around the edge of the base; a second occurred on the
wall of the vessel just above the base, the two rings separated at the junction of the base and
wall by a thin ring of unworn surface (Fig. 1). In another cup, the double-bodied or bellshaped Dragendorff 27, a spot of wear typically occurs in the centre of the cup, usually
removing the potters name-stamp and appearing to expand out with ongoing use (Fig. 2).
Similar wear-patterns have been recorded on the same types at other sites. A Dragendorff 33
cup from Seddlescomb, near Hastings, Sussex displayed the same rings of wear around the
inside of the vessel, as did a 33 from Northfleet Roman villa, Kent (E. Biddulph in prep.).
And, like those from Heybridge, a Dragendorff 27 cup from the Roman cemetery at Pepper
Hill, a little way south of Northfleet villa and outside the small town of Vagniacis
(Springhead), was worn internally in the centre (Biddulph forthcoming). Curiously, the wear
on the Heybridge cups was seen commonly on the south and central Gaulish products
generally later first and second century but was largely absent on later second and early third
century cups from east Gaul (Biddulph et al. forthcoming).
The uniformity of the patterns suggested that the cups were being used in standard ways,
but that the way that the vessels were used changed over time. Possible explanations might be
found in historical sources and on other pottery. The Roman writer Ammianus Marcellinus
alludes to the baseness of the Gauls for not drinking wine mixed with water. Perhaps added as
a reminder, or possibly for the sake of Gallic bravado, some fineware dark-slipped Rhenish
OXFORD JOURNAL OF ARCHAEOLOGY

92

2008 The Author


Journal compilation 2008 Blackwell Publishing Ltd.

EDWARD BIDDULPH

Figure 2
Dragendorff 27 from Heybridge, Essex, showing typical wear-patterns. Note the extended wear across the base of the
vessel on the right. Photo: Jeff Hobson.

ware beakers from central and east Gaul were decorated in white slip with the words, da merum
(served unmixed wine) and Misce mihi (mix for me!). Another one possibly reads, Parce
aquam (spare the water!). There is also mulsum, a drink which requires heated honey to be
mixed into wine; the drink was probably prepared at the table just before drinking, rather than in
advance in the kitchen (Grocock and Grainger 2006, 351). It is therefore possible that the
double-ring pattern in the Dragendorff 33 cups was made by people who liked to stir a drop of
water or sweetener, like honey, into their wine. Over time, the stirring lifted the surface slip and
exposed the pink fabric underneath. The wear-pattern characteristic of the Dragendorff 27 did
not appear to be consistent with stirring, and instead suggested an abrasive action, such as
grinding herbs or spices with a pestle, or eating yoghurt-like food with a metal spoon. But other
causes were possible. The ancient potters traditionally stacked the vessels inside the kilns. Sand
was sprinkled on the footring of the overlying vessel to separate it from the one below, often
resulting in a telltale ring-shaped imprint. Tiny breaks in the surface of the slip are sometimes
visible around the imprint, and so wear should be more likely to appear there first. Alternatively,
wear developed after routine stacking of the vessels on the kitchen shelf, or perhaps simply
through vigorous and repeated cleaning.

ready to wear: the experiments


One way to test these suggestions was to try them out on reproduction samian. Gilbert
Burroughes produces replica samian ware for museums and sale by mail order. For the body of
his pots, Gilbert uses a ready-prepared clay, but prepares the slip himself using similar methods
to those of the ancient potters, refining the clay to the extent that it shines when applied to the
vessel body. The shapes of the ancient pots are faithfully copied, and the completed vessels are
stacked and fired in an electric kiln, another modern compromise. But temperatures are carefully
controlled to reach those typically achieved by the Roman potters. Once fired, the kiln is left to
cool for 24 hours before drawing. A Roman kiln would have taken several days to cool, but the
quality of original and replica samian remains the same. The finished reproductions appear to be
OXFORD JOURNAL OF ARCHAEOLOGY
2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd.

93

FORM AND FUNCTION

Figure 3
A group of replica samian cups used in the experiments. The vessels with the double-curve profile are Dragendorff 27;
the conical cups are Dragendorff 33. Photo: Peter Lorimer.

identical to the Roman vessels; only Gilberts name stamp reveals the recently-made pot. His
replicas are a fair facsimile of the originals and it is reasonable to assume that they would behave
in a similar way when used (Fig. 3).
My use of the replica Dragendorff 33 cup began with stirring water into wine. The
experiment was enjoyable, although a sense of economy and sobriety led me to switch to milk
and milkshake powder; after all, it was the stirring action that was important. The contents
notwithstanding, I used a replica Roman silver-plated metal spoon to add to the authenticity of
the experiment. This was long-term work; after some two years of stirring for five minutes
perhaps two or three times a week, no sign of wear appeared. The pot did not scratch easily.
Fortunately, the effects of long-term use can be replicated with the use of food-colouring. I
coated the bowl of the spoon with the dye and stirred around the dry vessel. This left a blue ring
at the edge of the base and around the lower wall (Fig. 4). The pattern is a close, though not
perfect, match to the wear recorded on archaeological examples, but in any case, the distance
between the two rings would vary depending on the shape and size of the spoon. Indeed, metal
spoons need not have been used at all, with bone or wood stirrers being favoured. Bone objects
with long shafts and bulbous ends and conventionally identified as hair pins would make ideal
stirrers. To gain an impression of how the wear-pattern might vary depending on the shape of the
stirrer, I coated a glass stirrer with a short shaft and bulbous end in food-colouring. The resulting
pattern is a better match to the wear-patterns (Fig. 5).
A reasonable match was achieved when I stacked two Dragendorff 33 cups. I applied the
blue food-colouring to the footring of one vessel and placed it inside the second. The footring,
with its wider diameter, did not quite meet the base, and so left a ring just above the base around
the lower wall. I repeated the exercise, reversing the order of the vessels. This time, a ring was
visible around the edge of the base. If I had a number of such cups, enough for a household, and
each with slightly different footring diameters, then over time and repeated stacking, wear might
appear on the lower wall and edge of the base in the same cup (Fig. 6). However, the Dragendorff
33 was produced in a range of sizes, so another pair of cups might extend the wear to the junction
of the wall and base and therefore not result in the double-ring pattern, or conversely produce
OXFORD JOURNAL OF ARCHAEOLOGY

94

2008 The Author


Journal compilation 2008 Blackwell Publishing Ltd.

EDWARD BIDDULPH

Figure 4
The pattern created by stirring food dye with a metal spoon in a replica Dragendorff 33. Photo: Author.

Figure 5
The pattern created by stirring food dye with a glass stirrer in a replica Dragendorff 33. Photo: Author.

OXFORD JOURNAL OF ARCHAEOLOGY


2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd.

95

FORM AND FUNCTION

Figure 6
The patterns created after stacking one Dragendorff 33 cup on top of another (left), then reversing the order (right).
Photo: Author.

a much smaller ring closer to the centre. The evidence of the cups from Heybridge does not
appear to permit such variability.
Moving to the use of reproduction Dragendorff 27 cups, I ate yoghurt or similar food
out of one vessel. I used a metal spoon, which gradually lifted away the slip at the border and
ridges of the central stamp. In another vessel, I ground spices, such as peppercorns, coriander
seeds and fennel seeds, using a marble pestle. As they were crushed under the pestle, the seeds
moved away from the centre to the edge of the base as it gently curves into the wall. The
shape of the wall at that point allowed the seeds to fall back under the pestle after they were
pushed upwards. As might be expected, the centre of the vessel rapidly become worn
(although the cup itself again very hard-fired showed no sign of cracking), and much of
the central stamp eroded away. Moreover, spots of wear began to appear towards the curved
wall of the vessel, anticipating how the more developed wear of archaeological examples
might have formed (Fig. 7). Again, the wear could not have been caused by stacking. I applied
food-colouring to the footring of one Dragendorff 27 and placed it on top of a second cup. The
resulting blue ring surrounded the central part of the vessel, never coming into contact with it.
Also, it can be noted that stirring produces a similar effect, and the centre of the cup stays
clean.
I can also discount the theory that vigorous washing in an age lacking detergent removed
the slip. A few drops of a wine and honey mixture that I poured in a Dragendorff 27 cup settled
and covered the name-stamp. I left the cup unwashed for over three months until the mixture had
become sticky and mould had formed. I then washed the cup with nothing more than water and
a non-abrasive sponge. The vessel proved to have something of a non-stick, easy-clean surface,
as the wine was washed away instantly. No scrubbing was required, making it unlikely that
cleaning created the wear.
OXFORD JOURNAL OF ARCHAEOLOGY

96

2008 The Author


Journal compilation 2008 Blackwell Publishing Ltd.

EDWARD BIDDULPH

Figure 7
Dragendorff 27 used as a mortar, showing slight wear on the central stamp and around the curved edge of the base.
Photo: Author.

discussion: form and function


While the experimental use of the replica cups cannot prove how the wear-patterns seen
on archaeological examples occurred, the results are nevertheless highly suggestive. The wear in
the Dragendorff 27 cup was caused through use, rather than cleaning or stacking. Both food
preparation and eating remain possible functions. Both remove the surface slip in the centre, but
given the very robust nature of the vessel, and that wear begins to develop beyond the centre
relatively quickly, the former grinding and mixing seems on balance to be the more likely
explanation. The use of the replicas indicates that the central stamp and gently-curved wall gave
adequate purchase to the ingredient under the pestle. In contrast, the sharp angle at the junction
of the base and wall made the Dragendorff 33 unsuitable for this function, since the pestle could
not reach the seeds in the corner. Instead, Dragendorff 33 was used for drinking, stirring or
mixing liquids.
The conclusions support recent research on Roman texts. Graffiti found on the south
Gaulish samian production site at La Graufesenque record kiln loads for single firings, giving the
names and sizes of vessel types and the quantities of each (Dannell 2002). Few of these types can
be matched with known vessel types with certainty, but of the three vessels that fit the size
range offered by cups, the acetabulum, or acitabulum, sometimes with accompanying words
describing a dual aspect to the shape, appears to be best identified with Dragendorff 27 (Dannell
2006, 151). In Apicius, the collection of early to late Roman recipes, the acetabulum was
primarily a measure (Grocock and Grainger 2006, 83). This was expressed as a portion large,
small, or very small, for example in recipe 8.7.17, a sauce recipe for suckling pig (aceti
acetabulum minus a smaller cup of vinegar) or a single unit, as in 7.5.4, a sauce for roasted
OXFORD JOURNAL OF ARCHAEOLOGY
2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd.

97

FORM AND FUNCTION

meat (olei acetabulum 1 cup of oil). The double-curve profile may have been designed with
measurement in mind; the volumes of the lower and upper portions of the replica Dragendorff 27
are each approximately 100 ml, making 200 ml in total. The size of 27s found on archaeological
sites varies, and the volume of the replica vessel cannot represent a standard measurement, or
even a standard 1:1 ratio between the two portions. But the form would nevertheless serve as
a useful guide to the cook to ensure consistency of measurements when presented with the
requirements of the recipe. Small or cup could be up to the top of the lower portion, and
large or generous the whole vessel. Very small or smaller is more difficult to accommodate
in this scheme, though may be an appeal for the cook to judge. But it is also clear from Apicius
that the acetabulum was also used for food preparation, in this case mixing sauces. A sauce to
serve with Celsinian piglet (8.7.12) required the cook to blend pepper, fish sauce (liquamen), and
a little wine in the cup (Grocock and Grainger 2006, 277). If the identification of the Dragendorff
27 with the acetabulum is correct, then it is certain to have found a place in the kitchen, rather
than the dining table.
Data on vessels known from kiln-load graffiti as licuiae best fit Dragendorff 33 (Dannell
2006, 162). The term seems to relate to licui, perfect tense of liquesco, to melt, and in a
household sense may refer to the habit of mixing water with wine figuratively two liquids
melting into one or more literally warm honey with wine to create mulsum. Alternatively, licui
is more generally connected with liquid (liquor), and one cannot help but make the connection
with liquamen, the fish sauce mentioned so often in Apicius. Indeed, Grocock and Grainger
(2006, 386) draw our attention to evidence, albeit slight, that oenogarum, a type of sauce usually
including liquamen, oil and wine, could be prepared at the table to suit the tastes of individual
diners, rather like mixing an individual serving of French dressing. If so, then the image of diners
pounding away at the dining table with their personal pestle-and-mortars seems ridiculous, and
it is more probable that the mixing was achieved through stirring using pre-prepared ingredients.
Finally, two other aspects of the wear-patterns deserve comment here. The first is the
relationship between the chronology of the Dragendorff 27 and the wear seen on the cup. The
form was made in south Gaulish factories from c.AD 10 (Genin et al. 2002), reaching Britain
from AD 50 (Tyers 2003, fig. 91). Production continued in the second century at central Gaulish
factories, but numbers declined and by c.AD 160 the form had all but ceased to be made. This
was met by an increase in the proportion of Dragendorff 33 vessels, which apparently filled the
gap in the cup market (Willis 2005, section 5.3.2.3, chart 1). That this represents a change
in fashion from bell-shaped to conical cups is attractive but over-simplistic, as there was
considerable overlap. In Britain, Dragendorff 33 became popular in the late first century, while,
as noted above, the 27 continued to the mid-second century (Tyers 2003, fig. 91), giving a period
of some 100 years when the two forms appeared concurrently. If this was a change of fashion,
then it was especially prolonged. But if the decline of the 27 was based less on aesthetics and
more on function, then the fortune of one type need not be related to the other. The solution may
be found in the expansion of mortarium production, and introduction of other purpose-made
forms. While samian mixing-bowls were known from the mid-first century and available in
Britain mainly in the form of the Ritterling 12 spouted bowl, replaced by the Curle 11 flanged
bowl (cf. Webster 1996) the second half of the second century saw the development of a greater
range of purpose-made forms. The flanged bowl and successor to Curle 11, the Dragendorff 38,
reached Britain in significant numbers after AD 140. Among the Heybridge samian, and at other
sites, this form showed evidence of use-wear consistent with that of the mortarium the
specialist gritted mixing-bowl and similar to that of the Dragendorff 27, suggesting that the
OXFORD JOURNAL OF ARCHAEOLOGY

98

2008 The Author


Journal compilation 2008 Blackwell Publishing Ltd.

EDWARD BIDDULPH

form also saw kitchen use as a mixing-bowl (Biddulph 2005). Other mortar-like samian vessels
to be introduced during this time included Dragendorff 44. Samian versions of the mortarium
were also produced in the later second century, chiefly Dragendorff 43 and 45; another form,
Curle 21, was ungritted but took the form of a mortarium in all other respects. This development
was coincident with an increase in coarseware mortarium production in Britain. Centres emerged
or expanded, and the mortarium was added to the repertoires of small-scale producers serving
local markets only, among them Ellingham in Suffolk (Hartley and Gurney 1997), and
Heybridge, Essex (Biddulph et al. forthcoming). The implication is that the 27 lost its usefulness
as a result of the greater availability of specialist forms, and fell into terminal decline.
The second aspect to consider, though perhaps of a similar nature to the first, is the
apparent absence of the distinctive ring wear-pattern on east Gaulish samian Dragendorff 33
cups. This suggests that the form was not serving the same role in the later second and early third
centuries as it had done in earlier periods. In this case, the 33 may have been replaced by other
forms at this time, notably the Rhenish ware beakers mentioned at the start of this paper. The
beakers were manufactured during the later second and early third centuries (Symonds 1992),
and, given their mottos, clearly were intended for wine-drinking. The beakers may represent a
wider move away from the cup to the beaker as the main wine vessel.

conclusions
Two important inferences may be drawn from these observations. One is that the two
types of cups were used for different purposes. Samian cups cannot be cited solely as evidence
of wine-drinking; food preparation and drinking may be represented together in one ceramic
assemblage. Another is that the inhabitants of Roman Britain used the vessels seemingly
as intended. More work is required to discover whether the wear-patterns described here are
regularly seen among the samian from continental sites, and especially around samian
production regions where presumably the intended purpose of these vessels was most keenly
appreciated, but the results of this survey suggest that Britons understood Roman-style cooking
and dining, and were not simply attracted by the exotic and glossy appearance of the pots. To
what extent these Roman habits were adopted across the province is unknown, and in this
respect, quantified data are essential. Certainly, comparatively few vessels found appear to show
clear wear-patterns, and not all owners were necessarily using the pots in the correct ways.
However, this picture may be rectified if evidence of wear be routinely recorded. Wear-patterns
should be noted as a matter of course, but a note of the presence or absence of use-wear is
just as useful, especially if the assemblage is too fragmentary for fuller identification. More
generally, the experimental work begins to challenge our perception of samian pottery. It was
certainly of the highest quality, but its non-stick surface, and fine, hard-fired, fabric gave the
pottery a robustness that made it immensely durable as tableware or kitchenware. Samian ware
was not intended to stay on the shelf, to be brought out only on special occasions. It was designed
to be used. Its privileged status in the museum display cabinet reminds us of the tremendous skill
of the ancient potters in refining the clay, throwing and firing the pot. But the potters were
achieving this less for aesthetic effect, than practical excellence.
Acknowledgements

This long-gestating paper is the culmination of not just the experimental work, but many
discussions with archaeologists, Roman pottery specialists, and food historians at the conference hall, in
OXFORD JOURNAL OF ARCHAEOLOGY
2008 The Author
Journal compilation 2008 Blackwell Publishing Ltd.

99

FORM AND FUNCTION

the pub, or at the recreated convivium. I would like to record my appreciation of those contributions,
and in particular wish to thank Sally Grainger and Geoffrey Dannell for their continued interest and
encouragement. Geoffrey very kindly read and commented on the text. I am also indebted to Joyce
Compton and Mark Atkinson, both of Essex County Council Field Archaeology Unit, for their support.
Last, but not least, I am enormously grateful to Gilbert Burroughes for supplying the reproduction pottery
and for giving me the advice of an expert potter. But, of course, any errors in this paper are entirely my own.

Flat 4 Victoria Court


15 Windmill Road
Haddenham
Buckinghamshire HP17 8JA

references
atkinson, m. and preston, s.j. 1998: The late Iron Age and Roman settlement at Elms Farm, Heybridge,
Essex, 19935: An interim report. Britannia 29, 85110.
biddulph, e. 2005: Samian wear. Current Archaeology 196, 1913.
biddulph, e., compton, j. and martin, t.s. forthcoming: The late Iron Age and Roman pottery. In
Atkinson, M. and Preston, S.J. Heybridge: a late Iron Age and Roman settlement, excavations at Elms
Farm (Chelmsford, East Anglian Archaeology Report).
biddulph, e. forthcoming: The Roman cemetery at Pepper Hill, Southfleet, Kent (CTRL Integrated Site
Report Series, Archaeology Data Service).
dannell, g.b. 2002: Law and practice: further thoughts on the organization of the potteries at la
Graufesenque. In Genin, M. and Vernhet, a. (eds.), Cramique de la Graufesenque et autre productions
dpoque romaine. Nouvelles recherches. Hommages Bettina Hoffmann (Montagnac, Archologie et
histoire romaine 7), 21142.
dannell, g.b. 2006: Samian cups and their uses. In Wilson, R.J.A. (ed.), Romanitas: essays on Roman
archaeology in honour of Sheppard Frere on the occasion of his ninetieth birthday (Oxford), 14776.
genin, m., hoffmann, b. and vernhet, a. 2002: Les productions anciennes de la Graufesenque. In Genin,
M. and Vernhet, A. (eds.), Cramique de la Graufesenque et autre productions dpoque romaine.
Nouvelles recherches. Hommages Bettina Hoffmann (Montagnac, Archologie et histoire romaine 7),
45104.
grocock, c. and grainger, s. 2006: Apicius: a critical edition with an introduction and an English
translation of the Latin recipe text Apicius (Totnes).
hartley, b.r. 2005: Pots for tables; tables awaiting pots: an exercise in speculative archaeoeconomy.
Journal of Roman Pottery Studies 12, 11216.
hartley, k.f. and gurney, d. 1997: A mortarium kiln at Ellingham, Norfolk (Norfolk, East Anglian
Archaeology Occasional Paper 2).
symonds, r.p. 1992: Rhenish wares. Fine dark coloured pottery from Gaul and Germany (Oxford, Oxford
University Committee for Archaeology Monograph 23).
tyers, p. 2003: Roman pottery in Britain (2nd ed.) (London).
webster, p. 1996: Roman samian pottery in Britain (York, Council for British Archaeology).
willis, s.h. 2005: Samian pottery, a resource for the study of Roman Britain and beyond: the results of the
English Heritage funded samian project. An e-monograph. Internet Archaeology 17 (http://intarch.ac.uk/
journal/issue17).

OXFORD JOURNAL OF ARCHAEOLOGY

100

2008 The Author


Journal compilation 2008 Blackwell Publishing Ltd.

You might also like