Production Within and Beyond Imperial Boundaries : Goods , Exchange , and Power in Roman Europe

After Rome had conquered much of temperate Europe, the administration dir ected the establishment of industries important to the maintenance of militar y and economic control of the new provinces. These included stone quarries, pott ery manufa ctures, and metal industries. Recent research shows that much production wa<; not a<; centrali zed a<; ha<; been believed; diverse industrial sites throughout the provincial landscap es indicat e a variety of arrangements for supplying the needs of the empire. In many instanc es, Roman production system<; relied upon indigenous traditions of manu facturing. The provincial economics depended also upon materials collected and processed beyond the imperial frontiers. Analysis of Roman imports in German y, Scandina via, and ca<;tcrn Europe, and of the contexts in which they occur, suggest<; that goods produced outsid e of the empire played a major role in the imp erial economy. Th ese commercial links, over which Roman authorities had no effective control, contributed to substantial changes in economi cs and in social and political configura tions in societies beyond the Roman fronti er. INTRODUCTION Most of what we know, and think we know, about empires in world history concerns the actions, moti vations, and institutions of the imperial societies; we know relatively littl e about the peoples who arc drawn into imperial contexts, through conquest or other means. Most empir es [Page l] Journa I of World-Systems Research have been developed by societies that possessed wr iting, and the written accounts upon which understanding ha<; been ba<;cd by Romans in the Greater Mediterran ean world, Spanish in South America, British in lndia have been principally by writers who belonged to the imperial society and who represented the perspective of elite members of that society. 1n the ca<;c of early empires, where archaeology plays a major role in our understanding, the material remains of the imperial culture arc usually more substantial and more apparent than arc those of the societies impacted. Only relatively recently have historians, archaeologists, and others begun to make systematic attempts to understand the experience of the other peoples involved in interaction with empires the groups conquered by the expanding empires and those otherwise brought into close contact with them. I distinguish here between the more established research question what impact did the conquering society have on the indigenous peoples?; and a newer concern in what ways did the indigenous peoples a<;scrt their identities and maintain or reinforce their cultural systems in response to the challenges and opportunities offered by the expanding empire? Some notable examples of studies that focus on such indigenous groups include Smith's (1986) investigation of elites in societies on the periphery of the Aztec Empire, D'Altroy's (1992) studies of indigenous populations within the lnca Empire, and Alcock's (1993) research into Roman-occupied Greece. These studies show that the indigenous societies had important effects on the imperial culturcs. ln the ca<;c I shall discuss below, indigenous groups played major roles in guiding the course of imperial conquest and in the establishment and subsequent management of provincial systems of administration and supply . The question can now fairly be put, to what extent do the central authoriti es in empires determine the course of events, and to what extent arc empires dependent on compromise and negotiation with the societies they incorporate? A world systems approach to the question of the role of the conquered and nc ighboring peoples in empires can help to draw attention to the interactive a<;pcct of all relations in imperial situations (sec useful recent discussions in D'Altroy 1992:14-16; Ferguson and Whitehead 1992:4-8). Conquered peoples, andpcoples situated beyond imperial frontiers with whom empires interact through trade, all need to be viewed a<; part of the same ("world") system. We need to view indigenous peoples, not just in terms of how they react to the imperial power, but rather as active participant<; in the construction of the contexts of interaction (Hall 1986). From this perspective, we can shift our central question from "what effect did the empire have on group X?", to "what effect did group X have on the empire?" [Page 2] Journa I of World-Systems Research Archaeology can contribute to this developmen t in theory in two important ways. First, archaeology can examine cases of early empires and the changes associated with them over long periods oftimc. ln a landscape for which a good databa<; c of archa e ological material exists, we can examine changing circumstances and adaptive patterns from pre conquest times, through the period of conquest, and in different phases of post-conquest time. Such processes of change can take place over several centuries, and the availability of comparable archaeological materials from different periods makes broad-scale studies of change possible. Second, in contrast to historians dependent upon textual sources, archaeologists can examine all levels of society, not just elites and major communities, to gather information about change. 1n the study of the material manifestations of "everyday life" among the majority of people in a society, and changes in the patterns over time, archaeologists can make their special contribution to research into the broad impact of empires. Whereas historians working with texts depend upon the subjects that interested early writers, archaeologists can consult a theoretically unlimited range of material evidence pertaining to settlement, manufacturing, trade, status expression, and everyday life. The question that will form the focus of this paper is that of imperial control over resources and production. ln the World Systems model, an essential dynamic is that between imperial systems that expand over space and in their capacity to consume resources, and indigenous societies that interact with the imperial powers (Schortman and Urban 1992:18; Sinopoli 1994). A wide range of different patterns of interaction can be identified in different circumstances. As I shall demonstrate here, from the evidence of some early empires, the imperial societies had far less control over interaction with other groups than many analyses that have used the World Systems model might predict. The case I examine here is the Roman Empire in temperate Europe. THE ROMAN EMPIRE IN TEMPERATE EUROPE Communities in Italy and in the lands north of the Alps had been engaged in trade interactions since at least Neolithic times, and by the first part of the Late Iron Age, 500300 B.C., both trade and extensive movements of persons across the Alps is apparent in the archaeological evidence (Wells 1980). At the beginning of the fourth century B.C., invaders from north of the Alps attacked towns in Italy, even sacking Rome in 387 B .C. 1n the subsequent two centuries, Rome built up its defenses and embarked on milita ry expansion throughout the peninsula ofltaly. Rome extended its domain across the southwestern Alps [Page 3] Journal of World-Systems Resear ch into southern Gaul, where it established the colony of Gallia Narbonensis around 120 B.C. (Rivet 1988). Between 113 and 101 B.C. a group called the Cimbri, apparently from northern Europe, moved into central and southern parts of Europe. Together with other groups that join ed them, the Cimbri defeated Roman armies in a series of battles until they were lillallybeatenat Fe1rara i1111orthen1Italyi11 lOl B.C. The early fourth cc11trny B.C. attack on Rome, and to an e-.'i!n greater extent the incursions by the Cimbri and their allies iJ.1 the late secomt cenu:ny B.C., had profoi:n1d efl"ccl, 011 Romai1 thiJ.1kiJ.1g about the security of11orthen1 Italy amt ofRome itself, amt about the character of the little-know11 peoples bey011d the Alps (Timpe 1989:241-343; Christ 1995). Romai1 trade gooll~ arc well represented thrnughout cemral ai1d westent Europe .fiom the start of the secomt ce11li:ay B.C. 011 (Will 1987). But the Romai1 decision to embark on the c011quest of(',aul il158 B.C. represented a major llcparti:trc fu,m earlierpaltent~ ofi11teracti011. There continues to be llcbate about the reas011~ for Caesar's llccision to nivade (',aul. Mi:tch recent thnikn1g has emphasized the powerpolilics iJ.1 Rome at the time, ai1d Caesar's desire for a llcci~i-.'i! advai1tage over hi~ political rivals. But 011e major factor iJ.1 Caesar's deci~ion to fight ill Gaul, ai1d other Romai1 leaden;' subsequent action.~ elsewhere in temperate Eurnpe, was concem a0-,ut estabfahi11g a seci:trc fu,ntier to the 11orth, to prntect Rome agailt~t ftttttre threats of attack by groups like the Cimbri (ChrM 1995 ). Belwee11 the yean; 58 ai1d 51 B.C., .lnlfrL~ Caesar led Romai1 armies in the c011quest of Gaul the laiul~ of modem Frai1ce, Belgium, ai1d Gennai1ywest of the Rhine (Drinkwater 1983). In the year 15 B.C., the Romai1 general~ Dm~,L~ ai1dTiberii:L~ led the c011quest of the laiuls that compri~e (',crmai1y ai1d Austria sonth of the Daimbe (Schcii1 986). Forays across the Lower Rhine nlt:o th,! region between the Rhil1e ai1d the Elbe, mallc iJ.1 ai1 attempt to extend Romai1 imperial control to the Elbe, were called oJT when three legions i:nulcr the lea<.lcrship of the general Varns were aintihilated in the Teutoburg Forest in A:D. 9. Fn1ally, i11 A:D. 83 Romai1 aimies completed the c011qw!st of southwestem Gennai1y, establishn1g a11ew imperial boi:n1daiy lil1e amt wall the lim<'.~ to lilik their Rhn1e ai1dDaimbe fro11lien; (Fignre 1).


INTRODUCTION
Most of what we know, and think we know, about emp ires in world history concerns the actions, moti vations, and institutions of the imperial societies; we know relati vely littl e about the peoples who arc drawn into imperial contexts, through conqu est or other means.Most empir es [Page l] Journa I of World-Sys tems Research have been develop ed by societies that possessed wr iting, and the written accounts up on which understanding ha<; been ba<;cd -by Romans in the Greater Mediterran ean world, Spanish in South America, British in lndia -have been principally by writers who belonged to the imperial society and who represented the perspective of elite members of that society.1n the ca<;c of early empires, where archaeology plays a major role in our understanding, the material remains of the imperial culture arc usually more substantial and more apparent than arc those of the societies impacted.
Only relatively recently have historians, archaeologists, and others begun to make systematic attempts to understand the experience of the other peoples involved in interaction with empires -the groups conquered by the expanding empires and those otherwise brought into close contact with them.I distinguish here between the more established research question -what impact did the conquering society have on the indigenous peoples?; and a newer concern -in what ways did the indigenous peoples a<;scrt their identities and maintain or reinforce their cultural systems in response to the challenges and opportunities offered by the expanding empire?Some notable examples of studies that focus on such indigenous groups include Smith's (1986) investigation of elites in societies on the periphery of the Aztec Empire, D'Altroy's (1992) studies of indigenous populations within the lnca Empire, and Alcock's (1993) research into Roman-occupied Greece.These studies show that the indigenous societies had important effects on the imperial culturcs.ln the ca<;c I shall discuss below, indigenous groups played major roles in guiding the course of imperial conquest and in the establishment and subsequent management of provincial systems of administration and supply .The question can now fairly be put, to what extent do the central authoriti es in empir es determine the course of events, and to what extent arc empires dependent on compromise and negotiation with the societies they incorporate?
A world systems approach to the question of the role of the conquered and nc ighboring peoples in empires can help to draw attention to the interactive a<;pcct of all relations in imperial situations (sec useful recent discussions in D 'Altroy 1992:14-16;Ferguson and Whitehead 1992:4-8).Conquered peoples, andpcoples situated beyond imperial frontiers with whom empires interact through trade, all need to be viewed a<; part of the same ("world") system.We need to view indigenous peoples, not just in terms of how they react to the imperial power, but rather as active participant<; in the construction of th e contexts of interaction (Hall 1986).From this perspective, we can shift our central question from "what effect did the empire have on group X?", to "what effect did group X have on the empire?"

Journa I of World-S ystems Research
Archaeology can contribute to this developmen t in theory in two important ways.First, archaeology can examine cases of early empires and the changes associated with them over long periods oftimc.ln a landscap e for which a good databa<; c of archa e ological material exists, we can examine changing circumstances and adaptive patterns from preconquest times, through the period of conquest, and in different phases of post-conquest time.Such processes of change can take place over several centuries, and the availability of comparable archaeological materials from different periods makes broad-scale studies of change possible.Second, in contrast to historians dependent upon textual sources, archaeologists can examine all levels of society, not just elites and major communities, to gather information about change.1n the study of the material manifestations of "everyday life" among the majority of people in a society, and changes in the patterns over time, archaeologists can make their special contribution to research into the broad impact of empires.Whereas historians working with texts depend upon the subjects that interested early writers, archaeologists can consult a theoretically unlimited range of material evidence pertaining to settlement, manufacturing, trade, status expression, and everyday life.
The question that will form the focus of this paper is that of imperial control over resources and production.ln the World Systems model, an essential dynamic is that between imperial systems that expand over space and in their capacity to consume resources, and indigenous societies that interact with the imperial powers (Schortman and Urban 1992:18; Sinopoli 1994).A wide range of different patterns of interaction can be identified in different circumstances.As I shall demonstrate here, from the evidence of some early empires, the imperial societies had far less control over interaction with other groups than many analyses that have used the World Systems model might predict.The case I examine here is the Roman Empire in temperate Europ e.

THE ROMAN EMPIRE IN TEMPERATE EUROPE
Communities in Italy and in the lands north of the Alps had been engaged in trade interaction s since at least Neolithic times, and by the first part of the Late Iron Age, 500-300 B.C., both trade and extensive movements of persons across the Alps is apparent in the archaeological evidence (Wells 1980).At the beginning of the fourth century B.C., invaders from north of the Alps attacked towns in Italy, even sacking Rome in 387 B .C. 1n the subsequent two centuries, Rome built up its defenses and embarked on milita ry expansion throughout the peninsula ofltaly.Rome extended its domain across the southwestern Alps C. attack on Rome, and to an e-.'i!n greater extent the incursions by the Cimbri and their allies iJ.1 the late secomt cenu:ny B.C., had profoi:n1d efl"ccl, 011 Romai1 thiJ.1kiJ.1gabout the security of11orthen1 Italy amt ofRome itself, amt about the character of the little-know11 peoples bey011d the Alps (Timpe 1989:241-343;Christ 1995).Romai1 trade gooll~ arc well represented thrnughout cemral ai1d westent Europe .fiomthe start of the secomt ce11li:ay B.C. 011 (Will 1987).But the Romai1 decision to embark on the c011quest of(',aul il158 B.C. represented a major llcparti:trc fu,m earlierpaltent~ ofi11teracti011. There continues to be llcbate about the reas011~ for Caesar's llccision to nivade (',aul.Mi:tch recent thnikn1g has emphasized the powerpolilics iJ.1 Rome at the time, ai1d Caesar's desire for a llcci~i-.'i! advai1tage over hi~ political rivals.But 011e major factor iJ.1 Caesar's deci~ion to fight ill Gaul, ai1d other Romai1 leaden;' subsequent action.~elsewhere in temperate Eurnpe, was concem a0-,ut estabfahi11g a seci:trc fu,ntier to the 11orth, to prntect Rome agailt~t ftttttre threats of attack by groups like the Cimbri (ChrM 1995 ).Belwee11 the yean; 58 ai1d 51 B.C., .lnlfrL~Caesar led Romai1 armies in the c011quest of Gaul -the laiul~ of modem Frai1ce, Belgium, ai1d Gennai1ywest of the Rhine (Drinkwater 1983).In the year 15 B.C., the Romai1 general~ Dm~,L~ ai1dTiberii:L~ led the c011quest of the laiuls that compri~e (',crmai1y ai1d Austria sonth of the Daimbe (Schcii1 1986).Forays across the Lower Rhine nlt:o th,! region between the Rhil1e ai1d the Elbe, mallc iJ.1 ai1 attempt to extend Romai1 imperial control to the Elbe, were called oJT when three legions i:nulcr the lea<.lcrship of the general Varns were aintihilated in the Teutoburg Forest in A:D. 9. Fn1ally, i11 A:D.83 Romai1 aimies completed the c011qw!st of southwestem Gennai1y, establishn1g a11ew imperial boi:n1daiy lil1e amt wall -the lim<'.~-to lilik their Rhn1e ai1dDaimbe fro11lien; (Fignre 1).Shortly after the conquest of these regions, Roman administrators organized the division of the landscapes into provinces of the Empire (Filtzinger 1976).Military camps were constructed, particularly along the frontier lines, provincial capitals and other towns established, and a system of roads and bridges built.The broad outlines of these processes, [Page 4] Journal of World-Systems Research and specific dates of conquests and in some ca~es of the establishment of military ba~es and towns, we lmow from written documents.But the essential issue of how the Roman army, comprising tens of thousands of soldiers, a~ well a~ administrators and other imperial representatives in the provinces, were supplied with goods is only sporadically recorded in the written sources (Whittaker 1994).
The establishment of frontiers on the edges of the landscapes conquered by Rome, and the character of the political, military, social, economic, and religious patterns that developed in the :frontier territories, have been subjects of active research in recent years.Important studies of the frontier zones and the changes that took place in them include those by Dyson (1985); Barrett, Fitzpatrick, and Macinnes (1989); Maxfield and Dobson (1991);and Whittaker (1994); these works contain extensive bibliographies of pertinent literature .The present essay is intended to be a modest contribution to this broad and rapidly-developing field of research on Roman frontier issues.While political, religious, and other factors also played important roles in the interactions between indigenous peoples and Roman occupying forces, my treatment here focuses on a~pccts of the economy.
There is some debate among Roman historians and archaeologist~ about how many individuals moved from Italy to temperate Europe following the conquest.The dominant opinion now is that relatively few made such a move (Dyson 1985 :5).The principal representatives of Rome in the new provinces were the soldiers; other categories of persons from Roman Italy includ ed administrators and merchant~ (Dyson pers. comm.).Thus, the question of how the new imperial presence north of the Alps wa~ suppli ed is essentially that of how the army wa~ supplied.

SUPPLY AND PRODUCTION
The question of supply for the Roman army can be divided into two main categories, food and manufactured goods.My focus here is on the manufactured products.While the provisioning of the troops with food wa.., overseen by the state (Peacock and Williams 1986:58;Whittaker 1994:101-108), for manufactured products, the state seems not to have played an active part in supply, at lea..,t during the first and second centuries A.D. (Oldcnstcin World-Systems Research 1985).The manufactured goods needed by the troops included items required for military service, such a.., weapons, tool..,, clothing, dress paraphernalia, leather straps, belts, and tent..,; and everyday items such a.., pottery.For information about supply systems forthcsc goods, we arc almost totally dependent upon archaeological evidence, since the available text.., do not say much about this subject (Oldcnstcin 1976;Sommer 1988).

Within the Imperial Boundaries
Recent research indicates that in general for manufactured goods, the Roman state did not maintain centralized production facilities for the military.Instead, each military camp had to make its own arrangements, either setting up its own workshops or arranging w ith local craftworkcrs to provide the needed goods (Oldcnstcin 1976:75-84;Sommer 1988:596;Whittaker 1994: 112).Much of such manufacturin g wa.., carried out by workers in the vicitowns that were commonly a ... sociated with the Roman camps and that provided a wide range of goods and services to the troops (Somm er 1988).
Production of pottery wa.., sometimes organized on a large scale, though there were many small workshops a.., well; often, numerous different enterprises manufactured the same kinds of pottery (Greene 1986: 158-167).Much of the fine pottery in use in the provinces, especially terra sigillata, wa.., imported, initially from Italy and subsequentl y from ncwl ycstablishcd production centers in southern Gaul, such a.., at Lyon and at La Graufcs cnquc, and in central Gaul at Lczoux.Later, a.., demand for such fine pottery continued to grow, both within the provinces and across the frontiers in the unconquered area..,, manufacturing facilities were founded further north and ca..,t.Some of them produced great quantities to supply many different communities, includin g both Roman military camps and civilian settlem ents.At the manufacturing center at Rhcinzabern in southwest Germany, for example, it is estimated that over a million vessels were produc ed in the workshops every year (Garbsch 1982:l l) .A number of substantial pottery depots have been identified, where pottery wa.., stored for trade.One such depot found at K cmpt cn in southern Bavaria contained large quantities of terra sigillata made in the Rhineland (Czysz 1986:158).It wa.., buried when a fire destroyed the building in the l60s A.D. The depot wa.., situated in the house of a merchant, located in the center of the Roman town, j ust across the street from the forum; this situation suggests a relatively high status for this merchant .Investi gators disagree a.., to [Page 6] Journal of World-Systems Research whether such pottery circulated in a free enterprise market (Peacock and William " 1986:58) or under some degree of state control (Whittaker 1994: 110).Small potteries also existed, and some pottery was produced at military camps (Sommer 1988:594).Soldiers at the military sites, including both legionary troops from Italy and auxiliari es from the provinces, also used pottery manufactured in local communities, both plain and painted wares.These ceramics frequently represent a direct continuation of Late Iron Age pottery traditions (Fingcrlin 1981;Wieland 1993); often the indigenous pottery on Roman sites is indistinguishable from pottery on local pre-Roman settlements.It seems, therefore, that the defenders of the empire were dependent upon indigenous craft industries for some of the ba"ic necessities of their daily existence.
Although some limited metalwork wa" done at the military camps, the archaeological evidence strongly suggests that during the first and second centuries A.D. most wa" don e outside, in civilian-run establishments, often in the towns (vici) a"sociatcd with th e military ba"cs (Sommer 1988).Evidence includes large numb ers of metalworkin g tools that arc commonly recovered in such towns, but not in the military camps themselves (Sommer 1988:597).Some investigators argue that for the most part, the military wa" not involved in metal production at all, but arranged all supplies of metal goods through indigenous manufacturers (Fischer 1985:482;Sommer 1988:597) .In some ca"cs, evidence for the production of metal goods for Roman troops is recovered at places that otherwise have no apparent link with Roman sites.Oldcnstcin (1976:65) cites an unpublish ed find at Stcinhcim on the Main.A house contained the remains of a chest in which wa" found scrap metal from Roman military equipment.Iron tools found nearby suggest that this building wa" a workshop that produced metal implements for Roman troops stationed somewhere in the area.Oldcnstcin (1976) andC. Well" (1995) provide some numerical informati on that puts the question into perspective quantitatively.Along the Upper Gcnnan-Ra ctian frontier, Oldcnstcin estimat es about 20,000 Rom an soldiers served at any one time in the lat e second and early third centuries A.D. C. Wells (1995:611) estimates about 90,000 troops stationed on the middle and lower Rhin e.Typical weaponry for each soldier included helmet, body armor, shield, spear, sword, and dagger.Helmet" and armor were made of iron and leather, shields of wood and leather, with iron bands across the front and iron hand-guard.Spear, sword, and dagger were of iron.In addition to the actual weaponry, each soldier wore an average of 10 or more bronze objects, including pins, buckles, strapcnds, and various [Page 7] Journal of World-Systems Research ornaments (Oldenstein 1976).Thus the total quantities of iron, leather, and bronze required by the Roman soldiers in temperate Europe were va..:;t.
If each Roman military base needed to arrange the supply of all, or most, of the se goods from indigenous producers -and the evidence suggests that supply worked this waythen the Roman occupying forces were very much dependent upon the local groups.Without the constant cooperation of the local producers, the Roman venture would have failed.In such a relationship between dependent occupiers and local producers, negotiation and compromise arc likely to have played a greater role than exercise of power over the indigenous peoples.Thus the question of the character of relations between representatives of the imperial power, and local indigenous craftsworkers and local leaders, becomes a critical issue.The Roman troops must in turn have introduc ed considerable wealth into the communities that supplied the needed goods.Such reciprocal arrangements surely contributed in a major way to the growing intensity of economic activity in the provinces of Gaul and Germany during the first and second centuries A.D.

Beyond the Frontier
Interactions between the Roman provinces and the lands beyond the :frontier are well docum ented.The clearest indication of the chronology, extent, and character of the interactions arc the large quantities of Roman objects found all over the lands from the imperial :frontier northward a..:; far a..:; Norway, Sweden, and Finland, and ca..:;twar d a..:; far a..:; Russia (Hansen 1987;Hedcagcr 1987).Thes e Roman -made goods includ e vessels of bronze, gla..:;s, pottery, silver, and gold; coins; statuettes; and je welry.These objects often occur in exceptionally rich burials, but also on settlement sites .
Roman writers, particularl y Tacitus and Dio Ca..:;sius, provide another perspective on Roman interactio ns with the peoples ca..:;t and north of the frontier (Hansen 1987:234;Whittaker 1994:113-127).The written accounts concern mostly interactions w ith peoples in area..:; close to the frontiers, and they mention as trade goods coming into the Roman lands ox hides, oxen, horses, slaves, weapons, grain (a..:; tribut e), and amber.Th e textual sources arc not very precise, and they do not provide much information about source locations or [Page 8] Journa I of World-Systems Research quantities traded.Other goods that arc likely to have been involved in such cross-frontier trade (given what we know about interactions in comparable situations in other times and places), but that have not yet been identified archaeologically, includ e furs, woo l, textiles, honey, and wax.The thousands of Roman-made object s that have been found across the :frontier indicate that interaction must have been substantial, and there is every reason to think that supplies for the Roman troops were being traded across the border.One key piece of evidence wa-; recovered at Tolsum in the Netherlands.A tablet wa-; found, bearing a Latin inscription that document-; a transaction in which Roman buyers purcha-;ed cattle from native sellers (Boeles 1951: 129-130).Tacitus, writing around AD. 100, mentions ox hides a-; tribute paid by the Frisians to Rome (Annals 4, 72;Whittaker 1994:113).
The present state of research makes it difficult to link directl y goods that Roman suppliers obtained from producer communities across the frontier with archaeological evidence for such a supply system.But there is good rea-;on to think that the rapid expansion of production activities in iron and in cattle in regions close to the frontier wa-; directly related to this provisioning.Roman troops needed large quantities of iron for weapons, tools, nails, and other purposes.The archaeological evidence in regions across the frontier shows rapid and widespread expansion of iron production during the first, second, and third centuries AD. (Grunert 1988;Lcubc 1989;Henning 1991 :72), at the time that the Roman army wa-; establishing and outfitting its frontier posts.Examples of such expansion arc recently-excavated iron-smelting sites at Gcra-Tinz in Thuringia (Dusek 1989:561-562), Ricstcdt in Saxony-Anhalt (Grunert 1988:478), and Bark.ow in Mecklenburg-Western Pomerania (Lcube 1989:162).At Gcra-Tinz, for example, investigators found 21 smelting furnaces in an area mca-;uring 10 x 25 m, a-;sociatcd with remains of a small settlement dated by a-;sociatcd pottery to the first, second, and third centuries.Significantly, the production at these sites, and at others in the lands beyond the frontier, wa-; carried out in numerous very small-scale operations.No sizabl e, specialized iron-producing facilities have been identified in the lands near the imperial frontier, but instead many small farming communities that produced surplus metal.Only at a distance from the frontier, in the Holy Cross Mountains of southern Poland, do we sec clear indication of the growth of a large-scale center for th e production of iron at this time (Jazdzcwsk:i 1965:153-154).
Many sites, particularly on the sandy soils of the North European Plain north and ca-;t of the Lower Rhine frontier, show incrca-;cd production of cattle.Among the best evidence is that from Feddersen Wicrdc on the North Sea coa-;t near Brcmcrhavcn, Germany.The settlement wa-; established around the middle of the final century B.C., and it wa-; occupied throughout the Roman Period .The settlement surface wa-; built up over time to form a wurt,

[Page 9]
Journal of World-Systems Research or artificial mound, above the surrounding flat low-lying land (Haarnagcl and Schmid 1984:204-212).Preservat ion of wood wa-; exceptionally good in the wet environm ent.The characteristic structure on the settlement is the Wohnstallhaus, a long, rectan gular building divided into a habitation area for the human occupants at one end, and a barn with partitions for livestock at the other.Analysis of the foundations of stalls in the buildings from the different pha-.cs of habitation at the site indicates an incrca -.c in the total livestock capacity on the settlement from 98 stalls at the beginning of the occupat ion to 443 stalls during the second and third centuries (Haarnagcl 1975 ).Imports from th e Roman lands arc abundant at Feddersen Wicrdc; they include terra sigillata pottery, gla-.sbeads and vessels, coins, and millstones (Haarnagcl 1975(Haarnagcl , 1979)).The evidence suggests an intensification of the production of cattle during the first and second centuries at the site, and a concomitant incrca-.c in quantities of imported Roman trade goods.Around the end of the first and beginning of the second century, the excavator identifies evidence indicating increasing social differentiation.One building was constructed that is larger and more substantial than the others on the settlement.During the second and third centuries, greater concentrations of Roman imports arc associated with this structure and its successors, and greater quantities of metal-working debris arc found in and around it.At the end of the second and start of the third century, the large structure on this special part of the settlement was separated from the rest of the site by a palisade.Nex t to it was a fenced area with granaries and places where metalworking was don e (Haarnagcl and Schmid 1984:208).Haarnagcl suggests that the occupant of this special precinct directed craft production and trade for the community.In the course of the third century, a decrease in economic activity is apparent at Feddersen Wicrdc, a process that continues during the fourth century and results in the abandonment of the settlement in the fifth.

DISCUSSION
The maintenance of the Roman Empire's frontier in temperate Europe depended upon supplies produced by local groups, working in manufacturin g traditions that had developed in the prehistoric Iron Age.In fact, a substantial proportion of the goods provided to the Roman troops were versions of prehistoric Iron Age material s .Pottery and fibulae arc two categories [Page 10] Journa I of World-Systems Research of goo ds that illustra te this pattern.Pottery provides importan t evidence, since it is we ll preserved and abunda ntly represented on both native and Rom an sites.Monochrom e handmade pottery virtuall y identical to that at Lat e Iron Age sites such as Manching (Stocldi 1979) , Altenburg, andKclh cim (Wells 1993), has been recovered on num erous Roman Period settle ments, for example at the milita ry camp at Dangstcttcn on th e Upper Rhine (Fin gcrlin 1986;Wieland 1993), at Rottwcil in Wiirttcmberg (Planck 1975), and at Kcmptcn in Bavaria (Mackcnscn 1978;von Schnurbcin 1993).Fine whe el-mad e pottery decorated with horizonta l and vertical red painted bands matching a typical cera mic category from the Late Iron Age similarl y occurs on numerous Roman sites, including Kcmptcn and Straubing.Many fibulae (ornamental clothing fasteners) from t he military sites are of forms that are identical to local Late Iron Age types or that deriv e directly from them (e.g.Planck 1975, plate 67, from Rottwcil;Ricckhoff 1975, plates 1 and 3, from Hufingcn).It is thLL-; apparent that a substantial portion of the objects used in everyday life at the Roman military sites were manufactured by the indig en ous groups in the surrounding landscape, working in their traditional techn ologies and styles.
A-; Roymans (1983:58) argues from the Dutch evidence, the Roman occupiers surely stimulated the economics of the indigenous communities -both within and beyond the imperial :frontier -by arranging to acquire from them the goods they needed.As th e n eeds of the Roman army grew, some local groups shifted their technology and style of production to suit the wants of the Roman occupying troops, as well, of course, as of others who desired the new "Roman" fashions.This process oftransfonnation of indigcnoLL-; craft traditions is well illustrated in the sequence of pottery production at Schwabcgg in Bavaria (Czysz 1987).Kiln debris, including typical Late Iron Age pottery, on the site attests to production at Schwabcgg before the Roman Period .Early in the first century AD. a specialized pottery manufacturing communit y was establish ed at the site, and by the end of that century it had become a highly specialized center, producing a variety of kinds of pottery and serving a wide mark et.Among the pottery manufactured was a fine ware with white paint and red painted bands, a type that represents the continuation of a characteristic Late Iron Age ceramic.From thi s period, 55 kilns have been identified on the site .Continuity in the manuf act urin g traditions is apparent not only in the form and decoration of vessels produced, but even in the ident ity of the personnel -a high proportion of the personal names repr esented in stamps on the pottery arc local Celtic names.Toward the end of the second centu ry, the pottery industry at Schwabcgg began to specialize in the manufacture of terra sigillata.The product-; were shipped to communities in all directions from Schwab cgg , and they have been identifi ed at forts on the limes and a-; far ca-;t a-; the province of Pannonia.

[Page 11] Journa I of World-Systems Research
This example of po ttery pr oduction at Schwabc gg indic ates that we need to view the pattern of supply along the fron tier a-; an interacti ve system, with the Roman occupiers dependent upon local producers, and the local manufacturers adjustin g the output of their traditional industries to suit the scale and ta-;tc of their customers.Some loca l indi v idual-; gained in wealth and status throug h this commercial arrangement, and they arc rep resented in unusually rich burials durin g this period -burials reflecting the practi ces of the prehistoric indi genous peoples, but with Roman a-; we ll a-; native goods in them.An example is grave 8 at N ijm cgcn in the Netherlands , datin g to between AD. 80 and 100 (Koster 1993) .The grave contained the remains of a cremation, placed insid e a gla-;s urn, and numerous oth er goods, in traditi onal Iro n Age, and distinctly non-Roman, fa-;hion.The goods includ ed weapon s (three spears and a shield), a 23 -piece dinin g set of terra sigillata from the prod uction center at La Graufcscnquc in south ern Gaul, num erous ornate gla-;s vessels, five bronze vessels, and a set of writing implements.The identity with Rome is cmpha-;izcd by the pottery, gla-;swarc, and writing utensils; but the composition of the grave a-;scmblagc shows that it belongs to the native, pre-Roman, tradition.
The examples cited above arc only a very small portion of the rapidly-accumulating evidence that indicates that much of the material culture known a-; "provincial Roman" wa-; in fact made by indigenous peoples in the conquered territories, often using manufacturing techniques and expressing styles that developed directly from their pre -Roman, Iron Age craft traditions.Such evidence, which is only now gaining serious, focused attention among investigators (e.g. Millett 1990;Wieland 1993) raises the fundamental question, what docs the word "Roman" actually mean in this context (Freeman 1993)?It is clear now that most of the architecture and everyday material culture that is cla-;sificd a-; "Roman" in temperate Europe wa-; not made by individuals from Rome nor even by Roman citizens resident in the provinces, but rather by indigenous parties who, after the conquest, found themselves living under the Roman political structure and amidst the pcrsua-;ivc influence of Roman fa-;hion.As the archaeological evidence makes abundantly clear, after the conquest (and even before it, to a limited extent) Roman material culture and style became extremely popular with the majority of the populations of the provinces.Most people seem to have wanted to be a-; "Roman" a-; they could, displaying this new identity through the adoption of all possible a-;pects of Roman material culture, including pottery, personal ornaments, clothing, tools, and architecture.Agachc (1978) demonstrates the indigenous adoption of the "Roman" villa a-; a style of habitation, and Jones (1987) argues that essential features of the "Roman" cities and towns in temperate Europe were sponsored and construct ed by local elites in the context of [Page 12] Journal of World-Systems Research indigenous rituals of competition.We must therefore understand the word "Roman," when used in reference to temperate Europe, to designate a style -of archit ecture, pott ery, weaponry, ornaments, and so forth, that wa-; eagerly adopted by local people a-; a means of demonstrating their feelings of identity with the new cosmopolitan civilization under whose dominion they lived.This fashionability of the Roman style among the indigenous peoples did not la-;t long .Already during the first century A.D., within a century after the conquest, new styles of material culture, often with strong clements of the prehistoric Iron Age traditions, developed in the Roman provinces.Among the best examples arc several new categories of pottery that were established during the first and second centuri es in temperate Europ e.One is "Ractian ware," a type of hard-fired, polished pottery with relief decoration, that became immensely popular late in the first century AD. and wa-; produc ed by numerous local worbho ps in Bavaria (Czysc 1986:159-160).'Norican war e" wa-; another new product of the first and second centuries.It wa-.made in small-scale potteries and is characterized by hard-fired, wheel-made vessels with coarse temp er and rough surfac es.Decoration is in the form of comb incisions, incised wavy lines , and profiled rid ges.Maier (1983) argues that Norican ware, which embodies clements from th e Late Iron Age ceramic tradition, represents the expression of indigenous identity , a-.scrting it-.clf against the growing homogeneity of much of Roman material culture.
Beyond the :frontier, there is no evidence of "exploitation," in the sense of th e Roman Empire draining resources away from the indigenous communities.The evidence suggests rather what Hall (l 986, 1989) ha-.called "incorporation."According to Hall's model, incorporation is a process by which non -state societie s that int eract with imperial states become linked economically with the imperial states.As a result, both societies undergo certain changes in social and political configurations.The non-state societies play active roles in such changes.The archaeological evidence shows an incrca-.c in local industrial and livestock production, for supply to Roman provinces, a-.noted above in the examples of iron-working and cattle raising, along with a wide range of changes a-.sociatcd with the economic upswing (Lcubc 1989: 164 ).These changes, which arc apparent in many different regions across the imperial frontier, include the formation of larger communities, development of new technologies, adoption of new styles from the provincial Roman world, and greater expression of status differentiation, largely through display of Roman luxury imports (Hansen 1987), but also through extravagant employment of indigenous architecture and craft product-..There is no evidenc e, either archaeological or textual, to suggest that the Roman Empire exercised any kind of control over the peoples beyond the :frontier or over their production of th e goods that were desired by the Roman provinces.

[Page 13] Journal of World-Systems Research
The evidence that I outline in this paper pertaining to the situation in Roman Period temperate Europe suggests that we need to question even the extent to which the Empir e controlled resources and supply 1vithin its own borders.If each milit ary base was dependent upon production of foodstuffs, pottery, and metal equipm ent by communities of indi genous peoples, then a model of power and control is not the best way to examin e this rela tionship .lnstcad, we need to address issues of negotiation, interaction, and mutual self-interest in order to come closer to understandin g relations between the imperial power and the indigenous groups .These considerations lead to the question, who exactly is the empire, when we speak of provincial Roman actions north of the Alp s? Whose interests arc represented by the concept of the empire, and who carries out decisions to further those interests?The material evidence suggests that a wide variety of different interests were involved, and that treating the empire as a united entity is not helpful in understanding the dynamics of the relationships.The techniques of archaeology allow us to examine these rel ationships in detail and over time.The example of Feddersen Wierde illustrates how instructive such cases can be when evidence for long-term proccssual change in patterns of settlement structure, local production, and long-distance trade can be examined.

MODEL-BUILDING: FROM THE SPECIFIC CASE TO AN ANALYTICAL FRAMEWORK
The Roman Empire in temperate Europe provides an instructive case for examining questions of imperial relations with indigenous peoples, both within and beyond the empire, because of its unusually rich and well-studied data base.We can use this cont ext to develop an analytical framework that can be profitably applied to other situations of imperial relations with indigenous peoples.For this study, I have examined the issue of the provisioning of the empire's military force, since that problem should provide insight into the most extreme concern of supply for the Roman administration.The very basis of the security of the Empire in temperate Europe was the army; and the security of Rom e rested upon the security of the provinces to the north.

[Page 14] Journal of World-Systems Research
The evidence regarding the ways in which the Roman army solved this problem of supply , outlined above, makes nec essary a full rethinking of notions of imp erial pow er and control.World Systems Theory, when applied to empires, ha<; sometimes viewed imperial activity too rigidly and narrowly, without taking into account the myriad local interac tions betwe en representatives of the empire and indig enous groups.Th ese interactions -in all their variety -can be examined through focused archaeological research on settl ement and cemetery sites at different locations, both within th e imperial territories and beyond them.In the case considered above, the nec essity of supplying troops on the frontier indicates the need for negotiation and adaptation on the part of the army and its personn el.This ca<;c points up the need to reorient our investigations of imperial situations away from questions of power and how it is used, to questions of interests, mutual intcrdcpcnd cncc, and interactions maintained to further the intere sts of all involved.
The insights offered by this ca<;c can be used to develop a general :framework for analysis of relations between imperial powers and the indigenous peoples with whom they interact.It is clear that we cannot accept uncritically the cont ents of surviving written documen ts from the imperial socie ties, but mus t examine the material evidenc e on the ground.The Roman ca<;c presented here shows that relations between the empir e and indig enous groups were situation al -Roman troop s needed to supply themselves w ith goods, and they needed to secure these goods through arrangements establish ed with local groups.Any more detailed analysis of these relations needs to focus on the economic, social, and political configurations among the local peoples and to inc ludc examination of such variables a<; environment, community size and organization, and craft traditions, in order to reconstruct the development of commercial relations between Roman consumers and indigenous producers.
Analysis must begin by a<;scssing the need<; of the imperial power in the particular environment, then turn to establishing potential sources for fillin g those need<;.Production sites in the land<;capc will provide the clearest evidence of the sources -in the case above, kilns, iron-smelting furnaces, and barns for raising cattle.After the sources have been identified, then analysis can turn to evidence for the response of communities to the opportunities presented by interaction with representati ves of the imperial power.Evidence will be in the form of imported goods, and of local craft products that show effects of interaction such a<; adopted technologies and styles.Change evident in the indigenous communities, such a<; growth in community size, expansion of production facilities, and increa<;ing differentiation expressed in houses and graves, can be integrat ed into this analysis.

[Page 15] Journal of World-Systems Research
My intention in presenting this particular ca<;e of Roman troops in temperate Europe is to use the specific instanc e to draw attention to some of the different kinds of evidence that can help us to analyze the content and character of interactions between empires and indi genous peoples.

[
Page 3] Journal of World-Systems Resear ch into southern Gaul, where it established the colony of Gallia Narbonensis around 120 B.C. (Rivet 1988).Betwe en 113 and 101 B.C. a group called the Cimbri, apparently from northern Europe, moved into central and southern parts of Europ e. Together with other groups that join ed them, the Cimbri defeated Roman armies in a series of battles until they were lillallybeatenat Fe1rara i1111orthen1Italyi11 lOl B.C.The early fourth cc11trny B.