WRITING THE ROMAN EMPIRE

The paper offers: a) A critique of traditionalist Roman archaeology, including its lack of contact with such overlapping issues as an archaeology of material culture, gender relati01lS, structuration, the social meanings of power, and human agency. It is suggested that this is linked to the dominant concerns and social strategies of the influential traditionalist Romanist Establishment. b) Some examples of remedial research in the field of villa studies are the Brislington wel~ the Hambleden infant burials, and the distribution of villas in Britain. c) And, linked to the above, a brief a discussion of the opportunities for theory-building in Roman archaeology.

History cannot proceed without philosophical presuppositions of a highly complex character.It deals with evidence, and therefore makes epistemological assumptions as to the value of evidence; it describes the actions of historical characters in terms whose meaning is fIxed by ethical thought; it has continually to determine what events are possible and and what are not possible, and this can only be done in virtue of some general metaphysical conclusions ....It is equally certain that philosophy is impossible without history; for any theory must be a theory of facts, and if there were no facts there would be no occasion for theory.

Reading and Writing the Narratives
The Roman Empire once was.This is what we might call a fact, not a fantasy (although some people have fantasies about it).But the Roman Empire 1 suspect this is a rhetoric with which Collingwood may well have felt uncomfortable.
The 'theories' and the Jargon' originate for the most part in the disciplines of philosophy, anthropology and history, and the theories incorporate vital issues such as epistemology and gender theory.Professor Charles Thomas too has complained of 'awful jargon', when reviewing Gregson (1989): 'what ... does the neologism ernie mean?' (Thomas 1990, 184).If I may be forgiven for smashing back his gentle lob, I know what emic means because I have read a book on anthropology.

INTERPRETIVE FRAMEWORKS AND POWER PLAYS ----
A number of influential 'traditional' Romanists present their work as either or both of two distinct aspects of archaeology: I) discovery and recovery; 2) interpretation (or explanation).Yet there is an intermediate stage of crucial importance, which should be more explicitly debated, and which evokes Collingwood's epistemological assumptions: the formulation and adoption of the conceptual or interpretive frameworks with which to interpret that which is discovered.I'm not the only working archaeologist who is dissatisfied with many of the standard implicit conceptual frameworks for the Roman Empire.Largescale power, domination, control and submission are required subjects of study and interest, with an emphasis on the lives of the wielders of power, not on the means by which differential social power relations were created and maintained.As the means included the manipulation of meaningfully constituted material culture, including configurations of domestic and public architectural space, Roman archaeology is clearly failing to exploit its glittering wealth of material culture for theory building.
The description of powerful institutions, rather than analysis of the origins, manipulation and artificial maintenance of power, indicates all too well the nature of the concerns of the dominant group in Roman archaeology in Britain.This is partly a hangover from the days of Victorian antiquarians and some later archaeologists who saw the Roman Empire as some kind of natural precursor of the British Empire (see Hingley this volume), and were obsessed with the idea of military control and gracious living.It is clear from many writings that the Roman army was viewed unquestioningly as a positive force; and Roman villas were interpreted as the country houses of gentlemen farmers -indeed the word 'villa' entered the vocabulary of an aspiring middle class from the late nineteenth century onwards.Chedworth villa, Gloucestershire, was redesigned by Victorian antiquarians in a stylised and idealised manner, through the levelling and landscaping of the site, the perching of little roofs on surviving walls, and the building of a country house -now the museum -in the middle of the site.
It is disturbing that many Roman archaeologists have not yet eschewed Victorian-Edwardian notions of human relations (Scott 1990a).It has also been mooted that a patriarchal hierarchy of Establishment figures is especially evident in Roman archaeology, and that the Romanist hierarchy seems to 'reflect' the perceived hierarchies it studies.
I would argue with the 'traditionalists' that there are powerful and exciting reasons for change.a) Roman archaeology can avoid the more painful adventures of the fundamentalist-theorists (e.g.processualists).b) A concern with gender relations is currently revolutionising archaeology in a way that a concern with social structures did in the eighties.c) It is increasingly being accepted that political and social theory SeOTI has an influence on archaeological practice and theory, and thus on archaeological interpretation.d) Archaeology has matured enough as a discipline for us to allow ourselves alternative hypotheses to explain the same data, rather than searching for The Answer.e) Roman archaeology is so 'rich' (see below) that it can, if it so desires, make a major contribution to theory within the archaeology of material culture.
All of the above are healthy academic developments, and theoretical debate in Roman archaeology should be encouraged as an intellectual exercise, at the very least.

Dominant Concerns: 'the Comfort Zone'
Because people have their own particular idea of what their present means, depending upon such factors as the social theories which they are prepared to accept, and because this is their primary knowledge, they tend to project this present back into their past; their present is their perspective for finding meaning in the relics of the past; and with some influential Romanists this present includes the Comfort Zones of dominant social groups.
The past is seen as a powerful place, where structures and events were somehow more 'natural' and 'right' than now; and people use this past to legitimise their present.They say, 'that's the way things happened; that's the right way to do things; that's the way things should happen; these are natural and good aspirations.'It is a circular story, that becomes more intense with the telling.It is a means of social control, and has been used to regulate the activities and aspirations of 'naturally' subordinate groups.There are many presents and many pasts to be narrated, but the losers in the game are often the same subordinated groups.One group's legitimised present is gained at the expense of another group's past and future; and it is for this reason that critique and remedial research in the archaeology of women, for example, are crucial, especially at a time of a general 'backlash' by the New Right against feminism (Faludi 1991).

The Marginalisation of Women in Roman Archaeology
The comments above are no mere abstractions.There is, I believe, a pressing need for an archaeology of gender relations.Romanists should not feel exempt from this call just because they have information about 'real' women in the form of texts, epigraphy, sculpture, paintings and small finds.On the contrary, the cultural meanings of this data need to be critically assessed.Further, the information presented for study tends to represent the existence of only a small proportion of women from particular social groups, leaving the vast majority of women historically disenfranchised.These women, the majority living and working in agricultural communities, are continually screened out of archaeological narratives.Yet importance in the world -power -lies in whose stories get to be told and heard and credited with significance.What a different Roman archaeology we would have if we let those who otherwise go unconsidered tell their stories and be important within their small-world but universal lives.
But instead we have volumes like Who Was Who in the Roman World (ed.Bowder 1980), a work of 'meticulous scholarship' where 'every effort has been made to include all historical and cultural figures of importance' (ibid., 9).Only a small number of these figures are women -a sample of the sections A-D reveals a 7% share -though interestingly a much higher percentage appear in the supplementary index of persons mentioned in the text but 'not important enough to be given their own entry' (ibid.);but there is no discussion of whether this number is a fair, proportionate and useful representation of the available material evidence, so one is left to wonder how 'importance' is assessed.The entries of those women who are included tend to have certain common themes: they were the wives, mothers or daughters of important cultural figures (Julia Paula, Tranquillin a, Constantia); they were victims or perpetrators of violence (Julia Soaemias, LucilIa, Domitia Longina); they were reknowned for their beauty (Clodia, LolIia Paulina); they were reknowned for their chastity or their promiscuity (Antonia, Verginia, Clodia); their production or non-production of offspring (Agrippina the Elder, Sabina); or they were the 'real power behind the throne' (Julia Maesa, Julia Mammaea).There is a disparity between the length of the women's entries and the men's.Julia Domna receives only 17 lines, whereas relatively unknown male military and political figures receive 2 to 3 times as much text.Statilia Messalina receives only 5 lines, and Messallina only 9. The message which comes through is that women are of secondary importance.Within Julia Domna's entry, more space is given to the activities of her sons -Caracalla and Ceta -than to her studies of philosophy and literature.It is also notable that in Caracalla's own personal entry, he 'arranged to have Ceta stabbed to death', while in Julia Domna's entry 'Caracalla, having persuaded Julia to summon them for a reconciliation, stabbed Ceta to death in her arms', aSsociating her especially with a psychologically disturbing image of familial violence.

SCOTT
The Roman World (ed.Boardman et al. 1986, jacket), an Oxford History of the Classical World, is presented as the work of seventeen contributors [who] are acknowledged authorities in their field.They have taken great pains to present the latest position of modern scholarship in an accessible narrative, concentrating on those aspects of the Roman world that are both important for the understanding of the period and of lasting interest to today's reader.
But a depressing use of androcentric (sexist) and exclusionary language pervades the pages.Take Matthews's contribution, for example: 'in the cities of the Empire most men lived ... in the plain tenement blocks' (p.346).It is not good enough to counter this charge with the assertion that the word 'men' is used to denote 'people', and that the existence of women is implied.I do not want to be implied; I want historical existence.On the following page we learn that at Chedworth villa 'the owner's wealth was probably based upon stock-breeding and wool-production', but this statement begs many (undiscussed) questions of cultural structures.The assignation of ownership of wealth to a particular member of a rural working family should not be done automatically, almost by default.Matthew's paper assumes the primacy of patriarchal society throughout his discussion of the Empire, so one is left with the impression that the 'owner' of this British farmhouse raking in the 'wealth' is of course male.A promising paragraph (p.348) about the 'great majority of the population', the rural peasantry, drifts quickly into a discussion of landed magnates and 'men of substantial private means'.Women are to be found, within the confines of one page (p.357), which opens with: 'The women of the community [the Empire] would generally be seen in terms of the socioeconomic categories assigned to the men'.(Is this true?If so, why, and how?And what about ideological categories?)However, 'their rights at law were . . .much more extensive than one might have expected.'(Again, why?Why should Matthew's breathtaking assumptions lead us to expect anything?)Later in the page women appear 'sharing in their husbands' work and its organisation' (is women's work not their own work?Or are their expendable ergs the property of their husbands?),'particularly in the finer crafts ... such as perfumery'.Following this we are then launched straight into a twee passage on women's 'service occupations', viz.: prostitution, which concludes with mention of women's only other discussed activities, 'the home' and childcare.There is no discussion of the evidence for, and social and ideological implications of, the placing of the women of the Roman Empire firmly 'in the home'.I question the validity of Matthews's conceptual framework which categorises women into helpmates, whores and madonnas.
Another recent academic survey of the Empire, also titled The Roman World (ed.Wacher 1987) and written by 'acknowledged authorities' (ibid., jacket), reveals androcentric concerns amongst the contributions.In the light of large-scale ethnographic and United Nations data on agricultural societies which indicate that women do over 80% of the actual work, though they control a minority of the resulting 'wealth', Blunt's very first sentence in his 'Labour' chapter seems surprisingly forgetful of women: 'The economy of the Roman Empire was predominantly agricultural; for the rich land was the safest, and therefore the most honourable investment and most men worked on the fields' (Blunt 1987, 701).Women are at home, watching spinning wheels (ibid.).He continues, 'Most men of course lived in the country, or in small towns, many of whose inhabitants went out daily to till the adjoining fields' (p.702).All categories of labour, from 'freedmen' to debt bondage to slaves are discussed in terms of men and a male world (p. .His discussion of slave labour culminates in the observation that the presence of Black female slaves in the United States 'in healthy conditions' allowed the slave population to 'more than reproduce itself.He then asserts that 'To judge from the Roman agronomists, women and children were not much used out of doors, and ... it therefore seems probable that as a result of a deficiency in women slaves breeding could not have kept up numbers' (p.715).Leaving aside Blunt's irritatingly ethnocentric and adaptive framework for the study of slavery, it may be noted that he does not cite any specific Roman agronomists in support of his down-playing of the role of female (and child) agricultural labour.It is not wise to use unnnamed Roman literary sources to make generalisations about Empire-wide labour forces.Blunt's argument begs careful handling and critical assessment in light of the ethnographic data alone; we might also consider: the general unreliability of Roman agri cullum sources when writing about reality rather than an idealised world; their unreliability when writing about the daily lives of members of other social classes; the unrepresentative nature of these sources for non-Italian and non-slave-estate agricultural societies; and, as noted above, the tendency of some male writers, including perhaps Roman agronomists, to write of 'men' when they should perhaps write of 'men and women'.
The history of women fares little better in the hands of some women scholars, whose works on the Roman period have included desperately effete studies of 'the traditional concerns' of 'girls', 'ladies' and 'womenfolk' such as childcare, housework and gynaecological problems.

Breaking Out of the Comfort Zone
It would be infinitely more useful to study women in the Empire within critical frameworks of social classes, ideologies, changing configurations of social space, artistic and prosaic depictions of women, references to women in literature/rhetoric, and the supply and manipulation of 'female' material culture, and then to approach the 'male' world in relation to this.
In broad terms it may seem that women were and are socially and economically subordinated in very many societies.If this generalisation has any truth it is of supreme importance to the study of human societies, as are the reasons why.But is it a fact?Or is it just the way we write history which puts women down?Can writers who operate from a society which still maintains that it is a woman's biology which determines her poor economic status (Faludi 1991, passim) really hope to distinguish between natural and cultural 'reasons' for women's status in past societies?
Archaeologists may not be concerned with the future, and neither are they social engineers.Yet they have been, are, and will continue to be the pawns of social engineers, and must be explicit, possibly to the point of tedious repetition, about the origins of their interpretive frameworks.Archaeologists must particularly emphasise the cultural construction and historical specificity of social behaviour in the past, displaying a concern with how the dominant groups in past societies legitimised their decisionmaking processes through symbolic 'narratives' such as the manipulation of material culture, iconography, and mythology.Narratives have powerful social meanings, and they work because the 'reader' understands and is affected by them.Narratives drip with information and knowledge which range from the obvious and potent to the hidden, esoteric and subliminal; these can be structured as layers of interacting or even contradictory symbolism.Sometimes cross-cultural correlations may be recognised (such as the common human themes of fertility, and madonna-and-child iconography), but archaeologists should beware that what they recognise has emic meaning, and is not a carefully selected self-made narrative to support their personal or group perspective.
The dominant group in any society, when challenged with a new credo that is outside its Comfort Zone, does not easily change.Unfortunately there is much self-protection even in the bowery groves of academe, even though this self-protection is counter-productive to ongoing intellectual debate.The dominant group in academe, faced with the argument that their self-interests are actually devaluing historical analyses, will simply retreat into the Comfort Zone and protest their correctness.Hodder (1991, 15) understands power to be the ability to act in relation to interest, including the interest in controlling others and resources . . .power may be based on the control of social or esoteric knowledge rather than on the control of economic resources.
Human agents occupy the material world, permanent and decaying, constructed and demolished, exchanged and accumulated .. .a potentially powerful system of signification (Barrett 1988, 9).
On the theme of power Hodder (1991, 15) continues: goods, labour, and land have to be evaluated within a symbolic system before they can be used as the basis for social domination.
Thus he argues that the power to act assumes some knowledge of an interest, but that interest is itself constructed within a system of signification.From this he contends that therefore all power relations depend on the structures of signification within which they take place, anTI.that the central issue is the relative importance of prestige and economy in constituting power.In other words, are human actions determined more by cultural values and systems of prestige or by the distribution of economic resources?Archaeologists might hope to contribute to this debate by examining over the long term the durability of symbolic codes in the face of changing relations of production.As radical economic change takes place, associated with changes in structures of social domination, do the symbolic codes seem to determine or be determined by the changes in the economic structures?Is it possible, as Duby (1980) has argued in another study of long term processes, in Medieval Europe, that the superstructure can at times act as infrastructure, playing a dominant role?(Hodder 1991, 15-16).
The implications and potential for Roman archaeology are enormous, for here is a chance for Roman archaeology to contribute very dearly to theory building.Although Hodder is discussing social and economic transformations of the European Neolithic, he offers an underlying unifying theme: that it is sets of symbolic concepts that are central to social and economic change; we must interpret this symbolism however within its specific historical and cultural context, by teasing it out, identifying various associations, and, perhaps, oppositions (see below).
Perhaps also we might look forward to a change in the unequal social and economic power structures of Roman archaeology itself when conceptual transformations occur?
One of the failings of theoretical work in prehistorical archaeologyparticularly processualism -has been the lack of understanding of the importance of human agency.Romanists, on the other hand, tend to obfuscate the complexities of human behaviour with the introduction of 'historical' human stories, often in the style of the ripping yarn.This brings us back not only to the Comfort Zone, but to the issue of human agency, and the question of individual choice in the acceptance of codes and practices.
The degree of determinism involved in human action is discussed by Hodder.He asks what role can be given to human agents and appears to offer parameters.At one extreme he believes it possible to argue that all human actors are entirely determined by structures in relation to events (ibid., 14-15).Thus, given an array of symbolic and social structures, and given a set of concrete conditions, certain solutions can be predicted.Within this view, human agents provide the medium for the playing out of structures, but they are entirely determined by those structures.According to this view history could not have been otherwise, except perhaps for random variation or 'error'.
At the other extreme (ibid., 15) social life is generated by knowledgeable human actors, monitoring the results of their actions, conflicting in their demands, pursuing varied social goals.Even if agents act within and through symbolic and social structures, they are able to transform the structures strategically.According to this view, history could have been otherwise.It is not fully predictable.
Barrett might appear to regard human or social agency as more complex, and discusses the issue within an argument stressing that we must use archaeological evidence not as a record of past events and processes but as evidr ence for particular social processes (Barrett 1988, 6). Barrett (ibid.,[7][8] is seeking not only to break with the functionalism which characterises much of current archaeological thinking, but also with the more recent demands for an 'archaeology of meaning' as propounded by Hodder.In his own attempt to escape functionalist explanations Hodder has shifted the attention of archaeology towards considering the intentions and motivations of human agents ... we are simply moved from a position where social structures govern human behaviour to one which reasserts the primacy of the individual. And elsewhere Barratt (1990, 46-7) asserts that To engage in writing an archaeology of human agency which is understandable primarily by reference to the meaningfully constituted world within which that agency operated, does not require us to find out the meanings an object might have had for individual participants in the ancient cultural system.More to the point is the historical problem of understanding the way dominant meanings were sustained over time and space through those participants' engagement in different regions of social discourse.Clearly this argument emerges from a general theoretical consideration of exactly what we need to understand when talking about other societies.
Again, it would appear that Roman archaeology might have a valuable positive contribution to make to the debate on human agency.With its wealth of material culture in the form of houses, farms, public architecture, art and decoration, texts, an unbroken carpet of pottery over the landscape, etc., encompassing a period of many centuries, one might reasonably expect the data to provide evidence for the relationship of human agents with social and economic structures and systems of signification, and changes therein.

PROCEDURES, FRAMEWORKS AND THEORETICAL APPROACHES ---
Recent attempts by Romanists to 'get trendy' include the tactic of mentioning almost anecdotally the existence of 'the New Archaeology' (reviewed Scott 1990b: I mention the term here to dispose of it), particularly in terms of the Hypothetico-Deductive method or the testing of models in the scientific mode.But, as has been argued elsewhere, the kind of certainty hoped for by the H-D method's proponents can never be achieved.A blend of inductive and deductive reasoning is needed, with an active realisation that material culture is meaningfully constituted, that people continually create and re-negotiate narratives to maintain and subvert dominant meanings, and that the history we write about the past is itself such a narrative.
There is no one 'New Archaeology', but rather a pot brimming with ideologies and approaches, all rich with potential for making sense of the data, none of which is necessarily wrong, and none of which is necessarily right: structuralism, post-structuralism, post-processualism, critical theory, feminist archaeology, post-modernism, hermeneutics, Marxist and Marxist-structuralist theory, which all overlap to some extent, are commonly used frameworks.(Definitions of these terms are available now in many accessible publications, e.g.Hodder 1986; Renfrew and Bahn 1991.)For the orderly and ordered Romanist this may seem an intolerable state of affairs: is there no procedure?Well, no.But this is not new, as the Collingwood quotation at the start of the paper was intended to demonstrate.There are technical procedures for the recovery and processing of data, certainly, but even these are affected by interpretive priorities.Collingwood knew that the ways we deal with evidence are related to 'philosophical presuppositions' about that evidence, and our subsequent interpretations are always affected by 'epistemological assumption'.The archaeologist can taste from the pot, suck it and see.There is no hard science in archaeological interpretation; there is no hard science.Even physicists don't have 'proof -good practitioners obtain increasingly better and better descriptions of the data.
Could it then be true that 'anything goes' with regard to interpretations of the past?Could there not be fascist archaeologies, for instance?This is a question that the post-modernists have yet to answer satisfactorily, but suffice to say a socially responsible moral content is usually anticipated for theoretical approaches.For example, many conferences and publications specifically discourage discriminatory language of a sexist or racist nature.This particular issue is frequently denigrated and trivialised by senior academics, but they are presumably fighting for the territory of their Comfort Zone.I don't believe that they are stupid enough to believe recent feminist critiques to be bad academic excursions, but rather that they feel threatened.Gilchrist's recent discussion of women in archaeology (present and past) (1991,(495)(496)(497)(498)(499)(500) was a fine piece of work for both its intellectual content and its accessibility; it should interest the discipline of archaeology as a whole, as a subject per se and as a guiding strategy.The general framework for the 'engendering' of archaeology identified by other recent authors (Ovrevik 1991, 738;Cero and Conkey 1991,5) could usefully be taken on board by Roman archaeology: critique; remedial research; theory-building.

Symbols, Context, Associations and Categories: some Examples
How might we then go about critique and conduct remedial research and build theory?I believe this can best be done by methods which have a place for the traditional skills of observation and intuition: the recognition of broad archaeological contexts, associations, and categories.However the new ideas for Roman archaeology expressed above and throughout this volume may help to throw new light on old problems.

Symbolic Structures
For example, I was recently asked to consider the relative absence of villas in Britain north of the Severn-Humber line; by my reckoning only about 6 per cent of the known total of certain, suspected and possible 'villas' lie here (Scott 1988, fig. 85).The 'obvious' solution to this question is first that one shouldn't really expect to find villas in a military zone with inhospitable climate and terrain.But the gsur settlement in Tripolitania, for example, would appear to argue against this.Further, in Britain, one must consider that villas are absent from areas north of the Severn-Humber line where arable agriculture could have been and probably was practised; Collingwood (1923, 41) observed this long ago.Even if sheep and cattle farming were predominant in northern Britannia, it has always been accepted that these formed the basis of the wealth of the Cotswold villas (e.g.Matthews 1986, 347).The second reason usually given, after Rivet (1969), is that the pre-existing social and economic conditions were all wrong for villa settlement in the north of England and Wales.In fact he says (Rivet 1969, 204): Taking the distribution of villas as a whole, they present very much the picture we should expect . . .we find them to be fairly evenly distributed over that part of Britain which we defined as economically viable at the time of the conquest.
The south-east of England is deemed by Rivet to have become 'economically viable' through its late Iron Age contacts with the Roman world.Apart from the etic language, no one would really have a problem accepting that contact with the Roman world in pre-Roman Iron Age Britain was important.But unfortunately for Rivet's analysis, it ignores the fact that, after AD 43, the north of England had contact with the socio-economic structure of the Roman world for over 350 years, and the argument thus ignores its own belief in Romanisation as a civilising force and a mechanism for provincial change.There is also evidence that the earliest Roman villas in Britain were imposed, utilising a blueprint plan from Gaul and early Romano-British towns, and thus were not eagerly adopted by Britons seeking enhanced prestige (Scott 1990c, 158-9).This view is further supported by the simple fact that during the intense contact period of the late pre-Roman Iron Age, no villas were built in south-east England.
It may be that, where there is an apparent 'resistance' to the rural villa ideal in particular regions, it is not the pre-existing social and economic conditions which are all wrong, but rather the symbolic conditions which are wrong.How indeed should we expect people to take to life in a villa, a new building type involving different shapes, proportions and colours, with new ways of using social space, new ways of using fire, water and natural materials, new ways of processing and storing agricultural produce, new ways of stalling and butchering animals, new ways of dealing with the Roman economy and administration, and new types of art, decoration and furniture?Such ideological changes would have involved profound annotations to the existing cultural map.It is a lesson that many third world aid workers have been trying to teach the west for years.
Regarding northern Britannia, it has long been noted that there is a distinct paucity of Romanised material culture on the rural native sites (L.Allasonjones, pers.comm.), lending some weight to the idea that there was a conflict of symbolic codes between native and Roman.The question still remains why the Roman administration sought to impose or 'encourage' a villa landscape in the south of the province in the early years of conquest.It was of course imperial policy to build public buildings and monuments and private houses, and the effect of this would be the creation of an environment in which the administrators would feel more comfortable and in control.The distribution of villas and towns appear complementary throughout the Empire, and in Britain the early towns -Verulamium, Chelmsford, Colchester, Canterbury, Chichester, Winchester, Silchesterare in the south-east (Wacher 1974, 25), which is why the earliest villas were imposed here, and why the villa system always concentrated here.Gradually the imposition of Romanised culture did have an effect on the symbolic structures of southern Britannia, which in turn allowed for greater social and economic changes, and it can be observed that these changes, evidenced by the material culture, were a uniquely 'British' response to the new external and internal influences (Scott 1988).But the lesson is that Romanisation was possibly not the unstoppable cultural force and soughtafter commodity which some scholars have deemed it to be.

Contexts and Associations
It is perhaps ironic that whereas Romanists are adept at looking for 'parallels' and artefactual contexts for material culture, especially where small finds are concerned, they are less happy seeking associations and broad social contexts for material culture, other than trying to look one up in an historical text.This dependence on literary sources as 'objective text' is inherently problematic because the texts are themselves material culture, the authors being human agents acting within and through symbolic and social structures, whose work must therefore be interpreted through reference to these structures; but these structures in the Roman world have not yet been identified.The literary sources are also unrepresentative of the Empire's inhabitants and settlements.Varro does not travel well to northern England.And it is being demonstrated increasingly by field survey in the temperate and semi-arid lands of the Empire that cities and luxury villas were not isolated features in an otherwise empty landscape (e.g.Gaffney and Tingle 1985;Barker and Lloyd 1991); the Roman countryside was densley packed with farmsteads, small villas, villages and shrines (see van Dommelen this volume; Alcock this volume).We need archaeological theory to understand human behaviour in the Roman Empire.
It is suggested in this volume that Professor Branigan's analysis of the contents of the well at Brislington villa (1972) is undertaken in terms of a 'good story', rather than in terms of broad archaeological-social context (Poulton and Scott).That the deposits contained fragments of mosaics and a set of pewter vessels was taken as evidence of a clearing-up operation after 'the occupants were attacked in their dining room.' Parallels were sought, in the form of comparable well deposits from villas in south-western England, but only to confirm the interpretation, gleaned from the literary source Ammianus Marcellinus, that a war with the Picts occurred in Britain in AD 367.The broader archaeological context, of other pewter and debris deposits, in a variety of contexts, including those which might be termed ritual, was not explored; and the dating evidence used to assign all the well deposits to 367 is thus dubious.
Similarly, a great many conjectural stories have been put forward to explain the 97 infant burials at Hambleden villa: the burial of 'little corpses' 'secretly, after dark' (Cocks 1921, 150); the 'surreptitious evidence of unofficial births on the villa' (Johnston 1983, 11); 'evidence of infanticide' (Watts 1989, 373); and the exposure of unwanted female offspring of a slave-run establishment (Frere 1967, 266-7).If, however, the Hambleden infant burials are viewed within the broad archaeological context, we find that we are probably dealing with an infant cemetery (Scott 1988, 246ff.).At Barton Court Farm villa (Miles 1986), as at Hambleden, the infants were buried in an outer farmyard, and in an area of agricultural installations such as ovens and 'corn driers'.This means that these were not 'simply' cemeteries, for the associations between the burials and the agricultural features are part of the 'otherness' of Roman societies which is all too frequently unacknowledged.From this overall realisation it is possible to explore the meanings of these associations (Scott 1991).

Categories
This leads us to categories (see Rippengal this volume).Is the world organised by people into sets of categories (Miller 1982)?Information available for examination, such as the components and contents of a villa, can be put into categories of opposition (e.g.in, out; culture, nature; male, female; public, private; clean, unclean; sacred, profane) to enhance analytical understanding or 'reading' of the narrative.The problem, as Ian Hodder pointed out in the TRAC discussion, is how to recognise categories.But the accessible and vast material world of the Roman Empire should allow for much stimulating and rich debate on the primary contention that material culture sets reflect the organisational principles of human categorisation processes, and that it is through the understanding of such processes that we may best be able to intel'pret changes in material culture sets over time (ibid ., 17).

ENDPIECE----
It may well be that Roman archaeology with its wealth of m aterial culture can substantially contribute to the debates about archaeological critique, research and theory building.Roman archaeology should not and need not merely download the 'agendas' of prehistoric theorists into the discipline: it has its own contribution to make and its own future to build, whilst simultaneously improving its communications with the rest of archaeology.
I also hope that Roman archaeology will be able to bury the academic hierarchical elitism of the past, and become a more exciting and egalitarian THE QUESTION OF HUMAN STRUCTURES AND HUMAN AGENCY ---Roman archaeologists might fruitfully tackle the question of what is power.