Creolizing the Roman Provinces Author(s): Jane Webster Reviewed work(s): Source: American Journal of Archaeology, Vol. 105, No. 2 (Apr., 2001), pp. 209-225 Published by: Archaeological Institute of America Stable URL: http://www.jstor.org/stable/507271 . Accessed: 31/05/2012 12:04

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Archaeological Institute of America is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Archaeology.

http://www.jstor.org Creolizing the Roman Provinces JANE WEBSTER

Abstract has only just begun. New studies are now emerg- "Romanization,"a concept first discussed by the Brit- ing,4 focusing on the capacity of individuals to find ish scholar FrancisHaverfield in remains the 1905, domi- their own way of "becoming Roman" (or not). This nant for model interculturalchange in the Roman prov- work shares some fundamental characteristics with inces. on recent of Romanization,this Building critiques the which contributes to the paper suggests that Romanization-which is simply ac- present paper, replace- culturation-has meritsas a means of envisagingthe pro- ment of Romanization by putting forward a new cesses by which provincial elites adopted the symbols of framework for the analysis of contact and culture but that the is flawedwhen Rome, concept fundamentally change within the Roman provinces. This frame- applied to the majoritypopulations of the provinces. work is termed creolization. Drawing on in Caribbean and Ameri- developments In order to build new it is to can historical archaeology, it is suggested that the Ro- something helpful man provinces may more usefullybe regarded as creolized reflect on the factors informing the demise of the than as Romanized.Creolization, a linguistic term indi- old. The first part of this paper therefore reviews the of two into a cating merging languages single dia- the historical origins of Romanization and ex- lect, denotes the processes of multicultural adjustment plores the criticisms subsequently leveled against (including artisticand religious change) through which African-Americanand African-Caribbeansocieties were it. Several of these studies have incorporated his- created in the New World. It is argued here that a cre- toriographical analysis,5 and it is not my aim to re- ole perspective may fruitfully be brought to bear upon tread that ground. My purpose here is to review the material culture of the Roman provinces. Taking the decline of Romanization in order to propose a of Romano-Celtic as a case it aspects iconography study, new model. Thus, in the central section of this is argued that a creole perspective offers insights into I a new to Roman the negotiation of post-conquest identities from the paper suggest approach Britain, "bottom up" rather than-as is often the case in studies moving beyond the simplistic notion of Roman- of Romanization-from the perspective of provincial ization as a civilizing process, emulated at all lev- elites.* els of society. Building on my recent work,6 which has borrowed from developments in Caribbean "Romanization," a term first used by Francis Hav- and American historical archaeology, I suggest erfield,' defines the process by which the Roman that we should think of the societies that emerged provinces were "given a civilization."2 It remains the in the Roman provinces not as Romanized, but as dominant concept in the analysis of Roman provin- creolized.Finally, these arguments are drawn togeth- cial culture, but has recently been subjected to sus- er in a case study on the creolization of religion in tained critique, particularly in Britain.3 These crit- Roman . icisms have emerged for several reasons, but taken ROMAN VIEWPOINTS ON ROMANIZATION together they demonstrate that Romanization is a simplistic and outmoded model of provincial cul- We have a reasonably good sense of metropoli- ture change. tan Roman thinking on cultural interaction with Despite a decade of discussion on the weakness- provincial populations, and of the importance es of the Romanization model, work to replace it that the Romans attached to the dissemination

* I should like to thank DavidMattingly for his comments Romanizationstudies, from the influential earlywork ofJul- on earlierdrafts of this paper,and for his enthusiasticencour- lian (1908-1926) and de Coulanges (1891) to more recent agement of my workon creolization.Many thanks to the staff workby Goudineau (1979), Clavel-Leveque(1989), and oth- of the Departmentof History,University ofWest Indies (Mona ers, is usefullysummarized by Woolf (1998, 1-23). Campus,Kingston,Jamaica), who alsoencouraged my interest 4Recentstudies include Barrett1997a, 1997b; Forcey 1997; in the applicationof creole models to the Romanworld. Jones 1997. Two recent collections (Webster and Cooper 1996; 'Haverfield 1905-1906. Mattingly 1997) have also explored new approaches to identi- 'Haverfield 1923, 11. ty in the Roman provinces. I have chosen here to focus on the development of Ro- 5Richard Hingley's work has been particularly important in manizationstudies in Britain,since Britishscholarship has had this context. See in particular Hingley 1995, 1996. a particularinfluence on the studyof Romano-Celticart and 6Webster 1997a, 1997b, forthcoming. iconography,the theme of my case study.The historyof Gallic

209 AmericanJournal of Archaeology 105 (2001) 209-25 210 JANE WEBSTER [AJA 105 of their culture throughout the provinces.7 Ro- SHORTCOMINGS OF ROMANIZATION man culture was of course never static: it com- Romanization is simply another word for accul- prised a fluid repertoire of styles and practices turation: a concept seized upon by some Romanists altered, not least, by absorbing and adapting in- in the belief that it takes us beyond a one-sided fluences from the provinces.8 By the first century view of cultural change,'5 but comprehensively B.C., nevertheless, a set of Roman cultural values trounced in studies of intercultural contact in more had crystallized, encapsulated in the term human- recent colonial contexts precisely because it does itas ("civilization").9 Imperialism, in this context, not. To understand what is wrong with accultura- came to be regarded as a civilizing mission: it was tion, we may turn to the comments of a non-Roman- Rome's destiny and duty to spread humanitas to ist, Leland Ferguson: "Originally, acculturation sim- other races, tempering barbarian practices and ply identified mutual culture exchange between instituting the pax Romana."' Together with this people in contact. However, in recent years accul- civilizing ethos, on the other hand, went an ap- turation has commonly come to mean... 'the adop- proach to cultural interaction guided by politi- tion of traits of another group.' In social science cal pragmatism. It had been understood from the this generally means the adoption of European traits time of the conquest of Italy that political and or patterns by non-European people ... The central cultural assimilation went hand in hand, and that idea of this modern 'Eurocentric' view of accultur- the fostering of Roman cultural values among ation is that either through choice or through provincial elites was essential for the development force, non-European people in contact with Euro- of a unified ruling class. An understanding of peans gave up their traditional ways and became the importance that the Augustan and later ad- like Europeans."16 ministrations attached to provincial elites in this Ferguson is here discussing the shortcomings of respect has in turn informed recent studies of acculturative approaches to European contact with Romanization, and in particular the influential Native Americans and Africans in colonial Ameri- work of Martin Millett, discussed below." ca, but if we replace "European (s)" with "Roman (s)" Roman attitudes are important here because they we can grasp instantly that Romanization is an ac- made it possible for provincials to become Roman, culturative model of exactly the type described by not as a matter of ethnicity or even enfranchise- Ferguson. Richard Reece, who defined Romaniza- ment, but by wielding a specific cultural reper- tion simply as "foreign influence,""7 well under- toire.'2 The fact that some provincials came to iden- stood that, despite the rhetoric about cultural in- tify themselves fully with the values of Roman civili- terchange, what is really envisaged is a one-sided zation cannot be doubted,'3 but in terms of the ar- process (it is not, in the end, termed Romanization gument presented below, it is important to stress for nothing). Romanization thus does not conceive that efforts to naturalize Roman values were aimed of a two-way exchange of ideas: rather, it presuppos- by one elite (in Rome) at another (in the provinc- es a linear transfer of ideas from the center to the es).'4 A province is, however, more than simply the provinces, in the course of which provincial society sum of its elites. How, and with what success, did becomes cumulatively more Roman in its ways. Romanization operate at lower social levels? That is Why did such a model of contact and culture the question posed by this paper. change come to dominate the study of the Roman

7See Woolf (1998, 1-23; 48-76) for a helpful overviewon clusion operatedat all levelsof Romansociety. these issues. '3See here Woolf s (1998, 1-7) elegant account of the as- 8The developmentof the Romanpantheon is a casein point pirationsand deeds of the Gallicorator Eumenius, a wealthy here: Beard et al. 1998, 339-48. and powerfullate third-centurycitizen ofAutun. ' Most clearlyset out by Brunt (1976). 4Humanitas, in this context, wasa concept that specifically 10TheRoman concept of humanitasand the imperativeto defined elite ideals and aspirations:Woolf notes (1998, 55) disseminateit to the barbaroiare discussedin detail by Woolf thatdespite its applicability to humankind,humanitas embod- (1998, 54-60). Pliny the Elder (NH3.39) referred to Italyas ied "conceptsof culture and conduct that were regardedby "chosenby the power of the gods ... to gather the scattered the Romansas the hallmarksof the aristocracyin 15 particular." realmsand to soften their customs and unite the discordant Studies of the Roman provincesdrawing specifically on wild tongues of so manypeople into a common speech so that the notion of acculturationinclude Slofstra 1983; Millett 1990a, they might understandeach other, and to give civilizationto 1990b;Hanson 1994. mankind." "6Ferguson1992, 150, n. 22. ' Millett 1990a, 1990b. 7 Reece 1988, 3. ' Within limits:as Woolf notes (1998, 19), culturesof ex- 2001] CREOLIZING THE ROMAN PROVINCES 211 provinces? And in what ways has the notion been the towns, and among the upper classes. It was least modified since its inception? successful among the "peasantry," where aspects of native culture (such as religious belief) "seem to HAVERFIELD'S ROMANIZATION ( 192OS-1 96os) have survived more vigorously."25Because he regard- The German historian MommsenS' first coined ed the Romanization of urban elites as representa- the term "Romanizing," but it was the British schol- tive of the cultural development of the entire popu- ar Havefield19 who offered the first sustained anal- lation, Haverfield did not expand upon why pre- ysis of a process he termed "Romanization." Haver- Roman traditions were more vigorous among the field suggested that Rome maintained its empire peasantry. In a brief discussion of these issues, the in two ways: by organizing frontier defenses and by failure of nonelites to Romanize was put down to the fostering the growth of "internal civilization" with- "latent persistence" of folk customs-"superstitions, in the provinces.2" Haverfield termed this civiliz- sentiments, even language and the consciousness ing process Romanization, the means by which non- of nationality"26-which he argued lingered, at a Romans were "given"2' a new language, material dormant level, among rural communities. culture, art, urban lifestyle, and religion. His two COLLINGWOOD AND "FUSION" central conclusions regarding the Romanization (1930s) process were: "Firstly, Romanization in general ex- Haverfield's equation of Romanization with the tinguished the distinction between Roman and spread of superior (Roman) lifeways and material provincial, alike in material culture, politics and culture did not remain unchallenged. Colling- language. Secondly it did not everywhere and at wood's reading of the material culture of Roman once destroy all traces of tribal and national senti- Britain27 was in some respects a direct challenge to ments or fashions."22 Haverfield's viewpoint. As Collingwood wrote, "we These conclusions still inform the study of pro- cannot be content simply to assert that Britain was vincial societies today, but they present a paradox. Romanized. The civilization we have found exist- On the one hand, Romanization was deemed to be ing in even the most Romanized parts of Britain is an empire-wide process, molding diverse peoples by no means a pure, or even approximately pure, in the image of metropolitan Rome, and in the pro- Roman civilization bodily taken over by the con- cess creating new Romans: "One uniform fashion quered race. What we have found is a mixture of spread from the Mediterranean throughout cen- Roman and Celtic elements. In a sense it might be tral and western Europe, driving out native art and said that the civilization of is nei- substituting a conventional copy of Graeco-Roman ther Roman nor British, but Romano-British, a fu- or Italian art, which is characterized alike by techni- sion of the two things into a single thing different cal finish and neatness, and by lack of originality from either."2' and dependence on imitation."23 As a result, one of Collingwood's view of Romano-British culture as the lasting by-products of Haverfield's focus on the a syncretistic or hybrid culture, which resembles in uniformity of Romanization has been, as Woolf re- certain respects the creolization model proposed cently put it, "a tendency for [provincial] studies to here, has left an important legacy in Romano-Brit- focus on cultural homogenization, sometimes seen ish studies.29 But what is not found in Collingwood's as inevitable, rather than on the creation of cultur- work, or that of his inheritors, is a sense that fusion al difference."24 processes cannot be studied in isolation from the On the other hand, however, Haverfield argued consideration of power (that is, an acknowledgment that Romanization was arrested or assisted accord- of the fundamental inequalities of the relationship ing to the various political and economic structures between the colonizer and the colonized). This it encountered in the provinces. Within Britain, consideration underpins all recent writing on cre- Romanization was most successful in the lowlands, olization in the Americas, where cultural fusion,

" Mommsen 1885. " Haverfield 1923, 22. 't Haverfield 1905-1906. 27Collingwood 1932. 2"Haverfield 1923, 10-11. 28 Collingwood 1932, 92. ' Haverfield 1923, 11. 29Collingwood has had a particularinfluence on approaches 22Haverfield 1923, 18. to the development of Romano-Celticart and religion, most 2"Haverfield 1923, 19. clearlyseen in the work of Martin (1984, 1995) and 24 Henig Woolf 1998, 15. MirandaGreen (1996, 1989, 1997, 1998). 25Haverfield 1923, 21. 212 JANE WEBSTER [AJA105 among nonelites in particular, is studied in terms troduced for the first time the notion of resistanceto of the desire to maintain indigenous beliefs, tradi- the overtures of Roman culture, completely invert- tions, and language, while simultaneously, and of- ing Haverfield's conception. The first of these na- ten tactically, adapting aspects of a dominant mate- tivist strands emerged in North Africa, a region rial culture. For these scholars, the syncretisms de- which had experienced recent colonization by Eu- veloped by modern creole communities are a bal- ropeans, and within which the notion of empire in ancing act, in which the complex relationship be- general as "given" civilization was met with consid- tween power and identity is always to the fore. In erable skepticism.33 In the west, a different type of Collingwood's work, fusion is argued to have taken nativist countermodel emerged that can be traced place differentially,3" with the upper classes and back to Haverfield's own uncertainties34 about the the towns at one end of a sliding scale, and the extent to which the peasantry could really be said lower classes and the villages at the other. Fusion is to have Romanized. This nativist strand echoed seen as a problem-free process at all levels of soci- ideas expressed in the writing of Vinogradoff in ety, however, taking place beyond the politics of the early years of this century,35 but mainly flour- power, with a dynamic requiring no more elucida- ished in Britain in the 1970s and 1980s.36 For the tion than Haverfield had offered. Collingwood's nativists, who pointed to the slow uptake of Latin, view of the villages of Roman Britain is thus much the rapid demise of towns, and an apparent Celtic like that of Haverfield's image of the peasantry: "a revival in the later empire,37 Romanization was lit- stratum of population in whose life the Roman ele- tle more than a surface gloss beneath which Celtic ment appears hardly at all; if we must still call their lifeways survived unscathed. Side by side with Ro- civilization Romano-Celtic, it is only about five per manization, as Reece expressed it, the nativist cent Roman to ninety five Celtic."31We are no near- school thus envisaged a "British way, which most er here to addressing why that Roman 5%, and not people in Britain followed before Romanization more, was adopted, than we were with Haverfield. began, kept to while Romanization was in full flood, Nevertheless, although Collingwood nowhere ad- and which came back into fashion, or rather be- dressed the issue specifically, it is clear that he came the general way, when Romanization was no shared Haverfield's belief that Celtic survivals could more than a symmetrical memory."38 In this Roman not have represented either resistance to Roman Britain, to paraphrase Forcey,39a Roman veneer was culture or an adaptation of it, arguing that the Celt- applied to Celtic woodwork. That is, a tactical use ic revival among un-Romanized peasants during the of the symbols of Romanitas took place in public, late Roman period was in no sense a sign of disaf- but behind closed doors, the majority of Britons fection with Rome.32 On this level, then, Colling- declined to become Romans. For the British nativ- wood's work did not represent a shift in focus from ist school then, the Roman way was neither em- Haverfield's reading of Romanization as the rapid braced nor resisted, but largely ignored. civilization of native elites, and the more halting The difficulty with the British nativist model was dissemination of some aspects of that civilization to that, by polarizing Roman and native identities and the poor. material culture, it failed to explain the emergence of those features of provincial culture (including THE NATIVIST COUNTERATTACK aspects of religion and art) that were evidently Ro- (197os-198os) mano-Celtic hybrids, and that scholars such as Col- The 1970s and 1980s witnessed a new develop- lingwood, themselves challenging Haverfield's ment. "Nativist" counterapproaches emerged, polarized approach to Roman and native culture stressing the purity of indigenous cultural tradi- (civilization versus latent persistence), had so ably tions in the post-conquest era. Nativist scholars in- identified. It is worth noting here that in the Amer-

30 Collingwood 1932, 92. 36The work of RichardReece on the decline of the 31 rapid Collingwood1932, 92. symbolsof Romanitasinsome urban in RomanBritain 2 settings Collingwood 1932, 93. was particularlyinfluential here: see Reece 1980, 1988. A na- :3Laroui 1970;Benabou 1976a, 1976b. For a comprehen- tivistperspective also influenced studies of Romano-British sive overviewof the rise of the nativistposition in the Magh- settlement(Smith 1978; Hingley 1989) and ritual(Scott 1991). reb (which focused on overt resistance to Rome), see Mat- 37Haverfield 1923, 60-8. tingly 1996. 3 Reece 1988, 74. 34Haverfield 1923, 59. 39Forcey1997, 17. 3 Forcey 1997. 2001] CREOLIZING THE ROMAN PROVINCES 213 icas, a very similar charge to that levied against the at the same time placing the motor for the adop- nativists was also raised in reaction to early attempts tion of the symbols of Romanitas firmly in the hands to identify "Africanisms"(African-derived cultural of native elites. In Millett's model, native elites, who traits) in the archaeological record of slave com- had been given the power to govern, provided that munities, a practice which in the 1980s was an im- power was exercised in broad accordance with Ro- portant antecedent to the emergence of an archae- man principles, emulated Roman material culture ology of creolization.4" Some of these early studies in order to reinforce their social position.45 The fixed on the identification of "African survivals" as symbols of Romanitas, which provided status indi- a means by which Africans could be shown to have cators for the elite and set them apart from the re- resisted European lifeways. It quickly became ap- mainder of society, were gradually adopted lower parent, however, that this resistance-based ap- down the social scale through a self-generating pro- proach to African survivals actually hampered the cess of progressive emulation. In these terms, Mil- elucidation of the part African lifeways had played lett recast Romanization as an active process rather in the emergence of creole African American iden- than a passive one, and the native population (or tities. In Romano-British studies, for similar reasons, native elites at least) were given a part in shaping a the nativist model became a cul-de-sac, but in this new social structure that "owes as much to the na- case it was bypassed not by developing a new ap- tive as to the Roman ingredient."46 proach to the integration of Roman and Celtic life- Millett's approach to the Romanization of elites ways (as happened in the Americas), but by falling owed a debt not simply to nativist approaches, but back on the trend of thinking about Romanization to broader discussions about the role of Rome it- as the gradual triumph of one set of lifeways over self in this process. Throughout the 1980s and another, which has continued from Haverfield to 1990s, debate grew as to whether Rome had a delib- the present. erate policy of Romanizing her subjects.47 Some ten- At the same time, however, the nativist counterat- sion between notions of state-sponsored Romaniza- tack also left an important legacy. Its effects can be tion and hands-off Romanization (whereby the state seen in Martin Millet's reworking of Haverfield's was assumed to trust that the empirically persua- model in the early 1990s, in which he explicitly sive charms of Roman culture would be enough to addressed the nativists' criticism that Haverfield's ensure its adoption), has in fact always been present. Romanization gave indigenous Britons no active Haverfield also had been uncertain about this, stat- role in the development of post-conquest material ing that "the advance of this Romanization followed culture. manifold lines. Much was due to official encour- agement by statesmen who cherished the ideal of MILLETT'S ROMANIZATION (1990s) assimilating the provinces or who recognized more Haverfield's view of the Romanization process has cynically that civilized men are easier to rule than been superseded by Millett's influential model of savages. More, perhaps, was spontaneous."48 Romanization as native-led emulation.41 This mod- In recent years, it has become widely accepted el, as Millett himself stated,42 built on Haverfield's that Augustus ushered in a new era, during which foundations, but it made two major advances. First, the power of imagery emanating from the metro- Millett successfully attempted to reconcile Haver- politan center consolidated imperial authority and field's view that the provinces were "given a civiliza- aided the Romanization of provincial elites. At one tion"43with the British nativists' contention that the end of the spectrum, some scholars have seen this indigenous population played an active role in ac- new cultural program less as a deliberate policy than cepting or rejecting Roman culture. He achieved as a naturally evolving process;49 at the other, schol- this by accepting Haverfield's contention that Ro- ars have argued that Augustus and his successors manization was largely a spontaneous process,44 but were engaged in deliberate state intervention.:0

4"The transitionfrom the searchfor Africanismsto a focus tion process are set out explicitlyin Millett 1990b, 37-8. on AfricanAmerican creolization in the archaeologyof slavery 46Millett1990b, 37. in the United Statesis described by Singleton (1999, 7-8). 47See Millett 1990b;see also Woolf 1998, 22 n. 74. 41 Millett 1990a, 1990b. 48Haverfield 1923, 14. 42 Millett 1990a, 1. 9 Zanker 1990. 4 Haverfield 1923, 11. 50Aposition advanced most forciblyby Whittaker(1997, 44Haverfield1923, 14. 143-63). 45 The key features of Millett's reading of the Romaniza- 214 JANE WEBSTER [AJA 105

Millett came down firmly on the side of Zanker's the relationship between native elites and Rome, notion that the rapid adoption of the metropolitan and both are based on a belief that the impetus for cultural program in the west was a result of sponta- provincial change was emulation of Roman culture. neous competition between natives "within every social class."5' He thus saw Romanization "as a re- EMULATING ROME sult of accidents of social and power structures rather Haverfield was convinced that what Rome offered than deliberate actions."52 native societies was self-evidently better than their Another important feature of Millett's model was own culture. The reasons for this were, however, more to shift attention away from provincial homogene- complex than a simple faith in the virtues of under- ity (with which Haverfield had been concerned) floor heating and painted wall plaster. Haverfield's towards provincial differences. In the 1980s and writing was underpinned by the belief that in civiliz- 1990s, work on the western provinces placed a ing the provinces, Rome fostered the values of the much greater focus than Haverfield had done upon modern Western world.57 For Haverfield, Romaniza- the nature of interaction between later pre-Roman tion was inevitable because it was nothing less than Iron Age societies and the persistence and trans- the triumph of the classical cultural values on which formation of those societies following their incor- his own European, imperialist, turn-of-the-century poration into the empire.53 This important devel- worldview was itself based. Until quite recently, Ro- opment has been summarized by Jones, who has manization continued to be conceptualized in such argued that although this work has contributed progressive terms, summarized in Sellar and Yeat- significantly to our understanding of provincial man's comment that "the Roman Conquest was, how- societies, it has remained "almost exclusively con- ever, a Good Thing, since the Britons were only na- cerned with the emulation of Roman material cul- tives at the time."58Several factors informed this pro- ture in the legitimation of political power."54 gressive approach, not the least of these being that This leads to some major difficulties. First, it leaves the Romans believed it themselves.5' But does this intact Haverfield's implicit contention that the mean we must believe it too? Romanization of elites was the only Romanization As Richard Hingley has argued,6" it must be re- that mattered (as Haverfield wrote, "the rustic poor membered here that provincial Roman archaeol- of a country seldom affect the trend of its histo- ogy developed against the backdrop of modern ry").5 Second, it ensures that provincial heteroge- European imperialism, which reached its maxi- neity is largely explained as a reflection of the het- mum extent with the "Scramble for Africa" in erogeneity in Iron Age societies (and the differen- 1875-1900, and which was itself popularly regard- tial degrees of pre-Roman Romanization to which ed as a civilizing force. Common sense suggests those societies were exposed), rather than as a re- that these two imperial projects became to some flection of post-conquest choices. In other words, it degree conflated in the minds of the scholars who, is still assumed, as Haverfield had assumed, that in the early 1900s, formulated the cornerstone had provincial populations entered the period of texts on the Roman provinces. Considerable de- Roman fashions in what Richard Reece termed "a bate has raged on this topic,6' but for Hingley, the state of grace,"56Romanization would have proceed- most persistent critic of Romanization, the con- ed at an equal rate everywhere. cept says more about 19th-century perceptions of Things may have moved on in Romanization stud- European colonial culture and government than ies, then, but Millett's model is actually very similar it does about the Roman world. He argues that to Haverfield's: both are primarily concerned with Haverfield's understanding of Roman Britain

51Zanker 1990, 316. the civilizedlife found no period in which to grow firm and 52 Millett 1990b, 38. tenacious,civilization would have perished utterly. The culture 5:Influential studies on the transformationof Iron Age so- of the old worldwould not have lived on, to form the ground- cieties include Millett (1990a, 9-39) and the work of Colin workof the best culture of today." Haselgrove:see in particularHaselgrove 1984, 1987, 1990. 58Sellar and Yeatman 1975, 11. 54Jones1997, 35; see also 29-39 for a discussionof the re- 5"See supra n. 10. cent focus on the transformationof Iron Age societies. 6"Hingley 1995, 1996, 1997. 55 Haverfield 1923, 60. 6'Webster and Cooper 1996 contains two papers on this 5'Reece 1988, 5. topic,offering strongly contrastingviewpoints (Freeman 1996; 57This point is made explicitlyby Haverfield(1923, 10-1). Hingley 1996). He states the opinion that "hadRome failed to civilize,had 2001] CREOLIZING THE ROMAN PROVINCES 215 must be seen in terms of a cycle of interpretation, OPPOSITION TO EMULATION in which the past was used to inform the present, Since Haverfield's day, it has been acknowledged and in the process of which the Roman past was that at the bottom end of society (and in some re- itself reinvented. Among those inventions (which gions more than others) Romanization was far less also included defensive imperialism62 and laissez- widespread than at the top. Only in the towns, Hav- faire religious tolerance63) was the paradigm of erfield argued, could Romanization be shown to Romanization as civilization. have extended throughout several social strata, from Whatever we make of this debate, we can surely elites to the lower social levels.67 Elsewhere, civili- agree that Romanization was the product of a spe- zation took a firm hold only among the aristocracy. cific period in European history. Equally, the two The rustic poor, "covered with a superimposed lay- nativist reactions to it reflect other moments and er of Roman civilization,"68 comprised the least Ro- histories-the decolonization of North Africa and manized stratum. Efforts to explain this phenome- the new wave of critical post-colonial scholarship non, more than any other, highlight the weakness- this brought, and the backlash in some European es of Romanization as a model for cultural change. quarters against neoimperialism at the time of the The explanations put forward do not address local- Vietnam War. As Woolf has recently emphasized, ized choices or the development of countercultur- modern accounts of Roman culture have moved a al movements from the bottom up (a possibility dis- long way from the absolutist position represented cussed below). Instead, it is assumed that all levels by Haverfield.64 Indeed, much post-Haverfield of society must have desired the symbols of Roman- scholarship has sought reasons-beyond cultural ization, as did the elites, despite the fact that some superiority-as to why imitation might take place, social groups, such as the rural poor, had little of and has concluded that Roman symbols were not practical value to gain from them.69 This is not to slavishly copied for their own sake but were emulat- deny the existence of aspiring elites, whose role in ed for both aesthetic and pragmatic reasons. the uptake of Roman lifeways was no doubt particu- Millett's reworking of Haverfield,65 whereby elites larly important: it is simply to suggest that at some adopt Roman symbols as a means of retaining pow- point on the social scale, such aspirations would er and status, introduced the concept that emula- only be realized (or even entertained) in excep- tion has at least as much to do with the pragmatics tional circumstances. Exactly where that point lies of power as with the recognition of superior cultur- is an important question, but one little articulated al values. Though he does not explore the extent in Romanization studies. to which this form of pragmatic emulation was min- Haverfield also accepted that some features of gled with a recognition of superior cultural values, provincial life (such as religion) involved an inter- Millett is thus able to suggest practical reasons why action between Roman culture and surviving na- elites would emulate the symbols of Romanitas. Two tive traditions.7" As we have already seen, he regard- observations may be made at this point. First, we ed these survivals as evidence for the "latent persis- only have to think about the Druids to realize not tence"7' of indigenous superstitions. The passivity all elites shared that reasoning,66 and second, what implicit in the choice of the term latent is reveal- about provincial nonelites? It is when we address ing. It suggests that, however briefly expounded this issue that both the weaknesses of Romaniza- Haverfield's views on the Romanization of nonelites tion as a model for culture change, and the con- were, he did not consider the factors inhibiting the tinuing dependency on emulation as the motor for rustic poor from acquiring Roman material culture change, become most apparent. as having anything to do with choice. Indeed, Hav-

t2 See Harris1979 and, for a recent case studyhighlighting by Rome (Webster1999). the continuing repercussionsof this theme, see also de Souza 7Haverfield 1923, 16. 1996. 68Haverfield1923, 79. :3 See Webster 1997a, 1997b. 69Differential access to the opportunitiesoffered by Rome '4Woolf1998,5. The similarviewpoints of other keyfigures is discussedby Drinkwaterand Vertet (1992, 25-8). 70 in the developmentof Romanprovincial archaeology, includ- Haverfieldemploys the term "coalescence"in describing ingMommsen (1885) in GermanyandJullian(1908-1926) in this interaction (1923, 21). France,have also been superseded. 71On latent persistencesee Haverfield(1923, 22), where a " Millett 1990a, 1990b. contrastis drawnbetween the passivepersistence of indige- 'i I have discussedelsewhere the spectacularfailure of the nous culturaltraits and the concept of activeopposition. Druidicelite to embrace the "communityof interest"offered 216 JANE WEBSTER [AJA 105 erfield drew an explicit contrast between latent it at least suggests that without a push from the top persistence and the idea of active opposition. Such (whether Roman or native), Romanization was not concepts as resistance, the development of coun- the unstoppable cultural force many assume it to tercultures, or the birth of creole cultures simply have been. On the other hand, this reasoning goes did not exist in Haverfield's reading of the passive even further than earlier models in denying non- peasantry of Roman Britain. elites any say in their own cultural development. In The fact that Haverfield, writing at the turn of the this scenario of elite indifference, nonelites sim- century, regarded provincial elites as the only active ply become passive receptors of those random ele- forces in the history and culture in the Roman west, ments of Roman culture that trickle down to them. is perhaps unsurprising. What is more surprising is The unstated assumption here, again, is that these the persistence of this trend into the 1990s. Millett's Roman influences were always welcomed. model-while representing a significant advance in THE CURRENT STATE OF ROMANIZATION attempting to determine the processes by which STUDIES elites adopted Romanized material culture-has even less to say on the mechanisms by which Roman- It can be seen that at the end of almost a century of ization spread to nonelites than Haverfield's did. Romanization studies, emulation, assumed to be a Millett's contention, as we have seen, is that native spontaneous or self-generating process,77 remains the elites emulated Roman symbols to reinforce their dominant model of native engagement with Roman identification with Rome. Following this: "Progressive material culture at all levels of society. But what drove emulation of this symbolism further down the social this emulation? Elites and urban dwellers may well hierarchy was self-generating, encouraging others in have been motivated by pragmatic self-interest, as society to aspire to things Roman, thereby spreading Millett has persuasively argued, but what did the ru- the culture."72 Thus, nonelites were Romanized at ral poor have to gain by adopting the symbols of Ro- second hand, emulating the material culture of their manitas? Without an element of self interest, the only social superiors, who had set themselves apart from motor for change becomes the superiority of Roman the lower orders through the acquisition of Roman culture, bringing us full circle to Haverfield's origi- status indicators. Millett appears to accept-though nal (acculturative) conception of Romanization as the this is not discussed in any detail-that investment unstoppable march of civilization. in the symbolic capital of Roman material culture Equally importantly, continued faith in Haver- would have rapidly become the only way in which field's notion that anyone who could have Roman- aspiring elites, and those even lower down the social ized would have when given the chance means that hierarchy, would have been able to enhance their the failure to emulate must be explained, wherev- own prospects or express their identity in the new er encountered. Unfortunately, the point where the order. The existence of other currencies (that is, of trickle-down effect tapered off is never examined countercultural symbols of identity and status) is not in terms of localized choices. It is instead explained envisaged. in four ways: latent persistence of folk customs (Hav- Some recent writing, building on Millett, has even erfield/Collingwood); overt resistance (the north suggested that if we accept that Romanization was African nativist model); Romanization as a veneer the outcome of shared interest between native elites (the British nativist model); and pre-conquest re- and the metropolitan center, it also follows that gional differences (the 1990s approach typified by Rome would have shown no interest in extending Millett and Haselgrove). Roman culture to the poor. As a result rural non- Finally, but most importantly, the only alternatives elites experienced Rome entirely through the to wholesale Romanization suggested to date are mediation of Romanized elites.73 In such circum- negative ones: not to Romanize at all (nativist ap- stances, it is suggested, it is only to be expected proaches), or elite indifference. All models of Ro- that cultural assimilation in rural contexts will have manization thus lead us to the same place: a polar- happened more or less haphazardly.74 On the one ized provincial world of Romans (or Romanized hand, this argument represents an advance, in that natives) and natives,76 with no gray areas in between.

72Millett 1990b, 38; the emphasis is my own. 75"Spontaneous" is Haverfield'sterm for it (1923, 14); "self- 73Millett(1990b) implies this,Whittaker (1997, 155) states generating"is Millett's (1990b, 38). it explicitly. 76Woolf usefully terms this a dichotomy between Roman- 74Or as Whittakerputs it (1997, 155) "byosmosis." ization and resistance(1998, 15). 2001] CREOLIZING THE ROMAN PROVINCES 217

Woolf recently suggested that revisionist studies of has been reiterated by Woolf, who in his recent Romanization (such as this one) perpetuate that study of refines this position by stress- dichotomy by focusing on imperialism and Roman- ing that part of the reason why Roman culture was ization in terms of the interaction of Roman and widely adopted by provincial elites and would-be native cultures.77 It is my contention that only by elites lay in the fact that "those Roman aristocrats revising some of our cherished attitudes can we who had taken on themselves the burden of regu- move on from this "either/or" approach to provin- lating civilization had defined Roman culture in cial societies in order to understand the processes such a way that it might function as a marker of informing the emergence of new societies in the status, not of political or ethnic identity. In this way, Roman provinces. 'becoming Roman' could again be a strategy, and Perhaps the greatest problem with the polarized one strategy among others."79 view of Roman and native culture is that there can Thus, there is some recognition today that the use be no ambiguity whatsoever about what Roman or of Samian ware did not necessarily mean the user Roman-style material culture means: where we find aspired to the full gamut of Roman values (any more it, we do not find people making active use of it in than the millions around the globe who drink Coca- the negotiation of new identities-we find Romans Cola would necessarily view American cultural val- and aspiring Romans. Where we do not find Ro- ues as superior). But it must be emphasized again man-style material culture, or find less of it, we are that all these studies have focused on elites. Rome in the company of natives. (Such was the case in opened up windows of opportunity for elites and northern Britain, which according to the conven- aspiring elites that, at a certain point down the social tional reading failed to Romanize because pre-Ro- scale, simply disappeared (or never existed at all). man Romanization had not taken place. On the Equally importantly, none of these models allow other hand, pre-Roman Romanization in the south for adaptive synthesis, in which Romanized materi- fostered the community of interest upon which suc- al culture could be used in ambiguous ways, simul- cessful Romanization of indigenous elites depend- taneously creating new identities and maintaining ed.) Again, this polarized view of Roman and na- key aspects of pre-Roman belief and practice. It is tive material culture can be traced back to Haver- time to replace Romanization with a new framework field, who explicitly ruled out the possibility that that does address these issues. That framework is provincials, as they became acculturated, could re- creolization. gard Roman material culture with any degree of FROM ROMANIZATION TO CREOLIZATION ambiguity.78 For Haverfield, to Romanize was to be- come Roman, and to use Roman material culture Creolization is a linguistic term indicating the was to espouse all that Rome stood for. Theories merging of two languages into a blended dialect.80 have become more sophisticated since Haverfield, It has come to be used more generally for the pro- but, again, not enough. The British nativist model cesses of multicultural adjustment through and that of Millett both suggested that tactical, po- which-from the interaction between Europeans, litically-motivated ways of utilizing Roman culture Native Americans, and displaced Africans-African- may have existed. The former, however, supposed American and African Caribbean societies were cre- that use of Roman culture did not actually change ated.81 It is a term increasingly employed in colo- Celtic society-beneath a Roman veneer lay an un- nial archaeology in the Americas, and one that sullied pure Celticity-and the latter supposed that could usefully be brought to bear on the Roman these strategies were played out simply in terms of provinces. gaining or maintaining a particular social standing The development of creole languages is instruc- in the new Roman order. tive for those attempting to understand the negoti- Millett's reworking of Haverfield accepted that ations that shaped colonial societies. Abrahams ar- elities made a strategicuse of the symbols of Roman- gues that in the New World colonies, West African itas in order to advance their position. This point slaves had to substitute some of their own words

77Woolf 1997, 340. ceased to bear nationalhatred them"(1923, 78 any against 20). Haverfield'sviews here are quite explicit: "When [the 79Woolf1998, 239. 80 Romanprovincial] adopted, and adoptedpermanently, the use On creole dialectssee Abrahams1983. of thingsRoman, we maysay of him, firstly,that he had become 81The earliestand most influentialstudy of creole cultural civilizedenough to realizetheir value, and further,that he had developmentwas by Braithwaite(1971). 218 JANE WEBSTER [AJA 105 with English words because they did not have the (artifacts, icons, architecture) are viewed, like words equivalent terms (especially for the slave-master in a dictionary, as building-blocks in the "lexicon of relationship itself). But substitution of English for culture."87 In creole contexts, Ferguson argues, the African dialects was not the norm. Instead, their "grammar" informing the use of what at first sight language became a mixture of African and English appears to be an "acculturated" material culture elements, and English, where it was used alone, may be radically different from the material cul- was employed in a limited range of contexts: "The ture itself. Slaves, for example, typically received language of everyday discourse remained essentially their material culture from their European mas- creole. Only on certain ceremonial occasions was ters. Yet Ferguson's work demonstrates clearly that the European tongue called into play, and then these artifacts were frequently utilized according only the most formal style of that speech system."82 to sets of values that were principally African. Thus, Archaeologists working in the Americas have for while the artifacts of slaves might appear to be Eu- some time been extending the model of creole lan- ropean American, they were created and used ac- guage development to the study of material culture,83 cording to the underlying, African-derived rules of suggesting that, like creole languages, creole mate- slave culture. This dialectic, in turn, contributed to rial culture represents not the gradual replacement the emergence of a new, African American society. of one way of life by another, but the blending of Three points emerge here. First, just as it would both, in a clearly nonegalitarian social context. Like be impossible to understand what it meant to be a language, creole material culture exhibits a degree person of color in colonial America without exam- of mastery of two cultural traditions, which can be ining creole processes, we cannot understand what drawn on to differing degrees according to context. it meant to be a provincial subject of Rome without The result is a highly ambiguous material culture, focusing on similar processes of negotiation in the in the sense that it is imbued with different mean- Roman world. Second, in any colonial context, ings in different contexts. Anne Yentsch's attempts those processes are given material expression. In to identify the experiences of domestic slaves in one this sense, material culture encapsulates colonial elite household in Annapolis, Maryland, provides a experience. Third, while creole culture is an amal- good example here.84 Focusing on cookery, she ar- gam of different traits, creolization processes take gues that because slaves were expected to provide place in the context of asymmetric power relations. some of their own food (grown on allotments), Afri- Links with the past are maintained in opposition to can foodstuffs and ways of cooking were thus main- the elite-sponsored trajectories of a dominant cul- tained. They became, in turn, staple elements of ture, and maintaining tradition thus carries with it African American creole cuisine. Yentsch empha- an element of risk. Creolization is frequently, there- sizes that what resulted was not simply an African fore, a process of resistant adaptation. What emerg- nativist survival; rather, slaves negotiated a way of liv- es from this process is not a single, normative colo- ing that accommodated old and new, and out of this nial culture, but mixed cultures.88 a new ethnicity was created. Given its origins, how can creolization best be ex- Leland Ferguson's work on Carolina plantation ploited in studying the Roman world? Ferguson slavery85 represents perhaps the most detailed and emphasizes that creolization theory allows for wide- successful application of creolization theory to ar- ranging analyses of social and political interaction: chaeology to date. Building on the work of the his- "Since the emphasis of such studies is on the pro- torian and folklorist Charles Joyner,86 who applied cess of creolization, cultures need not be commonly the creolization concept to the lifeways of 19th-cen- known as 'creole cultures' for an analysis of creoliza- tury slaves, Ferguson has developed an approach tion to be applicable."89 I suggest that in the same to colonial material culture in which material things way that European artifacts could be used by slaves-

2Abrahams 1983, 26. 374 n. 4). "8Theparallels between the creolization of language and 8Yentsch 1994. material culture are drawn out explicitly by Yentsch in her 85Ferguson1992. studyof the materialculture of domestic slavesin Maryland: 8Joyner 1974. "Take,e.g., the process wherebyAfrican-born women living 7 Ferguson 1992, xlii. in the Chesapeakeadapted elements of English cuisine with 8 Ferguson (1992, xlii) envisages "a series of interacting Africancooking techniques and foodstuffs.What Abrahams subculturesrather than a single creolizedblend." [supran. 80] wroteaboutverbal performance in the Caribbe- 89Ferguson1992, xli. an points the wayto what happens with food" (Yentsch1994, 2001] CREOLIZING THE ROMAN PROVINCES 219 not because they aspired to become European but use of media apparently only rarely used in the west according to an underlying set of non-European before the conquest (sculpture in stone or bronze), rules-provincial artifacts in the Roman world may and the deities themselves are depicted in human likewise appear Romanized, but can in certain con- form. It has long been argued that the Iron Age Celts texts likewise operate according to a different, in- had a reluctance to construct anthropomorphic im- digenous, set of underlying rules. As creole artifacts, ages of gods, and it has therefore been assumed that they can negotiate with, resist, or adapt Roman styles after the conquest Celtic peoples adopted from to serve indigenous ends, and ultimately, they are Rome new gods and a new repertoire of anthropo- part of the emergence of creole societies. morphic representation, used to depict both Roman and . Anthropomorphic imagery is thus CREOLIZING ROMANO-CELTIC RELIGION argued to mark both a classically-inspired Roman- Religion has always had a central place in the ization of the representation of the Celtic gods, and, study of creolization processes. This is because, in through the process of (itself many colonial contexts, religious belief has either seen as a manifestation of provincial Romanization), been the focal point around which overt rebellion the welcoming of those deities into an accommodat- has crystallized,9' or it has been the aspect of indig- ing Graeco-Roman cosmology. As a result, Romano- enous cultural life most resistant to acculturation. Celtic religion has for many decades been interpret- For example, Santeria, the creole religion of Cuba, ed as a neutral syncretism that was not imposed upon developed as the beliefs and practices of diverse the provinces from outside, but reflected the spon- peoples (European Catholics, Christianized indig- taneous desire of polytheistic peoples to accommo- enous elites, American Indians, and Africans) jos- date each other's gods.92 This position has been tled together, creating a creole synthesis. As we modified somewhat,93 but the majority of work still examine Santeria, we need to bear in mind the fact lacks an acknowledgment that Romano-Celtic reli- that the ethnic make-up of Roman Britain was as gious and artistic syncretism took place within a co- complex in its own way as that of the Spanish colo- lonial context. nies. In the Roman west, indigenous beliefs en- While conventional models of Romanization have countered a Classical pantheon that was itself the no explanatory power regarding the emergence of product of centuries of religious interchange be- Romano-Celtic religion, nativist arguments for a tween Roman, Greek, and Eastern religions. Add- Roman veneer overlying an unsullied Celtic reli- ed to this, many of the Romans in the west (for ex- gion are of little help either. To regard Romano- ample, auxiliary soldiers) came from territories that Celtic religion-or its visual expression-as Celtic had also been conquered by Rome, and whose be- religion expressed in non-Celtic ways is not simply lief systems had already engaged with the Roman an error; it is a failure to recognize the emergence pantheon. Britons and , encountering these not of a problem-free syncretism, but of a creole already complex gods of Rome, brought other religion. While we hold on to the idea of an emula- strands of belief and practice into this arena. In tive Romano-Celtic iconography, we cannot begin Roman Britain and Gaul, as in the Spanish colo- to recognize the religion's true nature and under- nies, it seems reasonable to suggest that creole syn- stand provincial icons in terms of negotiation with theses may also have emerged. Roman beliefs and values. Needless to say, Romano-Celtic religion has never In order to explore the workings of creole reli- been conceptualized in these terms. The notion of gion we may turn to Santeria, the creole religion of Romanization has instead dominated the study of Cuba. Religion across the Caribbean today is a com- Romano-British religion, the main sources of evi- plex mixture of influences: American Indian sur- dence for which comprise inscriptions and iconog- vivals, Christianity (first Spanish Catholicism, later raphy.91 It is perhaps easy to see why this has hap- English Anglicanism), and also African (mainly pened. Post-conquest provincial iconography makes Yoruba) influences introduced by slaves. As Bran-

"0See, e.g., Webster (1999) on millenarianprotest move- Webster(1986) and Green'searliest studies of Celticreligion ments in post-conquestcontexts around the world, and the (1986, 1989). possibleexistence of thisform of overtresistance in earlypost- 93Green 1997 and 1998, on the genesis and nature of Celt- conquest Gauland Britain. ic and Romano-Celticimagery, are more concerned with the '1 I concentrate on the latterhere: on the former see Web- impact of culturalinteraction upon religiousart than wasthe ster 1995a, 1995b. case in her earlierwork. )9 Important studies advancing this viewpoint include G. 220 JANE WEBSTER [AJA105 don emphasizes in his study of Santeria,94 such reli- Creole religions like Santeria offer an important gions grow out of developing creole cultures as an lesson for students of Romano-Celtic religious syn- opposition between the ideals of indigenous elites cretism. They demonstrate that there are limits to syn- and the aspirations of nonelites. Colonial culture cretism, which represents an adaptation, rather than in Latin America and the Caribbean, he states, was an adoption, of new religious beliefs and practices. "participated in by most of the population regard- Syncretism is thus part of a process of intercultural less of ethnic or racial origin. Comprised of Span- negotiation, which scholars in the Americas call "re- ish, Native American, African, and Islamic influ- sistant adaptation" or "resistant accommodation."97 ences, it remained nevertheless oriented toward The iconography that emerges from such negotia- Spain and, as a result, was dominated proximally by tions is, as Cuban Santeria amply demonstrates, ex- the vision of Spanish culture current among the tremely complex, drawing some visual elements from island's hegemonic elite. This elite in its turn saw the metropolitan ideal, but conveying a very differ- the 'ideal,' the conquest culture promoted by the ent, countercultural visual message. Is it possible to union of the Crown and the church, as its model see similar processes at work in the iconography of and promoted it actively, selectively, and by exam- Romano-Celtic religion? One of the difficulties of ple to the rest of the population."95 The problem, studying Romano-Celtic iconography is that it is often as Brandon documents, was that the rest of the pop- very poorly contextualized and lacks the depth of in- ulation did not embrace the same ideal. Making formation available to scholars of recent historical pe- their own selections from Spanish Catholicism (the riods. Indeed, a significant proportion of the Romano- identity and iconography of the saints) and fusing Celtic data comprises stray finds, or poorly-contexted them with Yoruba deities, the urban poor, among material from 19th-century excavations. It is argued whom were large numbers of slaves and freed slaves below, nevertheless, that by comparing different cate- of African origin, created a new syncretistic reli- gories of iconographic representation and paying gion. This process took place over a period of sev- close attention to the degree to which different dei- eral centuries, centered on the years from 1760 to ties are associated with other new developments, such 1870. The visual expression of the resultant syn- as epigraphy, it becomes possible to isolate some di- thesis, Santeria, is highly idiosyncratic, making use vine figures that seem to be particularly resistant to of both Catholic imagery and Yoruba symbols. what (in conventional terms) would be seen as syn- Brandon's point above is that local elites (in the cretistic Romanization. I suggest below that , Spanish Caribbean) adopted an ideal to which the the horse goddess from eastern and central Gaul, is a Spanish had purposefully guided them, intending deity of precisely this type. to impose Spanish culture and values upon the New EPONA: A CREOLE GODDESS FROM ROMAN GAUL World colonies. MillettO`makes much the same point in his model of the Romanization of the Roman west, Eastern and central Gaul are areas particularly whereby a community of interest was fostered be- rich in post-conquest iconography. Three major tween native elites and Rome, ensuring that local groups of non-Classical deities occur: elites were the first to adopt the symbols of Romani- 1. Celtic deities paired or twinned with gods from tas. But at the root of Millett's model (as with Haver- the Graeco-Roman pantheon. The epigraphic in- field's) is the supposition that this colonial ideal trick- terpretatioRomana is frequently employed in these led down through society by a process of spontane- cases (e.g., Apollo Moritasgus), explicitly linking a ous emulation. Brandon's point is that different ide- Graeco-Roman and an indigenous deity, or identi- als and practices can emerge, as it were, from the fying one with the other. Another manifestation of bottom up, creating counter- or subcultures. It goes deity pairing is the divine marriage of a Graeco- without saying that the social contexts in colonial Roman male deity with a female Celtic partner. The Cuba and the Roman west were radically different, best-known example of this pairing is the divine but the general point is that what is missing from marriage of Mercury and .9 Millett's model, and from all acculturative models of 2. A repertoire of clay statuettes, mass produced in Romanization, is the possibility for this bottom-up the Allier region throughout the second century A.D. cultural development to take place. One of the most popular deities of this group was the

'4Brandon 1993. colonial Peru (Stern 1982, 11). The latteris discussedby Gar- ' Brandon 1993, 43. man (1998). ' Millett 1990a, 1990b. " Green 1989, 45-73; Webster 1997a. t7 The former term was employed by Stern in his study of 2001] CREOLIZING THE ROMAN PROVINCES 221 pipe-clay "Venus," clearly based on the Venus Pudica she is frequently associated with a horse and a foal. iconographic type. I have argued elsewhere"' that Indeed, the goddess's name apparently derives despite the Romanized appearance of Venus and oth- from the Gallic word for horse,1'4 and her identity is er deities in this category, it is possible to regard these dependent upon her horse emblem; she does not clay statuettes as a pantheon of popular culture, re- exist without it. Conversely, horse and foal imagery flecting indigenous fertility beliefs. does sometimes occur without Epona herself.1"" 3. A series of deities, which although depicted This zoomorphic trend is less tangible for Sucel- anthropomorphically, are clearly neither Graeco- lus, but there are occasions on which he is associat- Roman gods nor based upon classical exemplars. ed with dogs.""' It is very possible that the represen- The horse goddess Epona, the horned god Cer- tation of semi-zoomorphic deities in the post-con- nunnos, and the hammer god Sucellus, who wears quest period may be a continuation of pre-Roman a distinctive sagum (Gallic cloak), are the principal traditions, and this, I suggest, is one of the keys to deities of this group. understanding the transformation of the group 3 Here, I will principally be concerned with the deities in the post-conquest period. Drawing on deities in group 3, and above all with Epona. The Roman artistic traditions of naturalistic human im- iconography of this goddess is more widespread agery, pre-conquest zoomorphic deities began to than that of any other Celtic deity in Europe, and be either consistently represented in human form she is the one deity of this group for whom some (as in the case of and perhaps Epo- resistance to the Graeco-Roman pantheon has al- na),"'7 or realized for the first time in a way that ready been suggested in detail.1"' Epona's distribu- incorporated adoption of the human form (as may tion is concentrated in northern and eastern Gaul, be the case for Sucellus and perhaps also Epona). with notable clusters among the and Mandu- The large body of evidence for horse imagery in bii."" She was also popular with Gallic auxiliaries Iron Age contexts, together with Epona's insepa- on the border of Germania Superior, many of whom rability from the horse and her near-absence from are likely to have come from the Aedui. Dating of the most Romanized parts of Gaul (Narbonensis in these images is difficult, and usually performed on particular) has led to the claim that the goddess is stylistic criteria alone, but the majority of attempts "a Celt among the Romans.""'s I see this as a mis- at dating place the images in the second and third take, because to conceptualize Epona in these terms centuries A.D. Images of Epona extend from Brit- is to regard her rather in the way that the nativist ain to Dacia, and writing in the second century, school of thinking on Romanization regards pro- Apuleius mentions figures of the goddess on sta- vincial culture in general: as pure Celticity under a bles in Rome itself.1"2Her cult therefore appears to Roman veneer. In this sense, I completely agree have been long-lived and very widespread. with Henig's assertion"'"' that it is fallacious to as- It is important to begin by observing that, with sume that beneath a very thin veneer the Celtic the exception of Cernunnos,"'3 it is not certain that gods remained uncorrupted by Rome. For Henig, the deities in group 3 were represented in human however, this was because the Celtic gods, like the form before the conquest. Indeed, it is probable Celts themselves, had become Romans."" This I that they were not. I suggest this because one of the cannot accept. In my terms, Epona cannot be re- most obvious post-conquest characteristics of this garded as a Celtic deity or a Roman one. She is the group is the presence of animal imagery. Cernun- product of the post-Roman negotiation between nos is depicted as a semi-zoomorphic horned be- Roman and indigenous beliefs and iconographic ing, and Epona is represented in human form, but traditions: she is thus a Romano-Celtic deity but is consistently associated with one or more horses. not, as discussed below, the product of a problem- In Burgundy, where her cult probably originated, free or spontaneous synthesis. Creole deities en-

'"Webster1997b, following Vertet 1984. ""'Ona seriesof imagesfrom Nimes, e.g., the god is depict- ""By Linduff (1979). ed witha dog (Esp6randieunos. 434,436-7) or witha dog and ""Oaks 1986. a coiled snake (Esperandieuno. 435). 1)2 Met. 3.27. 1"7Green (1986, 171-2) suggeststhat Epona'spre-Roman "':Thehorned deity or deities grouped as Cernunnoscer- existence may be reflected in the frequent horse imageryof tainlyhave pre-conquestorigins, occurring on petroglyphsat the later Iron Age. Val Camonica,north Italy,dating from the seventh to fourth "'8Thisterm is used in the title of a paperby Linduff (1979). centuries B.C. (Green 1986, 193). "19Henig 1984, 43. ""As Green (1986, 173) notes. "" 1984, 93. 1"0 Henig See, e.g., Esp6randieunos. 2046, 2121. 222 JANE WEBSTER [AJA105 capsulate the limits to syncretism, and this is partic- tification of Graeco-Roman gods with deities who ularly true of Epona. have Celtic names but localized distributions. It is This point may be demonstrated by examining very interesting, therefore, that the group 3 gods, what is absent from Epona's imagery rather than who were widely worshipped in the Roman west, what is present. First, we may note that although were not paired with Graeco-Roman deities in this many of the stone images of Epona are the product way. As I have argued elsewhere,"' the practice of of competent artisanship, and although the god- epigraphic name-pairing was one enthusiastically dess is frequently depicted with classical at- embraced by high-ranking Roman officials and in- tributes,1" there are no instances in which she is digenous elites, but was less enthusiastically adopt- paired iconographically with a deity from the Grae- ed by other social groups. Nonelites may have re- co-Roman pantheon. Given the frequency with sisted the epigraphic interpretatioromana not because which other Gallic goddesses enter into divine they were ignorant, but because they were unwill- marriages with male Graeco-Roman partners,"2 ing to take syncretism this far: while the adoption Epona's resistance to the married state seems im- of the human form may have been acceptable, nam- portant. It suggests there are levels on which she ing perhaps was considered objectionable. operates beyond the Graeco-Roman pantheon; that It may of course be counterargued that the dis- despite her emergence into human form, there is tinctions made here are simply a matter of econom- a limit beyond which she does not syncretize or ics; Latin was only available to elites, and thus those Romanize. Cernunnos is similarly resistant to icon- lower down the social scale would not employ it."8 ographic twinning."3 It is also interesting to note, Some of the imagery I have discussed, however, is in this context, that the hammer god Sucellus is the product of good quality, and therefore expen- sometimes associated with a divine partner, but she sive, artisanship,"9 and it is unlikely that the indi- is not a classical one: his partner has the Celtic name viduals who commissioned such icons could not .'14 have afforded to incorporate epigraphy if they so Second, we may note that images of Epona, Cer- chose. More fundamentally, to reduce that element nunnos, and Sucellus rarely incorporate epigraphy. of choice to a simple economic necessity is simply Again, there are some exceptions to this generali- to fall, once again, into the trap of assuming that zation,1' but these are far from common. Epona, in Romano-Celtic iconography is emulative of a met- particular, lacks widespread epigraphic attestation. ropolitan ideal. We should by now be wary of assum- In his 1948 study of Celtic religion, de Vries noted ing that every provincial subject who could afford it a total of 26 inscriptions naming Epona."'l More would have epigraphized his or her iconography. examples may well have been identified since the Drawing these various strands of evidence togeth- 1940s, but given the widespread nature of Epona's er, it is possible to suggest that the group 3 deities, cult (over 300 images of the goddess are known), Epona, Cernunnos, and Sucellus, are neither Celt- this seems a very small number of occurrences. In ic deities'20 nor Roman ones. In suggesting that this context, it is important to note that neither Epo- they are Romano-Celtic gods, however, I have tried na nor Cernunnos is epigraphically name-paired, to move beyond the related concepts of Romaniza- in any instance, with a deity from the Graeco-Ro- tion and syncretism to suggest that these are creole man pantheon. Throughout Gaul, as in Britain, deities who encapsulate the limits to syncretism in epigraphic name-pairing (-Mars; -Min- Roman Gaul, and who were not actually incorporat- erva etc.) was widely employed for the explicit iden- ed into the Graeco-Roman pantheon. They repre-

"''Most commonly cornucopiae,but also paterae,from which Paris (Green 1986, 195). she feeds her foal: see Green (1989, 92). "6de Vries 1948, 276. ''"The iconographyof the Romano-Celticdivine marriage "7Webster1995a, 1995b. is summarizedby Green (1989, 45-73). "8 This is certainlythe view of MartinHenig (1984, 59), 'The exception which provesthe rule here is a veryhigh who like myselfhas argued that language lay at the heart of qualityrelief from Reimson which the god is flankedby Mer- interpretatioromana (the explicit identification of Roman and cury (Green 1989, fig. 38), but it is interestinghere that the non-Romangods). Henig, however,sees this as simplya mat- compositionof the reliefleaves no room for ambiguityregard- ter of status,arguing that "in order to be able to 'interpret,' ing the principaldeity depicted: Cernunnos, a seated figure one must firstbe articulate"(ibid.). pouring out grain or possiblycoins, dominatesthe image. 119 See, e.g., the stone images from Meusaultand Mellecy 14 Green 1986, 95-6. (central Gaul) in Green 1989, figs. 6-7. 15 E.g., the name Cernunnos appears on a relief image 120AsLinduff (1979) has suggestedfor Epona, and Vertet of an antlered god that forms part ofTiberian Sailors'Pillar, (1984) appearsto implyfor the entire group. 2001] CREOLIZING THE ROMAN PROVINCES 223 sent an alternative creole pantheon in Roman Gaul ambiguities of both syncretism and the colonial and provide some of the best evidence pertaining condition itself. to the complex negotiations that characterize Ro- The depiction of Celtic gods in the western prov- mano-Celtic religion. inces was neither a simple emulation of metropol- Turning back to Santeria, an additional lesson itan art, nor-at the other extreme-a visual ex- may be drawn out with reference to the group 3 pression of nativist opposition to Rome. Rather, as Romano-Celtic deity, Cernunnos. In order to read it came into being by negotiating a spiritual path- Santeria icons successfully-to understand the way between acceptance and resistance, much post- tensions and power play inherent in colonial reli- conquest iconography represented a creole art gious syncretism-it is necessary to pay as much form. Unfortunately, Romanization of form (the use attention to the juxtaposition of symbols (in this of anthropomorphic imagery) has blinded us to case Christian and Yoruba) as to the symbols them- what these icons meant to the people who fash- selves. This is a point we should remember when ioned and used them. Creolization, as a way to mod- we try to interpret the presence of Celtic attributes el these complex processes, allows us to reevaluate on representations of deities from the Graeco- these images as the active material culture through Roman pantheon. This practice, like its inverse which new social identities were forged. (the giving of classical attributes to Celtic gods) is CONCLUSION conventionally seen as reflecting the problem-free syncretism of two deities. Reference to Santeria I have suggested above that, as the inadequacies may allow us to see greater complexities to this of Romanization as a model for contact and culture process than formerly envisaged. In this context, change in the Roman provinces become increas- Green has pointed to the tendency for some pre- ingly apparent, this acculturative model should be and post-conquest anthropomorphic imagery from discarded in favor of the concept of creolization, Gaul and Britain to cross species barriers, mixing which offers a new way to approach provincial mate- human and animal traits together.'2' The horned rial culture in all its forms. god Cernunnos is the most obvious example here. The development of an archaeology of creoliza- This deity has definite pre-conquest origins,'22 and tion in the New World certainly offers pointers as to it seems likely that the representation of semi-zoo- where Romanists should now be looking as we at- morphic deities in the post-conquest period is a tempt a more nuanced study of cultural negotia- continuation, or strengthening, of earlier western tion in Roman Britain. In the Americas, the emer- European Iron Age traditions. It is thus particu- gence of creole communities has always been stud- larly interesting to note that in Rome's western ied by focusing on everyday material culture, from provinces, some images of classical deities are also verandahs, to pots, to clay pipes, to recipes. In the given this semi-zoomorphic treatment.'23 For ex- Roman provinces, too, colonial experience for the ample, a few representations of Mercury from Ro- vast majority of people would have been acted out 24 man Britain appear to have horns in place of the through the materiality of domestic life. Perhaps it god's more usual winged hair or petasos. As Green is time, then, to shift the study of intercultural con- observes, it is sometimes difficult for the modern tact in Roman Britain away from elites, on whom so observer to determine whether horns or wings much work has focused, to other social categories; were intended, but as she points out, this may rep- the urban poor, the rural poor, and that technically resent for Romano-Celtic observers a conscious most invisible of social groups in the ancient world, flexibility or ambiguity, allowing the spectator to the enslaved. see what it was appropriate for him or her to see; the classical god or a horned indigenous deity.25 SCHOOL OF ARCHAEOLOGICAL STUDIES Green herself126 places this ambiguity in the con- UNIVERSITY OF LEICESTER text of La Tene artistic traditions and Celtic con- LEICESTER LE1 7RH cepts of liminality and polarity. Perhaps it is also ENGLAND possible that this type of flexibility reflects the [email protected]

'~2Green 1997. (Buckinghamshire)(Green 1986, fig. 98) and another from 122See supra n. 103. Uley (Gloucestershire). 2' ''Green (1986,222-3) has also remarkedon thisphenom- Green 1986, 223; 1997, 906. enon. 126Green 1997, 906. 124Examples include a stone relief from a well at Emberton 224 JANE WEBSTER [AJA 105 Works Cited nal of Historical Archaeology 2.2:133-58. Goudineau, C. 1979. Lesfouilles de la Maison au Dauphin. Abrahams, R. 1983. The Man-of-Words in the West Indies: Recherchessur la romanisation de Vaison-la-Romaine. Gal- Performance and the Emergence of Creole Culture. Balti- lia Suppl. 37. Paris: Editions du Centre national de la more:John Hopkins University Press. recherche scientifique. Barrett,J. 1997a. "Romanization: A Critical Comment." Green, M. 1986. The Gods of the Celts. Gloucestershire: In Dialogues in Roman Imperialism, edited by DJ.M. Alan Sutton. Mattingly, 51-64.JRA Suppl. 23. Portsmouth:Journal . 1989. Symbol and Image in Celtic Religious Art. Lon- of Roman Archaeology. don: Routledge. 1997b. "Theorizing Roman Archaeology." In . 1997. "Images in Opposition: Polarity, Ambiva- TRAC 1996: Proceedings of the Sixth Annual Theoretical lence and Liminality in Cult Representation." Antiq- Roman Archaeology Conference Sheffield, edited by K. uity71:898-911. Meadows, C. Lemke, andJ. Heron, 1-7. Oxford: Ox- . 1998. "God in Man's Image: Thoughts on the bow Books. Genesis and Affiliations of some Romano-British Cult Beard, M.,J. North, and S. Price. 1998. Religionsof Rome. Imagery." 29:17-30. Vol. 1, A History. Cambridge: Cambridge University Hanson, W. 1994. "Dealing with Barbarians: The Ro- Press. manization of Britain." In Building on thePast, edited Benabou, M. 1976a. La Resistance Africaine a la Romani- by B. Vyner, 149-63. London: The Royal Archaeolog- sation. Paris: Maspero. ical Institute. . 1976b. "Resistance et romanisation en Afrique Harris, W.V. 1979. War and Imperialism in Republican Rome du Nord sous le haut-Empire." In Assimilationet Resis- 327-70 BC. Oxford: Clarendon Press. tance a la Culture Greco-Romaine dans le Monde Ancien, Haselgrove, C. 1984. "Romanization before the Con- edited by D.M. Pippidi, 367-75. Paris and Bucharest: quest: Gaulish Precedents and British Consequenc- Les Belles Lettres and Editura Academiei. es." In Military and Civilian in Roman Britain, edited Braithwaite, E.K. 1971. The Development of CreoleSociety in by T.C.F. Blagg and A.C. King, 1-64. BAR-IS130. Ox- Jamaica: 1770-1820. Oxford: Clarendon Press. ford: Tempus. Brandon, G. 1993. Santeria from Africa to the New World: . 1987. "Culture Process on the Periphery: Belgic The Dead Sell Memories. Bloomington and Indianapo- Gaul and Rome during the Late Republic and Early lis: Indiana University Press. Empire." In Centre and Periphery in the Ancient World, Brunt, P.A. 1976. "The Romanization of the Local Rul- edited by M. Rowlands, M. Larsen, and K. Kristian- ing Classes in the Roman Empire." In Assimilationet sen, 104-24. Cambridge: Cambridge University Press. Resistance a la Culture Greco-Romaine dans la Monde An- . 1990. "The Romanization of Belgic Gaul: Some cien, edited by D.M. Pippidi, 161-73. Paris and Bu- Archaeological Perspectives." In The Early Roman charest: Les Belles Lettres and Editura Academiei. Empirein the West,edited by T.C.F. Blagg and M. Mil- Clavel-Leveque, M. 1989. Puzzle Gaulois: les Gaules en lett, 45-71. Oxford: Oxbow Books. Memoire, Images, Textes, Histoire. Annales Litteraires de Haverfield, F. 1905-1906. "The Romanization of Roman l'Universite de Besancon 396. Paris: Universitc de Britain."ProcBritAc 1905-6:185-217. Besancon. . 1923. The Romanization of Roman Britain. 4th ed. Collingwood, R.G. 1932. RomanBritain. Oxford: Claren- Oxford: Clarendon Press. don Press. Henig, M. 1984. Religion in Roman Britain. London: Bats- de Coulanges, N.D.F. 1891. Histoire des institutions poli- ford. tiques de l'ancienne France. Paris: Hachette. . 1995. The Art of Roman Britain. London: Bats- Drinkwater, J., and H. Vertet. 1992. "'Opportunity' or ford. 'Opposition' in Roman Gaul?" In CurrentResearch on Hingley, R. 1989. Rural Settlement in Roman Britain. Lon- the Romanization of the Western Provinces, edited by M. don: Seaby. Wood and F. Queiroga, 25-8. BAR-IS575. Oxford: . 1995. "Britannia, Origin Myths and the British Tempus. Empire." In TRAC 1994 Proceedings of the Fourth Theo- Esperandieu, E. 1907-1966. Recueil general des bas-reliefs retical Roman Archaeology ConferenceDurham 1994, edit- de la Gaule romaine et pre-romaine. Paris: Ernest Leroux. ed by S. Cottam, D. Dungworth, S. Scott, andJ. Tay- Ferguson, L. 1992. Uncommon Ground: Archaeology and lor, 11-23. Oxford: Oxbow Books. Early African America, 1650-1800. Washington, D.C.: . 1996. "The Legacy of Rome: The Rise, Decline Smithsonian Institution Press. and Fall of the Theory of Romanization." In Roman Forcey, C. 1997. "Technologies of Power in Roman Brit- Imperialism, Post-Colonial Perspectives, edited byJ. Web- ain." In TRAC 1996: Proceedings of the Sixth Annual ster and N. Cooper, 35-48. Leicester Archaeology Theoretical Roman Archaeology Conference Sheffield 1996, Monographs 3. Leicester: School of Archaeological edited by K. Meadows, C. Lemke, andJ. Heron, 15- Studies. 21. Oxford: Oxbow Books. . 1997. "Resistance and Domination in Roman Freeman, P. 1996. "British Imperialism and the Roman Britain." In Dialogues in Roman Imperialism, edited by Empire." In Roman Imperialism, Post-Colonial Perspec- D.J.M. Mattingly, 81-100.JRA Suppl. 23. Portsmouth: tives, edited byJ. Webster and N. Cooper, 19-34. Le- Journal of Roman Archaeology. icester Archaeology Monographs 3. Leicester: School Jones, S. 1997. The Archaeology of Ethnicity: Identities in the of Archaeological Studies. Past and Present. London: Routledge. Garman, J.C. 1988. "Rethinking 'Resistant Accommo- Joyner, C. 1974. Down by the Riverside: A South Carolina dation': Towards an Archaeology of African-Ameri- Slave Community.Urbana: University of Illinois Press. can Lives in Southern New England." InternationalJour- Jullian, C. 1908-1926. Histoirede la Gaule.Paris: Hachette. 2001] CREOLIZING THE ROMAN PROVINCES 225

Laroui, A. 1970 [1977]. L'histoire du Maghreb, un Essay du ter Archaeology Monographs 3. Leicester: School of Synthese.(Trans. R. Manheim). Princeton: Princeton Archaeological Studies. University Press. Stern, SJ. 1982. Peru's Indian Peoples and the Challenge of Linduff, K. 1979. "Epona, A Celt among the Romans." Spanish Conquest.Wisconsin: University of Wisconsin Collection Latomus 38 (4):817-37. Press. Mattingly, D.J.M. 1996. "From One Colonialism to An- de Vries,J. 1948. "Le Religion des Celtes." In LesReligions other: Imperialism and the Maghreb." In RomanImpe- de l'Europe Ancienne III. Paris: Presses Universitaire de rialism, Post-Colonial Perspectives, edited by J. Webster France. and N. Cooper, 49-69. Leicester Archaeology Mono- Vertet, H. 1984. "Religion populaire et rapport au pou- graphs 3. Leicester: School of Archaeological Studies. voir d'apres les statuettes en argile sous l'empire ro- , ed. 1997. Dialogues in Roman Imperialism.JRA Sup- main." In Archeologie et rapports sociaux en Gaule, edited pl. 23. Portsmouth: Journal of Roman Archaeology. by A. Daubigney, 72-122. Paris: CNRS. Millett, M. 1990a. The Romanization of Britain. Cambridge: Webster, G. 1986. The British Celts and their Gods under Cambridge University Press. Rome.London: Batsford. . 1990b. "Romanization: Historical Issues and Webster,J. 1995a. "Interpretatio:Roman Word Power and Archaeological Interpretations." In TheEarly Roman the Celtic Gods." Britannia 26:153-61. Empirein the West,edited by T. Blagg and M. Millett, . 1995b. "Translation and Subjection: Interpre- 35-41. Oxford: Oxbow Books. tatio and the Celtic Gods." In Different Iron Ages: Stud- Mommsen, T. 1885 [1968]. The Provinces of the Roman ies on the Iron Age in TemperateEurope, edited by J.D. Empire: The European Provinces (edited by T.R.S. Hill and C. Cumberpatch, 170-83. BAR-IS602. Ox- Broughton). Chicago: University of Chicago Press. ford: Tempus. Oaks, S. 1986. "The Goddess Epona: Concepts of Sover- . 1997a. "A Negotiated Syncretism: Readings on eignty in a Changing Landscape." In Pagan Godsand the Development of Romano-Celtic Religion." In Di- Shrines of the Roman Empire, edited by M. Henig and A. alogues in Roman Imperialism, edited by D.J.M. Mat- King, 77-83. Oxford University Committee for Ar- tingly, 165-84.JRA Suppl. 23. Portsmouth:Journal of chaeology Monograph 8. Oxford: Oxford University Roman Archaeology. Committee for Archaeology. . 1997b. "Necessary Comparisons: A Post-Colo- Reece, R. 1980. "Townand Country: The End of Roman nial Approach to Religious Syncretism in the Roman Britain." WorldArch12.1:77-91. Provinces." WorldArch28.3:324-38. . 1988. My Roman Britain. Cirencester: Cotswold . 1999. "At the End of the World: Druidic and Studies 3. other Revitalization Movements in Post-Conquest Scott, E. 1991. "Animal and Infant Burials in Romano- Gaul and Britain." Britannia 30:1-20. British Villas: A Revitalization Movement." In Sacred . Forthcoming. "Art as Resistance and Negotia- and Profane, edited by P. Garwood, D. Jennings, R. tion." In Provincial Art and Roman Imperialism, edited Skeates, andJ. Toms, 115-21. Oxford University Com- by S. Scott and J. Webster. Cambridge: Cambridge mittee for Archaeology Monograph 32. Oxford: Ox- University Press. ford University Committee for Archaeology. Webster, J., and N. Cooper, eds. 1996. Roman Imperial- Sellar, W.C., and R.J.Yeatman. 1975. 1066And All That. ism, Post-Colonial Perspectives. Leicester Archaeology London: Methuen. Monographs 3. Leicester: School of Archaeological Singleton, T. 1999 "An Introduction to African-Ameri- Studies. can Archaeology." In I, Too, Am America: Archaeological Whittaker, C.R. 1997. "Imperialism and Culture: The Studies of African-American Life, edited by T. Singleton, Roman Initiative." In Dialogues in Roman Imperialism, 1-17. Charlottesville: University Press of Virginia. edited by D.J.M. Mattingly, 143-63. JRA Suppl. 23. Smith,J.T. 1978. "Villasas a Key to Social Structure." In Portsmouth: Journal of Roman Archaeology. Studies in the Romano-British Villa, edited by M. Todd, Woolf, G. 1997. "Beyond Romans and Natives." World- 149-85. Leicester: University of Leicester Press. Arch28.3:339-50. Slofstra,J. 1983. "An Anthropological Approach to the 1998. Becoming Roman: The Origins of Provincial Study of the Romanization Process." In Roman and Civilization in Gaul. Cambridge: Cambridge Universi- Native in the Low Countries, edited by R. Brandt andJ. ty Press. Slofstra, 71-104. BAR-IS184. Oxford: Tempus. Yentsch, A. 1994. A Chesapeake Family and their Slaves. de Souza, P. 1996. "They are the Enemies of all Man- Cambridge: Cambridge University Press. kind: Justifying Roman Imperialism in the Late Re- Zanker, P. 1990. The Power of Images in the Age of Augustus. public." In Roman Imperialism, Post-Colonial Perspectives, Ann Arbor: University of Michigan Press. edited byJ. Webster and N. Cooper, 125-33. Leices-