Download file in PDF format: TRAC 1992: Roman Finds Assemblages, Towards an Integrated Approach? (pp. 33–58)
ROMAN FINDS ASSEMBLAGES, TOWARDS AN INTEGRATED ApPROACH?
Since this paper was to be given to a Theoretical Roman Archaeology conference an attempt has been made here to examine a little the relationship
between theoretical approaches and the generally pragmatic work of field
archaeology in the context of Roman finds assemblages. It is assumed for
the purposes of this paper that our basic task as archaeologists is to
examine and analyse Romano-British society by means of reasoning about
its material remains.
It does seem to this author, from his own basically pragmatic viewpoint,
that there is a rather over-wide gulf between the development and propounding of theories on Romano-British archaeology and the day to day
business of its recording and reporting, which in part reflects a division
between those employed in Universities and those working in field archaeology. These might perhaps be crassly described as the ‘two cultures’ of
Romano-British archaeology. This is not to denigrate theoretical approaches to Romano-British archaeology, they seem desperately needed as
we acquire more and more data, but seem to get bogged-down in recording it, without the commensurate expansion of our understanding of the
general processes taking place in Roman Britain.
Rather it seems to this author that theoretical approaches are generally
too divorced from the material evidence and that, in turn, is often presented in manners which do not aid its synthetic analysis. It seems far too
rarely that theories are tested against the evidence, which they usually can 34 j.EVANS
be either directly, or by deduction of their material consequences were
they to be true and then seeking these. It is recognised that the testing of
theories will be a Bayesian process, where they may not be completely refuted, but they can at least be considerably weakened by failing to be consistent with the evidence.
This author’s concern here is not to set about doing this, but to look at
the other side of the equation, that mass of data that we spend much of
our professional lives accumulating, and to suggest some lines of enquiry
we might consider developing and a general approach which might facilitate the examination of theoretical approaches.
THE SAMPLE —-
Turning to our data, the first point to be made is that we select a very odd
collection of sites from which to study the diocese. Most authorities estimate that around 90% of the population lived on rural sites, in villages or
isolated settlements, but these seem to have evoked little interest and
stimulated disproportionately few excavations and fieldwork projects compared with forts, towns and villas. Table 3.1 shows the numbers of rural
sites excavated as a proportion of all excavations recorded in Britannia
since 1970. Only excavations have been counted and those have been
counted by the year, so that sites with several seasons of excavation will
have multiple records. Most of the site classes are self-explanatory; ‘other’
principally comprises pottery kilns and temples. Excavations of field
systems have not been counted, nor have the too frequent trenches across
Table 3.1. The incidence of excavati01lS on Roman sites by site type and Britannia region
(see Fig. 3.1).
Region II III IV V VI VII VIII IX
Fort 60% 81% 62% 30% 12% 6% 0 6% 10%
Vicus 0 9% 15% 10% 0 0 0 0 0
Hillfort 7% 6% 0 0 0 0 0 0 1%
Town 5% 0 0 23% 27% 41% 60% 29% 30%
Villa 5% 0 0 15% 22% 10% 4% 23% 28%
Rural 14% 3% 21% 11% 21% 19% 8% 34% 17%
Other 10% 0 3% 11% 17% 25% 28% 9% 15%
n 42 32 39 133 180 69 25 101 178 ROMAN FINDS AssEMBLAGES 35
Region II III IV V VI VII VIII IX
Fort 49% 70% 55% 36% 8% 0 2% 14% 8%
Vicus 12% 15% 0 7% 1% 2% 0 1% 2%
Hillfort 2% 0 0 0 0 0 0 1% 0
Town 14% 0 21% 25% 41% 45% 76% 33% 30%
Villa 2% 0 0 9% 15% 5% 7% 22% 21%
Rural 14% 15% 18% 15% 25% 28% 2% 10% 18%
Other 7% 0 5% 7% 11% 22% 13% 15% 20%
n 43 47 38 151 165 65 46 72 142
Region II III IV V VI VII VIII IX
Fort 37% 85% 76% 32% 12% 3% 0 15% 4%
Vicus 2% 3% 0 15% 1% 0 0 3% 2%
Hillfort 7% 9% 0 1% 1% 0 0 3% 2%
Town 29% 0 21% 21% 30% 46% 84% 20% 22%
Villa 7% 0 0 12% 15% 5% 4% 18% 26%
Rural 19% 3% 0 14% 27% 26% 2% 21% 21%
Other 0 0 3% 5% 15% 21% 10% 20% 22%
n 59 33 33 108 122 39 49 80 134
Region II III IV V VI VII VIII IX
Fort 45% 93% 89% 37% 12% 9% 1% 8% 3%
Vicus 21% 3% 7% 11% 1% 2% 0 1% 1%
Hillfort 3% 0 0 4% 1% 2% 0 0 0
Town 18% 0 0 23% 30% 42% 73% 41% 42%
Villa 3% 0 0 2% 19% 5% 5% 25% 26%
Rural 7% 3% 4% 19% 19% 22% 0 14% 15%
Other 3% 0 0 12% 19% 20% 20% 11% 14%
n 62 29 28 91 108 65 79 92 122
Region II III IV V VI VII VIII IX All
Fort 47% 82% 71% 33% 11% 4% 1% 10% 6% 21%
Vicus 9% 8% 4% 11% 1% 1% 0 1% 1% 4%
Hillfort 5% 3% 0 1% 0% 0% 0 1% 1% 1%
Town 18% 0 11% 23% 32% 42% 75% 53% 31% 31%
Villa 4% 0 0 10% 18% 7% 5% 22% 25% 14%
Rural 13% 7% 11% 13% 24% 24% 3% 20% 18% 17%
Other 4% 0 3% 8% 15% 22% 16% 13% 18% 13%
n 210 146 140 511 589 247 210 356 607 3016 36 J. EVANS
Figure 3.1. Map of the regions by which Britannia reports excavations. ROMAN FINns AssEMBLAGES 37
Table 3.1 shows the data split into four blocks of five years-and together
as a twenty one year block. There are marked regional variations, and some
shifts of emphasis with time, but the first point to emerge is that rural sites
are grossly under-represented, overall comprising only 16.7% of the sites
listed. This global figure hides some fairly consistent regional variations.
Regions I-IV, ie Wales, Scotland, Hadrian’s Wall and northern England are
the regions in which rural sites are the most poorly represented, apart
from London (for which the reasons are fairly obvious). The home
counties, region IX, also has a fairly poor representation of rural sites,
whilst the best regions are V, VI, and VII, the Midlands, East Anglia and
the south-west. Worryingly the chronological trend seems to be against the
excavation of rural sites; in 1969-73, 1974-78 and 1979-83 they formed
18% of all excavations reported, but in the following quinquennium, 1984-
88, they fall to a mere 13% of excavations.
The totals of excavated sites shown in Table 3.1 reflect subjective perceptions of the impact of public expenditure policies since the later 1970s, falling from 799 in 1969-73 to 769 in 1974-78 and”to 657 in 1979-83 with a
slight rise to 676 in 1984-88 reflecting the rise of developer funding. It is
clear from the figures, too, that the benefits of developer funding have
been confined almost entirely to the Midlands and the south, adding
further regional bias to our national sample. The Welsh Office (region I)
seems to manage things better with a rise in the number of excavations
from 1969-1988, whilst the number of Scots sites excavated in 1984-88 was
only slightly fewer than in 1969-73.
Fulford (pers. comm.) rightly points out that these data on excavation
numbers do not necessarily reflect the expenditure on excavations on different site types and evidence of this would be interesting (although difficult to obtain including developer funded excavations).
The reluctance to excavate rural sites must reflect the attitudes of funding bodies and the planning processes which tend to assess the importance
of sites as individual units, without full regard to the development of
research strategies which might elucidate more of the economy, social relations and identity of the rural population, and of its relationships to
‘urban’ and ‘villa’ centres. As an example of this tendency the number of
fort excavations, 21 % of all sites, compared with excavations of vici, 4% of
all sites, suggests that Roman military archaeology still lacks interest in the
civilian communities associated with forts. 38 J. EVANS
QUESTIONS AND APPROACHES —-
Despite the devotion of a series of volumes to the civitates of Roman
Britain we still know little of what are the material cultures of the diocese.
The civitas volumes seem to have paid more attention to piecemeal chunks
of the history and pseudo-history of the diocese than to examining material
(culture) patterning within and across cantonal and regional boundaries.
Interesting patterns would seem to exist in the evidence, take, for example,
the division of the Dobunni into three groups on the basis of their coinage
(Selwood 1989: fig 13.11), and the inclusion of the southern of these in the
core distribution of BBI types (which rarely, if ever, travel to the Midlands,
Wales or the north), and the concentration of most hall villas within the
Dobunnic territory as defined by Selwood (Millett 1990: fig. 87; note the
civitas boundaries on this do not correspond with Selwood’s). There are
also enormous differences in the finds assemblages between highland and
lowland zone sites (see below).
THE RECORD —-
Turning to the published archaeological record, much is said of archaeological reporting as preservation by record, but the record, as compared
with the material and stratigraphic archive, is frequently incomplete and
often delphic. As with pottery reports 25 years ago, many finds reports consist of detailed descriptions and parallels of individual objects of ‘intrinsic
interest’, but with no indication of whether some, all, or a ‘representative
selection’ of the particular type has been made (or of how much the editor
Pottery reports have slowly become quantified, or rather pottery fabrics
are fairly often quantified relative to each other in publications, following
Young’s (1980) guidelines. This has been a fairly slow process, however,
Griffiths (1989: 67) commented ‘Exhaustive enquiries revealed that nowhere on the continent or in Britain could all these criteria be satisfied [of
holding the data in consistent format required for her study]. Of the five
most likely areas in Britain, only two local units could supply a consolidated
form/fabric series for their local coarsewares, only one of which was available for immediate use in 1982 when these enquiries were made. The publication record for recent excavations was overall fairly good, but the almost total lack of quantification, even on sites published since 19S0, was
surprising.’ Even when quantification has been done, and has included the ROMAN FINDS AssEMBLAGES 39
samian ware and mortaria, it is frequently impossible to discover how many
vessels of which form occur in which phase. Sometimes pottery reports
may be fully quantified, but very rarely is it then possible to establish how
many, of which finds types, occur with the pottery. An excellent example of
this is the good series of published pottery groups from a number of different rural and villa sites in the environs of Milton Keynes (Marney 1989). It
would be of interest in the light of these and their use by Griffiths (1989)
to quantify the glassware assemblages from the same groups against the
proportion of fineware and to examine other aspects of the finds assemblages. However, the full glass report by Price (1987) has been edited:
‘owing to limitations of space only the more interesting pieces have been
described below’ (Mynard 1987). Further analysis from the published
record is therefore impossible, despite the use of microfiche in the volume,
which one might expect to obviate such limitations. It is unclear which
other finds categories have been dealt a similar fate.
Fulford (1975: 134) has suggested that the absolute quantity of pottery
on Roman sites, especially in the south-east, may have been declining in the
later fourth century. This is, in principle, a testable hypothesis, with interesting implications for assertions made about the later Roman economy.
Unfortunately the test would require a record, or reasonably accurate estimate, of the volume of earth relative to the quantity of excavated pottery
from an acceptable sample of sites, but to this author’s knowledge, this information has been recorded on only two Romano-British excavations,
Lynch Farm Oones 1977) and Shipton Thorpe, even though this was done
on the former in 19771 No doubt, inter alia, excavators are deterred from
doing this by the knowledge that it is a tool of comparative method, and
that without any other sites with which to compare their data they can do
little with it apart from publishing and hoping that someone else will benefit. An analogous situation seems to have existed with pottery assemblages,
where, following Hull’s (1932) quantified report on the Yorkshire signal
stations further quantified reports followed on late Roman assemblages
(Gillam 1957; Corder 1961) but not on any earlier material.
What seems to be lacking in many published reports is any coherent
philosophy of preservation by record. Instead the finds assemblage is
fragmented into large numbers of individual specialist reports, but the
finds officer/excavator responsible for collecting these makes little attempt
to put them together again and examine trends in the assemblage as a
whole. Thus the basic aim of the reconstructability of the assemblage from
the record is often lost, despite the advent of microfiche which should 40 J. EVANS
allow reconstructability at reasonable cost. (Even if microfiche is an outdated technology that would be better replaced by read-only ASCII computer discs and printed drawings (cf. Hen Domen: Barker and Higham
It now seems that it is not only reports published some time ago to which
Pitt-Rivers (1887: xvii) comments are applicable:
Excavators, as a rule, record only those things which appear to
them important at the time, but fresh problems in Archaeology and
Anthropology are constantly arising, and it can hardly fail to escape
the notice of anthropologists . . . that on turning back to old
accounts in search of evidence, the points which would have been
most valuable have been passed over from being thought uninteresting at the time. Every detail should, therefore be recorded
in the manner most conducive to facility of reference.
The omens are still not too good as it appears that the basic ceramic evidence from urban sequences is supposed to be replaced by ceramic (only)
syntheses (Fulford and Huddlestone 1991). It also appeared that assessments of the post-excavation treatment of finds groups were likely to be
made on a material by material basis (Wainwright 1990), or, if not, then
only the more exceptional (and therefore atypical) groups were liable to be
funded, although this emphasis seems now to have been modified (Wainwright 1991a: appendix 4, section 4.2.1.ii). However, nowhere in The
Management of Archaeological Projects 2 (Wainwright 1991a) is any commitment made to the reconstructability of the assemblage and the general
system still appears to be designed to avoid publishing ‘uninteresting’ finds
and the danger of this has been emphasised by the recent Society of Antiquaries discussion paper on archives and publication.
One of the reasons that the data are treated as they are may be that there
is still little of an archaeology of Roman Britaih in the sense of an interpretation of it based primarily on the material record rather than inspired by
history or ‘pseudo-history’. Millett (1990) noted that ‘during the work on
this book I have become intensely aware that some established opinions
about the subject are based not on evidence, but on what have been called
‘factoids’. These are pieces of information which have been so commonly
repeated that they are almost indistinguishable from facts.’ Interpretation
in Romano-British archaeology is rarely ‘firmly based in material and in the
demonstrated relationships between different parts of that material’ (Reece
1988). Lacking the basic data we tend to assert its nature on the basis of im-ROMAN FINDS AssEMBLAGES
pressions and then go on to interpret these.
Collingwood (1946: 149) commented:
‘History’ said Bury ‘is a science, no less and no more.’ There is a
slang usage . .. according to which ‘science’ means natural science.
Whether history is a science in that sense of the word, however,
need not be asked; for in the tradition of European speech going
back to the time when Latin speakers translated the Greek episteme
by their own word scientia, and continuing unbroken down to the
present day, the word ‘science’ means any organized body of knowledge. If that is what the word means Bury is so far incontestably
right, that history is a science, nothing less.
He went on to say (1946: 251):
History then, is a science, but a science of a special kind. It is a
science whose business is to study events not accessible to our observation, and to study these events inferentially, arguing to them
from something else which is accessible to our observation, and
which the historian calls ‘evidence’ for the events in which he is interested.
In both of these quotations archaeology could, with little modification,
be substituted for history, but in large part a ‘scientific archaeology’ of
Roman Britain has yet to develop and the phrase is generally employed for
the piecemeal use of techniques borrowed from the natural sciences. The
study of data to observe trends for which interpretations might be advanced and tested against further data is hampered enormously by the lack
of fully and consistently published data.
INTEGRATED APPROACHES —-
This section is drawn principally from data easily at hand and does not
purport to be a review. Its aim is to illustrate some types of approach and
to sketch one or two possible trends in data which might repay further
Volume of excavated earth figures do exist for a Norman to post-medieval sequence from the Orange Grove excavations in Bath (O’Leary 1981;
Evans and Millett 1992). Figure 3.2 shows the number of roughly contemporary and Roman residual sherds per cubic metre in the medieval
sequence from these excavations and the numbers of roughly contempor-42
Figure 3.2. Bath Orange Grove pottery supply by volume of earth excavated
(after Evans and Millett 1992).
ary and Roman and medieval residual sherds in the post-medieval part of
the sequence. The sequence does offer some interesting possibilities with
the absolute quantity of Roman material being much greater than medieval. Even in the late seventeenth century in this sequence Roman residual
material is commoner than medieval residual sherds. This could merely be
a product of the nature of this specific sequence where the Roman
occupation was probably an intensively used area as it is less than 150m
from the spring, whilst the medieval site consisted of the Abbey foundations and a cemetery and the post-medieval one was a public park.
One reason why the Bath sequence might reflect a more general trend
comes from a functional analysis of the material from all periods (Evans
1985: tables 5.14 and 5.15) following the definitions used on Roman
material by Millett (1979) and Evans (1985). There is clearly much less
functional diversity in the medieval assemblage than amongst the Roman
one and this diversity is only restored in the post-medieval period. A
similar pattern can be seen in the similar analyses of a sequence from
Chester-Ie-Street, Co. Durham (Evans 1991) and the medieval to post-medieval sequence from the Newcastle Castle ditch (Ellison 1981). The lack of
functional diversity in medieval assemblages emphasises the point, more
easily accepted for aceramic periods, that pottery is employed for a different range of activities in different periods, and if the range is comparatively ROMAN FINns AssEMBLAGES 43
restricted the absolute quantity in use may well be less.
A further point emerges from these three medieval to post-medieval
sequences from the function figures (and from the sequences of basic technology (handmade/wheelmade/moulded) and surface treatment (glaze
type)). Developments seem to take place earliest in the Newcastle
sequence, followed by the Bath one, with Chester-Ie-Street trailing behind,
this would seem to reflect the social and economic status of the sites, a
castle in a major ‘port of trade’, a prosperous southern town, and a back
street site in a minor market town (cf. Weatherill 1988).
An interesting attempt to integrate interpretation of the finds assemblage from a site was made by Halstead, Hodder and Jones (1979), who utilised both variations in the levels of ceramic finewares and the classes of
bone waste, examining these in relation to the types of features excavated
and their spatial distribution. The authors concluded that ‘though in the
Iron Age there is little remaining evidence of spatial separation of activities
on the site, past separation of activities in space and/or time resulted in
segregation of different types of refuse. This is less the case in the larger
more complex Roman site where there is more mixture of different types
of rubbish’ (Halstead et al. 1979: 130).
A similar study, regrettably unpublished, was made of the well and pit
groups at Portchester utilising data on the principal pottery fabrics, fineware levels, functional groupings of the small finds and a rather more
sophisticated division of the animal bone assemblage than that employed at
Wendens Ambo (Creighton 1985). Consistent differences were demonstrated between the pit groups, characterised by a more industrial use, and
the isolated pits and wells receiving domestic debris. This, together clearly
defined discrete pit complexes, suggested a high degree of zoned activity
on the site, consistent with a well-organised, presumably military, occupation (Creighton 1985: section 6). Interestingly, this is the second re-examination of these features, the first being by Millett (1979), which seems to reflect not weaknesses in the original report but rather its usefulness in providing the full data necessary for subsequent re-evaluation.
Figure 3.3 (Evans forthcoming b) shows the ratio of various common
finds types relative to the quantity of pottery from the Roman fort at
Segontium: the numbers of nails, tile fragments and bone fragments
divided by the number of sherds of pottery. Figure 3.4 shows a similar
diagram, calculated on the same basis, for three third century sequences
from Gas House Lane, Alcester (Evans forthcoming c) and one of the later
fourth century. Ideally the figures would be best standardised by the 44
2 3 4
5 5a 5b 6 6a 7 7a 7b 8
Figure 3.3. Ratio of common finds types by period
at Segontium relative to the quantity of pottery.
9 10 lOa
volume of deposits excavated, but, this information was not available.
Standardisation by the quantity of pottery does assume that, overall, the
amount of pottery and the way it was discarded was similar in all parts of
the Roman period, not an entirely reliable assumption. However, if the
quantity of pottery is the main factor changing between phases then all
three of the ratios being examined should change together. In fact in the
Segontium sequence the data all change together only in period 7 A, for
which a fall in the quantity of pottery could be suggested.
Various interesting details appear in the Segontium sequence (Evans
forthcoming b) such as in the apparent demolition debris (period 5A) of ROMAN FINDS AssEMBLAGES
1 v 2
C12 C13 CI’ CIS C21 C22 C23 C24 C31 C32 C33 0 2 0
Figure 3.4. Ratio of common finds types by Period at Alcester,
Gas House Land. relative to the quantity of pottery.
V Bon •
• Til .. 46 J. EVANS
the period 5 structures, which appears to represent a dump of outdated
material, especially samian (King and Millett forthcoming) also contains
fairly few nails, and very little tile. This seems to suggest that nails, and
perhaps tile, were deliberately stripped from the building on demolition.
But there is a fairly consistent change in the sequence from period 6, with
the amount of animal bone rising sharply. The greater quantities of bone
may reflect a change in waste disposal practices, which would not have
needed to be as strict as previously since the population density of the area
had fallen very considerably. Periods 6A and 7 are fairly similar to 6, but
period 7 A saw an increase in both bone, tile and nails. This could reflect a
fall in the quantity of pottery in period 7 A, but given structural evidence
suggesting that the period 7 courtyard building was rather run down in this
period a real increase in the quantity of refuse left around would seem
The amount of bone further increases in period 7B, which is satisfactorily similar to period 8 with which it is at least partly contemporary. The
quantity of bone continued to increase in period 10, no doubt from the
back-filling of the pits with domestic waste, and there is a further rise in
period lOA with domestic waste being dumped in the large disused drain
across the site.
The Gas House Lane sequence (Fig. 3.4) shows similar basic trends to the
Segontium one for the 3rd and 4th centuries. As in the former sequence all
three third century sequences show a bone fragment to pottery ratio fluctuating around 1:1, this seems to be quite a general phenomenon (compare
the ‘Romanised’ sites on Fig. 3.7). The quantity of tile in the sequences
tends to suggest its use on the site from its inception and the proportion of
nails seems to remain fairly constant. The only really notable deviations in
the tile figures are in phases C21 and especially in phase C23. This latter
was the only point in the third century sequences where pilae occur and it
appears that a hypocaust has been demolished or refurbished somewhere
in the vicinity of the site. The late fourth century groups from area A,
phase D2, and phase D, dating to after c. AD 370 show an interesting
change. The lack of nails no doubt reflects the lack of buildings in the areas
with phase D deposits (whether in this phase or earlier) although there are
quantities of tile, perhaps dumped along with quantities of domestic waste.
The quantity of bone has risen markedly, as in the Segontium sequence, if
not as greatly. In both these cases it is suggested that this phenomenon reflects a breakdown in previous waste disposal practices at this period.
Similar changes can also be seen elsewhere: the tip of late Roman domestic ROMAN FINns AssEMBLAGES 47
waste on the berm outside the city wall in Lincoln (Darling 1977) or the
large spread of late fourth century material around the buildings and yard
of the villa at Beadlam, North Yorks. (Stead 1971); and the large deposit
including several pole-axed cattle skulls from the partially demolished praefurnium of the later fourth century hypocaust of the ‘Commandant’s
House’ of the fort at Binchester, Co. Durham.
It is possible to see these figures as reflecting Fulford’s (1975) suggestion
that the absolute quantities of Roman pottery was in decline in the late
fourth century. However, given that the quantity of pottery from period
lOA at Segontium was very large, as it was from the Lincoln and Binchester
deposits cited above this seems unlikely, as such, although only volume
figures will show this. It might be possible, however, that there is a dichotomy between deposits in northern and western England and those in the
south-east, given other suggestions (see below) of there being two different
ceramic cultures in these areas by the late fourth century.
Figures 3.5 and 3.6 show the proportions of finewares from assemblages
from various sites of early-mid and late fourth century date from northern
and southern England. The figures, in fact, are not quite comparable as the
northern ones (Evans 1985) include painted parchment wares, one of the
principal late fourth century finewares in the region, whilst the southern
ones are restricted to colour-coated wares, thus the northern figures are
slightly higher than they might be. It is, however, clear that whilst there is a
slight rise in the proportion of finewares in the north in the later fourth
century, this does not generally amount to a level greater than 15% and the
maximum value is below 25%.
In the south in the early fourth century the fineware level seems to be on
average more like 10-20% and on, many, but not all, sites the fineware
level rises markedly in the late fourth century. The average level is probably
in the order of 25%, but quite a number of sites produce levels over 40%.
A similar effect may be observed by comparing Hodder’s map of the proportions of samian ware (Millett 1990: fig. 54) from assemblages in the
south-west with his maps of the proportions of 3rd and 4th century finewares from the same area (Hodder 1974: fig. 8). (As Fulford (pers. comm.)
has rightly commented there are some sub-regional trends within the broad
brush approach being taken here, and these may prove of interest when
regional studies of finewares and site status are made.}
A similar dichotomy can be seen developing between north and south in
the functional use of pottery. In the third century the level of jars on sites
in the north (Evans 1985: table 5.2) seems to be similar to those in the 48
‘V 11- 201
EARLY 4th CENTURY
• Major town
• Small town
Figure 3.5. The proportions of jirurwares from various sites
in the early fourth century.
J. EVANS ROMAN FINns AssEMBLAGES
“‘l 30+ ‘I.
LATE 4 th CENTURY
• Major town
• Small town
o 100 kin
Figure 3.6. The proportions of finewares from various sites
in the late fourth century.
49 50 J. EVANS
south (Going 1987; Millett 1979; Millett 1983; Evans forthcoming c), but in
the early fourth century this rises slightly in the north to a general range of
40-50% and in the late fourth century there is a further rise to levels generally above 50% and often markedly so. Meanwhile in the south the proportion of jars often falls to its lowest level in the late fourth century as the proportion of tablewares increases with the proportion of finewares. (Again
some sub-regional trends may also be present such as the high levels of
drinking vessels on some Severn Valley region sites) (Evans forthcoming a).
Undoubtedly the two phenomena sketched out above (Evans in prep.)
deserve further study, but it seems that generally two different ‘material
cultures’ of ceramic use were developing in ‘Roman’ Britain in the late
Figures 3.7, 3.8 and 3.9 show the finds assemblages divided by basic
material types standardised to finds per square metre (in default of volume
figures this is believed to be a reasonable measure for sites without deep
stratigraphy) from Roman and early post-Roman sites. Figure 3.7 shows the
assemblages from six Roman rural sites, three, Elsted in West Sussex (Redknap and Millett 1980), Ower in Dorset (Sunter and Woodward 1987) and
Bradley Hill in Somerset (Leech 1982), from the lowland zone, Roman
Britain proprii dictu, and three, Milking Gap in Northumberland (KilbrideJones 1938), Graeanog in Gwynedd (Kelly forthcoming) and Staden in
Derbyshire (Makepeace 1983) from the highland zone. Figure 3.8 in comparison shows the same data from three post-Roman sites, all with claims of
rather higher status than the above Roman sites: Cadbury-Congresbury,
Somerset, which has been suggested to be a monastic site (Fowler et aL
1970), Ban tham , Devon, is interpreted as a port site (Silvester 1981) and
phase 3C at Cowdrey’s Down, Hampshire (Millett 1983). (For the latter site
the area figures have been adjusted and only the roofed area of the settlement in this phase has been used.) Figure 3.9, for comparison, shows finds
from area 7 for four succeeding third century phases at the small town of
Alcester, Warwickshire. The deposits here are not deep, nor do they necessarily extend over the whole area of the trench in each phase and plotted
by phase they are felt to be reasonably comparable with the other data.
These figures, from which the rest of this paper arises, were constructed
to attempt to examine and illustrate differences in the finds assemblages
between Roman and post-Roman sites. The selection of the sites is reasonably random, the main constraints being the difficulty of finding out the
quantities of pottery and animal bone from sites and small finds reports
which appear to be complete, and sites which appear to have been reason-6′
V Elsted o Ower • Bradley Hill o Staden
… Graenog m Milking Gap
Tile & m 0
Pot m &
Coins o & m IV 0
Bronze o m IA”
I. & m I
0 ’05 ·1 · 2 · 4
·8 5 10 20 40
Figure 3. 7. The frequency of finds occurrence standardised by site area from six Roman rural sites.
o Cadbury Congresbury
o Cowderys Down
0 ’05 ‘1
·4 ‘6 ‘8
Figure 3.8. The frequency of finds occurrence standardised by site area from three post-Rmnan rural sites.
5 10 20 40
Phase o C12
o C14 A C15
lead o 0 A
o • 0
0 ’05 ·1 ·2
· 4 ·6
Figure 3.7. The frequency of finds occurrence standardised by site area from three phases at Gas House lAne, Alcester.
10 20 40
t.>O 54 J. EVANS
ably fully excavated within the limits of their trenches. Even so the number
of nails from Bradley Hill is unreported as is the quantity of animal bone.
It is reasonably clear that the Roman lowland rural sites have much more
pottery, as expected, but also more nails, sometimes more iron and definitely a wider range of material types than the post-Roman sites which
were probably of a higher status. Figure 3.7 further emphasises the considerable material dichotomy between highland and lowland zone rural
sites, the highland zone sites quite often having lower finds levels than on
the post-Roman sites, although they do have a rather wider selection of
finds classes and a little more pottery. Regional variations within highland
zone sites in fact seem to be exhibited in the quantity of pottery, with
Pennine and Cumbrian sites having very little, but those in North Wales, at
least, having rather more (cf. Gidney 1986; Dore 1983; Going and Marsh
forthcoming; Evans forthcoming d, e and f). Interestingly one of the four
categories of finds types both highland and post-Roman sites tend to have
more of than lowland Roman rural sites is stone artefacts, hones, whetstones, quernstones, etc., in part this is no doubt related to geological conditions, but also probably to the fact that this material type is available
everywhere and does not necessarily rely on trade or exchange for its provision.
One interesting implication of this illustration of the highland/lowland
division between ‘Roman Britain’ and the contemporary ‘Celtic West’, is in
the role of the military vici in the ‘Celtic West’, especially in north-western
England. These were clearly not centres providing goods and services for
the surrounding countryside in the same way as such settlements, in Yorkshire for example. Leaving aside possible reasons for this, the simple lack
of reasonable quantities of ‘Romanised’ goods on the rural sites demonstrates it. It is much more probable that the vicani formed an alien blot on
the landscape, along with the forts (quite probably even speaking a different language from the surrounding rural population). The disappearance
of these sites in the 4th century, compared with the continued occupation
of such sites in Yorkshire until the end of the century is eloquent testimony
to the differences of function between these sites in the two regions (Evans
The Alcester urban sequences, in contrast to those discussed above,
might suggest that in the third century this small town may still have functioned as a central node for the distribution of goods, with much higher
densities of finds and a wider range of finds types than on the rural sites
examined. (This is not a consequence of residual material, the particular ROMAN FiNDS AssEMBLAGES 55
site being occupied de novo at this time).
It would seem from this sketchy and very preliminary survey that many
data remain to be studied in comparative examination of the finds assemblages from Roman sites, however, Honorius’s letter of AD 410, to Bruttium
or Britain (Rivet and Smith 1979), is only the end of the world for Romanists and such comparisons with both early and high medieval sites (and
even post-medieval sites) might prove revealing in a longue duree approach.
Indeed it would seem to be in the early post-medieval period that many
Roman type features of assemblages reappear, perhaps one of the benefits
of the unseen hand of money taxation and government expenditure, rather
than providence? (cf. Evans 1990; Pearce 1942; Hill 1980).
This paper has attempted to show that there is considerable scope for the
development of the study of Roman finds assemblages and a RomanoBritish archaeology, but that one of the prerequisites for this is the capacity
to express sites as a matrix of associated finds and that some method of absolute quantification such as the volume of earth excavated is needed.
Volume figures are not a panacea, and will not necessarily provide the
grounds for simple comparisons between groups: factors such as context
type will also need to be taken into account. However, in the absence of
data, the difficulties which may be encountered when they exist remain
speculative. The ability to examine sites as a matrix of associated finds will
not in itself solve any problems, but will provide a basis for the realistic
comparison of, and the systematic examination of trends in, the archaeology. Hypotheses to explain patterning in the data might then be proposed and their material corollaries tested: a sine qua non for the development of a rigourous and mature archaeology, as opposed to a history or
culture-history, of Roman Britain.
The author wishes to thank Chris Scull, Martin Millett and Mike Fulford
for comments on drafts of this paper which have improved it considerably
(particularly the former for his sterling work on the English). The author,
however, remains solely responsible for the opinions expressed and all remaining errors. The illustrations are the work of Nigel Dodds and Steve
Rigby. 56 J. EVANS
Barker, Philip and R. Higham 1982. Hen Domen, Montgomery, v. 1. London:
Collingwood, Robin G. 1946. The Idea of Nature. Oxford: Oxford University
Corder, Philip C. 1961. The Roman Town and Villa at Great Casterton,
Rutland, Third Report.
Creighton, John 1985. Portchester Castle; an Investigation into the Analysis
of Site Finds. Unpublished undergraduate dissertation, University of
Cunliffe, Barry and David Miles (eds), 1984. Aspects oflhe Iron Age in Central
Southern Britain, 191-204. Oxford.
Darling, Maggi 1977. A Group of Late Roman Pottery from Lincoln. London:
Council for British Archaeology (for lincoln Archaeological Trust, 16.1).
Dore, John 1983. The Pottery. In Nicholas Higham and G. Barri D. Jones
(eds), The Excavations of Two Romano-British Farm Sites in North
Cumbria. Britannia 14:67-70.
Ellison, Anne 1981. The pottery. In Barbara Harbottle and Anne Ellison
(eds), An Excavation in the Castle Ditch, Newcastle-upon-Tyne 1974-6.
Archaeologia Aeliana (ser. 5) 9:95-164.
Evans,Jeremy 1984. Settlement and Society in North-Eastern England in the
Fourth Century. In Peter Wilson, Richard F. J. Jones and David M.
Evans (eds), Settlement and Society in the Roman North, 43-48. Bradford.
Evans, Jeremy 1990. From the End of Roman Britain to the ‘Celtic West’.
OxfOTdJournal of Archaeololfj 9.1:91-103.
Evans,Jeremy 1991. The Chester-Ie-Street Pottery. In Jeremy Evans, Richard
F. J. Jones and Percival T. Turnbull (eds), Excavations at Chester-IeStreet 1978-9. Durham Arch. Journal 6: fiche 83-130.
Evans, Jeremy forthcoming a. The Bidford Grange (BG91) Pottery. In
Nicholas Palmer (ed.), Report on excavations at Bidford Grange, Warks.
Transactions of the Birmingham and Warwickshire Archaeological Society.
Evans, Jeremy forthcoming b. The finds synthesis. In John Casey, John L.
Davies and Jeremy Evans (eds), Excavations at Segontium 1975-79.
London: Council for British Archaeology.
Evans, Jeremy forthcoming c. The Gas House Lane Pottery. In Stephen
Cracknell (ed.), Excavations at Gas House Lane, Alcester. In Stephen
Cracknell Roman Alcester, 3, the Defended Area. London: Council for
Evans, Jeremy forthcoming d. The Graeanog Pottery. In Richard Kelly (ed.),
Report on Excavations at Graeanog, Gwynedd. Archaeologia Cambrensis.
Evans, Jeremy forthcoming e. The Pottery from Bryn Eryr, Anglesey. In
David Longley (ed.), Report on Excavations at Bryn Eryr. Archaeologia
Evans, Jeremy forthcoming f. The Ewan Rigg Pottery. In Robert Bewley
(ed.), Report on Excavations at Ewan Rigg, Cumbria. ROMAN FINDs AssEMBLAGES
Evans, Jeremy in prep. Function and Finewares; Some Trends in RomanoBritish Ceramics. Journal of Roman Pottery Studies.
Evans, Jeremy and Martin Millett 1992. Residuality Revisited. Oxford Journal
of Archaeolog;y 11(2):225-240.
Fowler, Peter J., R. S. Gardner and Philip Rahtz 1970. Cadhury Congresbury,
Somerset 1968: an Introductory Report. Bristol.
Fulford, Michael G. 1975. New Forest Roman Pottery. Oxford: British Archaeological Reports (British Series 17).
Fulford, Michael G. and Karen Huddlestone 1991. The Current State of
Romano-British Pottery Studies: a Review for English Heritage. London:
English Heritage (occasional paper 1).
Gaffney, Vince and Martin Tingle 1989. The Maddie Farm Project: an
Integrated Survey of Prehistoric and Roman Landscapes on the Berkshire
Downs. Oxford: British Archaeological Reports (British Series 200).
Gidney, Louisa J. 1986. The Pottery. In Ken Fairless and Dennis Coggins
(eds), Excavations at the Early Settlement Site of Forcegarth Pasture
South, 1974-5. Durham Arch. Journal 2:34.
Gillam, John P. 1957. The Coarse Pottery. In E. J. W. Hilyard (ed.), Cataractonium, Fort and Town. Yorkshire ArchaeologicalJournal39:224-265.
Going, Christopher 1987. The Mansio and Other Sites in the South-Eastern Sector
ofCaesaromagus; the Roman pottery. London: Council for Brit. Archaeol.
Going, Christopher and Geoffrey Marsh forthcoming. The Pottery from
Cefn Graeanog. In R. White (ed.), Report on Excavations at Cefn
Graeanog, Gwynedd. Archaeologia Cambrensis.
Griffiths, Karen 1989. Marketing Roman pottery in Second Century
Northamptonshire and the Milton Keynes Area.Journal of Roman Pottery
Hodder, Ian R. 1974. Some Marketing Models for Romano-British Coarse
Pottery. Britannia 5:340-359.
Halstead, Paul, Ian R. Hodder and Glenis Jones 1978. Behavioural Archaeology and Refuse Patterns: a Case Study. Norwegian Archaeological Review
Hill, Christopher 1980. The Century of Revolution, 1603-1714. 2nd ed.
Hingley, Richard 1989. Rural Settlement in Roman Britain. London: Seaby.
Kilbride-Jones, H . E. 1938. The Excavation of a Native Settlement at Milking
Gap, High Shield, Northumberland. Archaeologia Aeliana (Ser. 4) 15:
King, Anthony and Martin Millett forthcoming. The Sam ian. In P. John
Casey, John L. Davies, and Jeremy Evans, Excavations at Segontium
1975-79. London: Council for British Archaeology.
Makepeace, G. A. 1983. A Romano-British Settlement at Staden near
Buxton. Derlryshire Arch. Journal 103 (1984):75-86.
Marney, Pauline T. 1989. Roman and Belgic Pottery from Excavations in Milton
Keynes 1972-82. Aylesbury: Bucks. Arch. Society (monograph 2). 58 J. EVANS
Millett, Martin 1979. An Approach to the Functional Interpretation of
Pottery. In Martin Millett (ed.), Pottery and the Archaeologist. London:
London Institute of Archaeology (Occasional Publication No.4).
Millett, Martin 1983. Excavations at Cowdrey’s Down, Basingstoke 1978-8l.
Millett, Martin 1990. The Romanization of Britain; an Essay in Archaeological
Interpretation. Cambridge: Cambridge University Press.
Millett, Martin and David Graham 1986. Excavatwns on the Romano-British
Small Town at Neatham, Hampshire 1969-79. Winchester.
Mynard, Dennis (ed.) 1987. Roman Milton Keynes: Excavatwns and Fieldwork
1971-82. Aylesbury: Bucks. Arch. Society (monograph 1).
O’Leary, Timothy 1981. Excavations at Upper Borough Walls, Bath, 1980.
Medieval ArchaeololfJ 25: 1-30.
Pearce, B. 1942. Elizabethan Food Policy and the Armed Forces. Economic
History Review (ser. 1) 12:39-46.
Pitt-Rivers, A. H. L. F. 1887. Excavations at Cranbourne Chase, vol. I. London.
Price,Jennifer 1989. Glass. In Mynard 1987, 147-157.
Redknap, Mark and Martin Millett 1980. Excavations on a Romano-British
Farmstead at Elsted, West Sussex. Sussex Arch ColUictions 118:197-229.
Reece, Richard 1988. My Roman Britain. Cirencester.
Rivet, A. L. F and Colin Smith 1979. The Place Names of Roman Britain.
Selwood, Lynn 1984. Tribal Boundaries Viewed from the Perspective of the
Numismatic Evidence. In Cunliffe and Miles 1984.
Silvester, R. J. 1981. An Excavation on the Post-Roman Site at Bantham,
South Devon. Devon Arch. Society Proceedings 39:89-118.
Stead, Ian M. 1971. Beadlam Roman Villa: an Interim Report. Yorkshire
Archaeological Journal 43: 178-186.
Sunter, J. and Peter Woodward 1987. Romano-British Industries in Purbeck.
Dorset Natural Hist. and Arch. Society (monograph no 6).
Thomas, Charles 1966. Rural Settlement in Roman Britain. London: Council
for British Archaeology (Research Report 7).
Wainwright, Geoffrey 1990. The Management ofArchaeololfJ Projects. London:
Historic Buildings and Monuments Commission for England.
Wainwright, Geoffrey 1991a. The Management of ArchaeololfJ Projects 2nd edn.
London: Historic Buildings and Monuments Commission for England.
Wainwright, Geoffrey 1991b. Exploring Our Past: Strategies for the ArchaeololfJ
of England. London: Historic Buildings and Monuments Commission for
Weatherill, Lorna 1988. Consumer Behaviour and Material Culture in Britain
Young, Christopher J. 1980. Guidelines for the Processing and Publication of
Roman Pottery from Excavations. London: D.A.M.H.B. (Occasional paper