Download file in PDF format: TRAC 1991: Theory and Roman Archaeology (pp. 29–38)
THEORY AND ROMAN ARCHAEOLOGY
The paper seeks to find the proper, or at least current, role of theory in Roman
archaeology. It sets up a project to study the settlement pattern of Roman Britain from
purely material sources and tries to investigate the presence or need for theory in each
of the successive steps of the project. A division is found between the gathering of
observations and their manipulation, which can be relatively explicit, and the
interpretation of the observations which has to have an an element of theory and
subjectivity. Three basic questions arise from this title. Which theories? To which areas
of Roman archaeology should they be applied? What is the purpose of such application?
Two simple, and polarised, objections against any such activities are clear:
(a) we know what happened, we don’t need theory; and (b) we do not yet
have a good enough data base to justify the application of theory.
Perhaps we ought first to sort out some objectives and so find a possible
archaeological problem which could act as an example. If our aim is to
describe a dinner-party-orgy as it might have taken place in South Italy
around the middle of the first century AD – a worthy and interesting aim –
we should apply to the Satyricon of Petronius rather than an archaeological
data base. If our aim is to analyse the changes in settlement pattern in
Britain between AD 1 and 500 then we might be on the right archaeological track. I would go so far as to say that the settlement question could
only exist in the context of information on material, that it could not have
occurred without a sample of basic material. In that sense it is a ‘material’
subject and can only be answered from material evidence. We are in the
business of archaeology. How would an archaeological answer to the question be found?
Taking nothing for granted we might be able to define’ a type fossil for
the period in question, a general Roman type fossil as a theme for those 30 REECE
five hundred years. If any recourse to history is allowed then Pompei might
be a good place to look for a specifically Roman type fossil, because there
you have a site sealed at a historically known date. If historical sources are
not allowed then tree-ring dating must be pressed into use. Rings must be
counted back from AD 1990 to AD 1, and a site with a sealed deposit of
that sort of date examined. Perhaps the London water-fronts will do if an
error of a decade or so is allowable. I do not see for the moment any
theoretical help for either tree-ring-dated deposits, or possible theoretical
refinements in the date of the destruction of Pompei. Let us, for the sake
of argument, take Samian pottery as our type fossil, our Roman Type
Fossil, or RTF; I like to use the S sometimes, and only a sickly pedant
would forbid it.
A map of finds of Samian would be a first step towards the settlement
pattern of Roman Britain. Given extensive field-walking and field-survey
the map could have quite a lot of dots on it. It would of course not be a
map of settlements in Britain 1-500, but a map of find-spots of Samian
pottery. Perhaps this is where theory comes in?
If so, our two polar objections come in as well.
(a) This map is clearly so far from the real state of Roman things that no
injection of theory could improve on what we already know from other
sources. Viz. Britain to the 18th century was agricultural, and the distribution of settlement mirrored the distribution of cultivable land. Work therefore on modern land-quality maps, and distribute on the maps random
dots in proportion to the usability of the land, and you will have a far
better, if generalised, map of settlement in Roman Britain than from any
other source. Set your parameters so that the number of dots is greater
than the number of Samian dots, and perhaps give four or five ascending
numbers of sites. Then moderate your possibilities by reference to one or
two examples of highly detailed field-work. The land will give the
generalised distribution; the moderating field-work will give the level at
which the sites are to be scattered.
(b) The Samian map is such obvious nonsense that to apply any theory to
it would simply compound the folly. We are clearly not yet ready for
Or am I already indulging fairly heavily in theory? If so I find the word
and the concept a nonsense, for all I have tried to do so far is to think in
stringently careful and analytical terms.
Can we agree to summarise so far? I doubt it, but I might as well try. The
first purpose of archaeology is to define material accurately, to know which THEORY AND ROMAN ARCHAEOLOGY 31
material is where, and when. This gives three main parts to the basic
primary functions of archaeology: definition of material, description of
spatial spread, and description of chronological context. The second
purpose may be to interpret this evidence gathered.
Before we try to go on theoretically let us try to improve our map. We
must improve our RTFs. A more common fabric than Samian pottery, but
sometimes in the same contexts, is Black Burnished Ware category 1, or
BB1, and its copies, BB2. We define it as a second RTF by its common
association with Samian, but we may note that it often occurs without
Sam ian. We can therefore plot new maps of Samian, with or without BBl.
When we take a closer look at BB 1 and take into account form as well as
fabric it becomes clear that certain forms occur with Samian in primary
contexts, and other forms never occur in those primary contexts but in
others, higher up the stratigraphic record, where Samian, if it occurs at all,
may be re-cycled. We therefore get Sam ian phase BBI dots and postSamian phase BBI dots. From constant observation of super-position of
deposits we conclude that this involves early BBI and later BBl. We now
have period maps. Some dots on the map can be associated with marks
such as soil-marks, plough-marks, aerial photographs or even humps and
bumps. Some of the sherds gathered from field spreads can be associated
with assemblages more or less described from sites more or less excavated.
So spreads can be measured and dated. Maps can reflect the dimensions
and dates of scatters. Is theory likely to elbow its way back in?
Objection (a) will now have changed to
(a2) We know the date and distribution of field scatters; we don’t need
This objection seems set to change and keep on cropping up as your
information is improved. It will not go away when the first set of problems
are overcome. It may well turn up in a different guise as follows.
(a3) If your sophisticated maps of distributions of RTF scatters is seriously, detectably, even ‘significantly’, different from what we know (i.e.
from agricultural land) then it is dearly nonsense and no amount of theory
will help it. I don’t believe it.
(b2) The data base is as good as it will ever be; why subject it to theory?
We begin to approach a possible area of agreement. In order to have a
good idea of how the past still exists in the present we need to know:
i) how the material is defined;
ii) where it is in space;
iii) when it was in time. 32 REECE
Our objectors will go on from there, in unison: and this is the nearest we
can ever come to what was, so why add theory?
Because, we answer, of the next stage. We asked questions not about
scatters of RTFs but about settJement patterns. We have to turn scatters
into settJements. In short we have to interpret. But before we go on to that
stage we need to have a closer look at some of the things we have done to
see whether theory has crept in unnoticed.
We move on now to the Definition of Material. Some ardent theorists
will insist that the material is already theory laden, and that to define it
further only adds to the theory. I see the force of argument for a flint
‘arrowhead’ or a bronze ‘razor’, where an attribution of use has been
foisted on a piece of material in order to define it, when that object is part
of an indivis”ible continuum. The dividing line between ‘spear-shaped
arrowheads’ and ‘arrowhead-shaped spears’ may either be totally arbitrary
or non-existent. In the Roman period I doubt whether this is a major problem if we stick to material objects. And it is part of my thesis that most
material of the Roman period in Britain can be described, and even
defined, in modern terms, without necessarily introducing interpretation.
For about thirty years excavators have sent me their Roman coins. They
want as many scraps of bronze to be coins as possible because far more information can be obtained from coins than from scraps of bronze. After
the first batch have come from a large excavation a second batch, much
smaller, usually follows of further hopeful scraps which have emerged from
cleaning. I therefore tend to see too many scraps and have to reject a few.
But the number of bronze scraps which I have to label ‘not a coin’ are few,
less than 1 in 1000, and the number of scraps about which I am not sure is
even less than that. For the objects I accept as coins there are places in
voluminous corpora. Note that I do not say that every ‘coin’ is described in a
corpus somewhere; this is because if you understand the material you can
find the place in a well constructed system in which your previously
undescribed coin belongs. Since the indeterminate scraps to which I can
attach no classification are so few, less than 1 in 10,000, I cannot take seriously any claim that there is a serious and important doubt about what is a
Roman coin. I therefore take the simple descriptive definition ‘Roman
coin’ to be relatively theory free. If anyone were to object that to define
something as a coin involved countJess theories of economics, market
forces, and politics, they would have completely missed one of my main
theme-songs, that we do not know what ‘Roman coins’ were to the
Romans, nor have we any idea, a priori, what they were used for. All that is THEORY AND ROMAN ARCHAEOLOGY 33
theory, and none of it follows from the descriptive definition ‘coin’.
Or Samian, which is the result of a highly centralised, organised and
standardised production. The number of borderline sherds – is it, or is it
not Samian – is small in anything other than a viciously acid soil. Snap
decisions probably affect the total archaeological record very little, and if
they do matter they could be decided chemically.
Or BBl, the result of highly standardised but loosely organised production. There is little or no doubt about fabric, and that could be resolved
petrologically. There is strong agreement on the general forms. That is a
BB 1 pie-dish, this is a BB 1 cooking pot.
The word Samian obviously carries no hints of the island of Sam os, big
or small s, pie-dishes suggest nothing about pies, nor cooking pots cooking.
But BBI is black and burnished.
Once the material has been well defined then it either has been found at
National Grid co-ordinates WXYZ or it has not. And if there is doubt, it
has not. It was found at that point in the stratigraphic sequence or this, and
if there is doubt choose the later. It was in context a or b and if uncertain it
was unstratified. It commonly occurs in upper stratigraphic groups, or
lower ones, or both.
All this, our data base, has been arrived at without any obvious application of explicit theory, and quite probably without direct involvement of
specific theory. Applications have involved certain principles which are
thought to be important in almost any endeavour. Any definition ought to
be clearly comprehensible, ought to avoid uncertainty, and ought to be
universally applicable. If possible it should be susceptible to testing by
physical or chemical means. If either definition or test involves theory the
theories are likely to be diverse rather than convergent, so any support of
definition by test will probably be in spite of the theories rather than due
to the theories. Add to definitions precise observations of what was found
where, and in what combination with other objects, and I think you have
the basis of the information.
If people wish to niggle they may. But if they wish their niggles to be
taken seriously they must show how the application of the niggle would
change the data base created by more than one per cent or keep quiet.
It is possible that potential critics will have followed me this far,just out of
bored curiosity, in the sure and certain knowledge that I shall collapse in an
untheoretical heap in the next stage – if I attempt it. For, to a number of
Roman archaeologists, the next stage must be interpretation. If I want to
pursue my original goal of changes in settlement pattern then I probably do 34 REECE
have to interpret. My tendency would be to stop at the data base and say
‘The settlement pattern is represented by the spreads of material mapped.
Changes over time can be seen on the phase maps provided.’ In fact I shall
briefly stop there, but I will go on after to see where interpretation leads.
I could say that scatters of BB 1 and Samian have been measured (5 yards
across and 20 yards long) and have been assigned to either an early group,
a late group, or a general group. All scatters of more than 50 yards across
are early or indeterminate; no late scatters are 50 yards across. Of course
no one would quote this, or perhaps even take it in. They would interpret
it, and Reece would have shown that large villas were exclusively an early
phenomenon. I would rather say that above ground rubbish heaps which
plough out to large scatters were the early fonn of rubbish disposal, and
rubbish pits, which plough out, if at all, to small scatters, were the late
form. But the villa interpretation is much more interesting.
So how do we interpret? By what theories can interpretations be constructed, compared, proved or disproved? What is the interpretative theory
which transforms what may be relatively objective categories or Roman
material into stories? What is done with the material and what ought to be
done with it?
To embroider material into interpretation we must have some concepts
in our minds, some ideas of parallels ready to apply, and we must have
some reasons for sticking the concepts to the material. You have to have
the idea of a threshing floor ready to hand in order to apply it to a layer of
worn stones set in a particular way. When it is pointed out to you that the
stones are set in, and covered with, water-borne silt you may have to give
up your interpretation in favour of a paved river ford. You must search
your mind, or the minds of colleagues, or the library, for analogies, but
give them up if they conflict with the material evidence.
We interpret by analogy. Some people are horrified at this, and want to
believe that archaeology can be more exact a science than this. They would
do well to heed Prof. Mary Hesse’s warning (given in seminars, but not yet,
so far as I know, published) that by doing this archaeology is working
alongside the ‘hard’ sciences. Analogy is not a poor relation, but, according
to a respectable professor of the History and Philosophy of Science at
Cambridge, what scientists have often done and do now.
The next question concerns the way in which analogies are chosen, and
then, how they are judged. The cynical and destructive, if truthful, answer
is that they are neither explicitly or even knowingly chosen and that only
the unpopular ones are judged. THEORY AND ROMAN ARCHAEOLOGY 35
Many Roman villas were dug up on estates. Many estates produced only
one villa, and, almost by definition, estates have one comfortable if not
stately home. There were clearly too few villas known to accommodate
more than a small part of the Roman population of Romano-Britons, and
they were too well appointed compared with other houses of the Roman
period to be for the poor. It was therefore obvious that the villa was the
home of the Roman estate owner just as the comfortable or stately home
was the house of the modern estate owner. Other facts followed, such as
occupation of the house – the house, singular – by a single family, with
dependants if necessary to fill up the rooms, and a fence round a nice
discrete property to give a comfortable few hundred acres at least. And the
majority of Romanists are too blind to distinguish this plausible and possible fiction from fact. That it has never been explained in a locus classicus
shows, not that it is a heap of rubbish, but that it is fact, for fact does not
need to be explained, only to be quoted and used.
Hence we arrive at a definition of theory in archaeology. Theories are
unsuccessful ways of arguing against facts. Facts are what are set out in the
text-books of Roman Britain. Therefore anything which goes against the
text-books is merely theory. This is neat, but it does hold a problem now
that Frere’s Britannia has imitators and dubious offspring. So as not to be
misunderstood I had better say that I regard all sub-Britannia offspring,
that is all books modelled on Britannia, as dubious, if I am polite. Britannia
by itself was internally consistent, and I admire it for that, while not agreeing with it. Imitators have jiggered with bits of it and the results contain
argumentative holes through which objectionable carts and horses can be
driven. With Britannia you either accept it whole and its world to go with
it, or you do it yourself. With various Roman Britains about, judgements
need to be made, and that can only be done with the help of theories. So,
in a sense, sub-traditional Roman Britains have fostered the need for a
theory of interpretative judgement. Nutty Roman Britains just make things
worse, or better, depending whether you want to bring interpretation out
of the closet or not.
So far have our standards of common decency declined that there are
now books on Roman Britain which show explicit interpretation. Richard
Hingley (Rural Settlement in Roman Britain, 1989) had the nerve to take a
Roman villa plan, and populate it as if it were a collection of savages’ huts.
North American savages, I think, at that. Which clearly makes it wrong,
since Rome was the colonial power, so villas ought to be interpreted by
analogy with weather-boarded, white-washed, colonial mansions. Which of 36 REECE
course helps us to understand the role of slaves.
So, to leave polemic for a little, we interpret by analogy. This has all the
known problems, in particular the impossibility of interpreting a pottery
kiln as a knuft-herping site since no one knows what knuft-herpers needed
for their job. Remains need to be consistent with the firing of pottery, as
known at present, or observed in primitive contexts, or read from early
manuscripts, or dictated by technological considerations, reaching at least
500 degrees C and staying there for some hours, in order to be interpreted
as pottery kilns. It may be that they also need to be consistent with human
beings. It is no good suggesting an interpretation which simply will not work
for human beings as we judge for ourselves. Here we may be seeing the entrance not of the interpretative theory, but the theoretical framework. It is
too loose a method to leave each person with his or her individual perceptions and peculiarities to judge the humanity of each interpretation, we
need a human scheme, a human theoretical framework, a common frame
of reference, within which to interpret. Several of these frameworks exist,
and we can both ease our own task, and that of our readers, by explicitly
siting our interpretations in a chosen model of the world. It may be the
Marxist model, or the market economy model, or the religious (flavour to
taste) model. It is not so much importing a load of theory into our
interpretation as doing our interpretation within a framework of theory.
This proliferation of theory worries me. We have theory level 1: interpreting by analogy, and all the needs of method, testing and evaluating. We
have theory level 2: the theoretical framework in which we see human beings operating. We have theory level 3: the theory to which we apply the
two lower levels, the theoretical construct which is the Roman Empire.
How do these levels relate in practice? To what extent are they explicit?
And to what extent are they controllable?
The Roman Empire is based partly in historical references and partly in
the pious hopes that material has been well gathered and used as a commentary. As it stands it is quite immune from valid theoretical niggling unless the nigglers were to start afresh and build their own Empire. Even that
would be of questionable value since it would be unique to them. We have
to live with the construct as it is, recognise that it has no more reality than
the physicists’ constantly changing pictures of the atom, and try to knock
some sense into it whenever we get the chance.
If we come down to the level of a province, of Roman Britain, then there
may be more hope. But we only have control over the theoretical construct
that is Roman Britain if we go back every time that we do anything to the THEORY AND ROMAN ARCHAEOLOGY 37
actual material. If we take secondary sources then we are incorporating
other people’s theories into our theory as if they were material facts. Thus
we may use a secondary summary, like John Wacher’s The Towns of Roman
Britain, 1974, provided we do no more than use it as a guide to the original
sources. We must draw a very hard line between talking about material and
talking about what people thought about that material. This almost restricts
us to bibliographies, for any connected text in the chapters of such books is
interpretation, and so is laced with personal theory. Such text is only usable
for a study of ‘twentieth century views of Roman Britain’, it says nothing at
all about Roman Britain.
This suggests that we can get at some aspects of Roman Britain if we go
back to our sources, but we have to push further. Wacher on towns may
guide us back to an original excavation report. Here we must be extra vigilant. We may trust a drawn section, though work at Stanwick shows how
Wheeler could bend his sections to fit in with his historical preconceptions.
We can probably trust explicitly described material since few reports describe material that was not found on the site, but found somewhere else.
Yet context is vital. It would be true for a report to say that Ptolemaic coins
had been found at Winchester; but it would be a suggestio falsi unless it
added that they were found on a spoil-heap and not in excavation. If the
context is given then it is clear that they have rather less validity than the
coins in grand-father’s box which he always said came from his allotment.
Where material is well described, this is good. Where it is not described it
is not necessarily proof of absence, for every printed and even archive
report is a selection of what was actually found.
At this point I need a parable, and the only one that springs to mind
involves the world of the word-processor. I hope it will not put off those
who are not yet properly programmed. A conference organiser decides to
publish the proceedings of the conference. Contributors are asked to supply their text on a word-processing disc. The discs come in and the editors,
not realising there are any difficulties, but thinking that all word-processing
is the same, send the discs off to a little publishing company which has
otherwise intelligent customers, and they manage to plug all the discs into
their system and print the results. The firm are surprised, but the organiser
did say it was a very abstruse conference. The editors are horrified, for
most of the papers are blocks of signs and letters and spaces which make
no sense at all. The company comes to their help and explains that it all
depends on the framework within which each contributor worked, an
Amstrad here, an IBM there and a Mac somewhere else. And even when 38 REECE
you have found out what general theoretical framework each contributor
was working in, you still have to find out the individual programme, Word,
Word Perfect, Words tar, Macwrite, that each was using to fit the basic information of words and phrases together, to interpret the material.
Our Roman Britain is just such a jumbled conference proceedings; the
trouble is that we think we can read it, and we never notice that each
person is working in an individual theoretical framework. We may sort out
the Mac user, the Marxist, from the IBM user, the liberal – I steer clear of
equating any computer with Thatcherism; it would be an actionable slur.
But we are not even aware that there are different programmes, different
ways of interpreting, and we read the results as if they all fitted together.
To try to sort things out we might have conversations more often in the
following vein. The scene is set by a non-Marxist arguing with a Marxist:
I could never work in a Marxist framework, I need freedom.
So you say that outside Marxism, and other like isms, there
is total freedom?
Which framework do you work in then?
I know you to be a Capitalist Imperialist Swine; I therefore
assume that you work in a CIS framework
That is rubbish; I work in post-Thatcherite, success oriented, Britain which is free from all theoretical constraints.
In other words, if they are needed, you cannot avoid a framework within
which to pursue archaeology. You, just by being you, have taken in so
many things that anything you do to material is heavily theory laden the
moment you begin to interpret it, and perhaps long before. Your framework may not be an identifiable, taggable framework, it might be all your
own. As a person living a full life, which is different from all other lives,
this may well be a very good thing; as an archaeologist helping to build a
picture of what went on in the Roman Empire it is disastrous. Disastrous
because no-one else knows exactly why you gathered this material in this
way and said that about it. And of course the trouble is that many archaeologists are people at the same time.
Is there a recipe for a better future? It sounds suspiciously as if everyone
must write Their Roman Britain, and all will be revealed.