Legitimation Crises in Premodern Worlds

Scholars employing world-system theory have tended to examine how world-system develop and expand, while few have addressed the fragmentation or collapse ofworldsystcms. This paper explores the conditions of world-system collapse using Habcnna-;'s concept oflcgitimation crisis a-; a starting point. The paper posits that legitimation crises arc a recurring problem in world-system-; and have led to collapse in a number of ca-;cs. Prehistoric North American and Pacific world-systems arc used a-; examp les. Introduction If the world-syst em perspective is indeed rooted in the proposition that "everythin g is process," a-; Wallcrstcin stated in his seminal 1974 article, then collapse (which I define following Tainter [1988: 4] a-; a significant loss of an established level of sociopolitical complexity) should be one of the processes we arc interested in. In fact, our work should equally weigh rise and collapse, centralization and decentralization, growth and decline, viewing them a-; alternate outcomes of a singular process of world-system operation , rather than a-; polar oppositcs--onc occurring when a world-system is functioning well, the other when it ha-; broken down. Despit e this, world-systems analyses have rarely focused on collapse, even though world-systems theory should be particularly useful for investigating collapse becaus e crisis in one part of the system could, due to the interdependency of politics in the world-system, lead to crisis in the system a-; a wholc--a process that seems common (Tainter 1988). Existing theories of collapse tend to a-;sumc environmental crises or failures in the subs istence economy arc the ba.:;is of collapse , and often this a-;sumption is not clearly articulated. A perusal of Taintcr's (1988: 39-90) comprehensive summary of these theories will demonstrate that many discussions of collapse, while not explicitl y concerned with the [Page l] Journa I of World-Systems Research environment or subsistence economy, implicitly describe collapse in term s of environmental degradation, overpopulation, increasingly marginal returns on the energy put into complexity (Taintcr's own vicw--1988: 118-123), or the like . In short, these theories implicitly link collapse to crisis in the subsistence economy; to a crisis in the ability of individuals to maintain life processes (also sec the papers in Yoffce and Cowgill 1988). In contrast I suggest collapse is equally likely to stem from a crisis in social rcproduction--from an inability of individuals to sustain themselves socially-and not from individuals' inability to sustain themselves physically. This paper considers the collapse of world-systems and posits a theory emphasizing social reproduction to explain why repeated patterns of rise and decline appear so common (Anderson 1994; Blanton ct al. 1996). The theory is rooted in the work of Jurgen Habcrmas, a controversial figure whose sociological writings arc difficult, uneven, and often flawed (sec, for example, Bernstein 1985), but one whose insights into the workings of the Capitalist world-system, and, by extension, other , prcmod crn worldsystems, have been overlooked. To be fair, both to Habcrmas and to world-systems theorists, Habcnnas has never embraced the world-systems persp ective and has never talked about the Capitalist "world-system ." However, Habcnnas' s work on the culture of late Capitalism is explicitly cybernetic in approach, and Habcnnas views Capitalism as an holistic system of interdependent politics much as Wallcrst cin docs. Habcnn as lacks the specific theories of worldsystem process, of geographic differentiati on and competition, of unequal exchange and uneven development , but his ideas are, I suggest, amenable to the world-syst em perspective, and can easily be incorporated into it . The primary distinction between Habcrmas's perspective and a strictly worldsystems one is that Habcn nas views the economic, political, and social subsystems of Capitalism as having equal importance . In other words, the economy is not seen to be the only, or even the most important, force in the system's operation. Indeed Habcrmas (1973) argues that these three subsystems arc so intertwined that they cannot be reasonably separated . His insistence that the political and social be given equal weight to the economic is a product both of his philosophy of "communicative action," which envisions all social forms as created through communication between "rational" individuals, and his understanding of how the Capitalist system operates (sec Habennas 1976) . Figure l is a diagram ofHabcnnas's (1973) conception of the Capitalist system. On the far left is the economic system, the privately-owned enterprises which produce goods and [Page 2] Journa I of World-Systems Research services for profit. The arrows going to and from it show how it is aided by the politi cal system, which develops laws and policies beneficial to economic interests, and which works with other politics to maintain favorable conditions for growth. In this way , the political system helps to steer the economic system to maximum performance for private, prn.fit-drivcn interests. In rct,,Qn, the economic system .financially supports the political S}'Stcm, -which cannot main min itself othciwi-;c since itprndw::cs nothing beyond steering the economic system and pro,iding social welforc, which lcall<; to the other side oftl1c diagram. Steering Social welfare r performances performances ---i Economic Political Socio-cultural system system system Fiscal skim-off ___j' t_ Mass loyalty ........................................ .. .................. ... ...... ... ................... .......... ....... ........ . .. f 1g,,irc l. Habc1ma<;'s conception of Capitalist soo10-p., Iiti cal organization . The right side offigurc l shows tl1c socio-cu.lt,,iral S}'Stcm, ba<;ically tl1c traditions, belief.<;, norms, values, expectations, and tl1c like, which arc shared by members of the polity. As tl1c .fig,,,re shows, these tlallition<; a:rc aided by tl1c polity tlrmi,gh social welfare prngrams -winch Sl.lJlPOrt tl1cm, leading, in t,,irn, to m!l!;S popt,lar Sl.lJlPOrt for tl1c polity. The political system i,L<;cs fmancial resoi,rocs generated thrni,gh tl.:: economic system to Sl.lJlPOrt the socio..:1ult,,,ral S}'Stem, -which, in t,,irn, legitimates tl1c polity's existence and right to go,em, i.e., its right to create and implement laws and policies beneficial to the maxiimim operation oftl1c economic system. Thi<; intenlcpcmlcncyis the ba<;is of Habcnna<;'s ,icw of Capitalist societies and tl1c bil<;is ofhi<; conception of legitimation cnscs. Uccim<;c the tl1rec systems arc tightlyintenlcpcndcnt, a crisi<; in any one of them may !call to a systemic cri<;i<; of the whole. However, lfabcnna<; (l 973.J suggests tl1attl.:: weak point in tl1c S}'Stem is in the 'ina<;s loyalty" anow !calling fu,m tl.:: socio-cult,,iral system to the political S}'Stem. His reason for tlris is complicatcl~ m,t i<; bil!;ically that the political S}'Stcm can control c,1!rytlring except people's "rational" miml<;, and crises in anypart of the system arc going to tend to prodiicc crises in legitimation, precisely bcca1L<;c it annot be rcallily contrnllcd. Habc1ma<;'s insight i<; 01,i.cial for the di<;mL<;sion of colbrpsc in world-systems. Wliat Habcnn!l<; arg,,ics is tlmt an cn,irnnmcntal calamity or cri<;i<; in the subsi<;tcncc economy i<; not a necessary, or even common, prccomlition for collapse.; rnthcr, a crisis in tl.:: socio-mllt,,,ral S}'Stem, a legitimation cri<;is, is a more likely soiircc of political collapse. Clearly there arc some iiniq,,,c fcattircs of Capitalism tliat botl1 ilhiminatc and confiL<;c Habcnna<;'s perspective as applied to non -Capitalist societies, partiC1'1arly the intimate link between knowledge, power, and capital (Lyotanl 1984), b11t it is my assertion that the interrelated Sy-stem Habcnnas prnposcs for Capitalist societies exists i11 all politically ccntralii.cd societies, albeit 11ot in 1hc same mmnicr as it docs ill Capitalist ones. One clear prnblcm witl1 Habcnnas's mod.cl is tliat he fails to emphasize tl1c importmicc of social reproduction. Habcnnas foc,,Lscs solely 011 what I call tl1c s,,ibsistencc economy, mul not Oil [Page :\] Jm1ma l ()( W()rl<l-Syste11i~ R< wyu .. ·h wliat I call the prestige economy, the system tlm,iLgh which indi,iduals create mid mail1tail1 social sta111.lilig, prestige, power, and the like. I arg11c tliat 1rulilturi1li11g these arc ,itally important (sometimes cvc11 more ilnport1u1t tluu1 maintai1li11g life) in all societies, c,-cn Ca:pilltli..~ ones. I suggest we cm1 re,isc Habcnnas's diagrmn to giv-c emphasis to ~,cial repn-xluction by rcphiciligthc "economic Sy-stem" with "prestige-Sy-stem", the "political system" with "elites", mul the "socio-c,,llt11ral system" 'INith "non-elites". The revised diagram is sho'INll in Fig11rc 1. Steering Social rperformances ~ reproduction ~ performances Prestige system Elites Non-elites Fiscal _j' t_ Mass : skim-off loyalty .. ................................................................................................ ....................... Fig11re 1. Habcnnas's conception of socio -political organization r,.,'\-iscd to fbc,,is on social rcprnduction. lil Fig11rc 1 ''prcstig,;:-systcm" reprc~nts the myriad ofway'S in which prcstig,;: is accrued mid mail11irined ill the society. Jt inclilllcS knowledge, rin1.1tls, mul Sy1nh,ls which 0:,n,-cy mid di..~lay stanlS. Elites help to "steer" the prestige-system tlm,iLgh S1.Qllpt11ary l 'IN'S, policies, a.ml rcgillatioiis. By giiitli11gthis ystem the elites effectively cns11rc their O'IN11 StlltilS, b11t 1hcy also prn,idc opportulitics for stanlS cnhm1ccmcnt to 1hcir followcTS, the "non-elites" il1 the sy-stem.111cn01i-clites, in rctum, pro...ide loyalty to the elites. The system work-; similar to the way Habennas sees Capitalism operating, except that profit here is in terms of prestige rather than of capit


Introduction
If the world-syst em perspective is indeed rooted in the proposition that "everyt hin g is process," a-; Wallcrstcin stated in his seminal 1974 article, then co llapse (which I define following Tainter [1988: 4] a-; a significant loss of an established level of sociopolitical complexity) should be one of the processes we arc interested in.In fact, our work should equally weigh rise and collapse, centralization and decentralization, growth and decline, viewing them a-; alternate outcomes of a singular process of world-system operation , rather than a-; polar oppositcs--onc occurring when a world-system is functioning well, the other when it ha-; broken down.Despit e this, world-systems analyses have rarely focused on collapse, even though world-systems theory should be particularly useful for investigating collapse becaus e crisis in one part of the system could, due to the interdependency of politics in the world-system, lead to crisis in the system a-; a wholc--a process that seems common (Tainter 1988).
Existing theories of collapse tend to a-;sumc environmental crises or failures in the subs istence economy arc the ba.:;is of co llapse , and often this a-;sumption is not clearly articulated.A perusal of Taintcr's (1988: 39-90) comprehensive summary of these theories will demonstrate that many discussions of collapse, while not explicitl y concerned with the [Page l] Journa I of World-Systems Research environment or subsistence economy, implicitly describe collapse in term s of environmental degradation, overpopulation, increasingly marginal returns on th e energy put into complexity (Taintcr's own vicw--1988: 118-123), or the like .In short, these theories implicitly link collapse to crisis in the subsistence economy; to a crisis in the ability of individuals to maintain life processes (also sec the papers in Yoffce and Cowgill 1988).In contrast I suggest collapse is equally likely to stem from a crisis in social rcproduction--from an inability of individuals to sustain themselves socially--and not from individuals' inability to sustain themselves physically.This paper considers the collapse of world-systems and posits a theory emphasizing social reproduction to explain why repeated patterns of rise and decline appear so common (Anderson 1994;Blanton ct al. 1996).The theory is rooted in the work of Jurgen Habcrmas, a controversial figure whose sociological writings arc difficult, uneven, and often flawed (sec, for example, Bernstein 1985), but one whose insights into the workings of the Capitalist world-system, and, by extension, other , prcmod crn worldsystems, have been overlooked.To be fair, both to Habcrmas and to world-systems theorists, Habcnnas has never embraced the world-systems persp ective and has never talked about the Capitalist "world-system ."However, Habcnnas' s work on the culture of late Capitalism is explicitly cybernetic in approach, and Habcnnas views Capitalism as an holistic system of interdependent politics much as Wallcrst cin docs.Habcnn as lacks the specific theories of world-system process, of geographic differentiati on and competition, of unequal exchange and uneven development , but his ideas are, I suggest, amenable to the world-syst em perspective, and can easily be incorporated into it .
The primary distinction between Habcrmas's perspective and a strictly world-syst ems one is that Habcn nas views the economic, political, and social subsystems of Cap italism as having equal importance .In other words, the economy is not seen to be the only, or even the most important, force in the system's operation.Indeed Habcrmas (1973) argues that these three subsystems arc so intertwined that they cannot be reasonably separated .His insistence that the political and social be given equal weight to the economic is a product both of his philosophy of "communicative action," which envisions all social form s as created through communication between "rational" individuals, and his understandin g of how the Capitalist system operates (sec Habennas 1976) .Journa I of World-Systems Research services for profit.The arrows going to and from it show how it is aided by the politi cal system, which develops laws and policies beneficial to economic interests, and whi ch works with other politics to maintain favorable conditions for growth.In this way , th e political system helps to steer the economic system to maximum performance for private, prn.fit-drivcn interests.In rct,,Qn, the economic system .financiallysupports the political S}'Stcm, -which cannot main min itself othciwi-;c since itprndw::cs nothing beyond steering the economic system and pro,iding social welforc, which lcall<; to the other side oftl1c diagram.
The system work-; similar to the way Habennas sees Capitalism operating, except that profit here is in terms of prestige rather than of capital.Also, there arc either no individuals who uniquely control the economic sector (i.e., who own the "means of production"), or if there are, they are the same individuals as the elites who run the political sector.In other words, elite manipulation is not focused on production to sustain life, but rather on sustaining system-; of social reproduction.An example may help illustrate this idea.

The Tongan Polity
The Tongan archipelago is located in the Pacific Ocean some 2000 miles ea-;t of Australia and 600 miles southeast of Fiji.It consist-; of about 160 islands strung out over 200 mil es on a roughly northca-;t-southwcst axis (Kirch 1984: 217).The population in the 1920s wa-; about 25,000, most of whom lived on the three large islands of the archipelago (Gifford 1929: At the time of first contact with Europeans (ca.1643), Tonga had a dual political structur e, with two major leaders: the Tu'i Tonga, who wa-; consider ed divine, and was the link betwe en humans and deities; and the Tu 'i Ka nokupolu , who answered only to th e Tu'i Tonga, and wa-; responsible for secular concerns in the chiefdom (Kirch 1984: 224 -25).A-; described byBa-;il Thompson (quoted in Kirch 1984: 225): [the spiritual king]--the Tu'i Tonga--was lord ofthe soil, and enjoyed divine honours in virtue ofhis immortal origin ... The temporal king --the Tu'i Ka nokubolu--was the irresponsible sovereign of the people, wielding absolu t e powe r of l(fe and death over his su~;ects, and was charged with the burden of the civil government and the ordering of the tribute due to the gods and their earthly representative, the Tu'i Tonga. Both the Tu'i Tonga and the Tu'i Kanokupolu had a group of four chicfa and attendants who were known asfalefa, and served a-; the Tu'i's courtiers (Kirch 1984: 230 -31).Subservient to the Tu'i Kanoku polu's falefa were a numb er of local, landholdin g chicfa, or eiki, their attendants, or matapule .Subservient to the landholding chicfa were lesser chicfa, also called matapule, hereditary titled craftsmen, or tohunga, and finally commoners, or tua (Kirch 1984: 231-32) .
Irving Goldman (1970: 314 -15) offers an interesting outline of the Tongan political structure through what he describes a-; four segments of political hierarchy.Th e first segment refers to the Tu'i Tonga and Tu'i Kanok:upolu, who maintain ultimat e power in Tonga.The second segment refers to the chiefa oflandholding lineages , who "were in virtually all respects sovereign in their own jurisdiction.Each major lineage was a replica of the entire administration" (Goldman 1970: 315).The third segment refers to sub-chic£.:; of the major landholding lineages, "and for its most part a replica in most respects of the major branch, dependent on the major a-; the major wa.:; on the Tu'i Tonga or his representatives" (Goldman 1970: 315).Finally, the fourth segment refers to the patrilocal household.Goldman ( 1970: 315) tells us that "A hierarchy of successive dependencies wa.:; the ancient Tongan scheme," and at the ba.:;e wa.:; the patrilocal household.
The organization of these patrilocal households wa.:; reproduc ed in the political hierarchy of Tonga.Generation and sex were the ba.:;is of rank within the family, and rank wa.:; the ba.:;is of political power: "The first-born son took the title, the social position, and the leadership in the family" (Goldman 1970: 289).Although sisters outranked brothers in formal honor, females did not hold political office or power in Ton ga, and the head of the family wa.:; the eldest mal e, and wa.:; succeeded by his younger brother or eldest child (Gifford 1929: 20, 290).
This patrilineal ranking between fathers and sons, older brothers and younger broth ers, wa.:; reproduced in the political hierarchy of Tonga, and formed a ba.:;e for legitimating power.As Gifford (1929: 19) (G[fford 1929: 128).

Primogeniture, Marriage, and Power in Tonga
Power in Tonga was held, at all levels, by social elders.In the family, the eldest member served a..:; head.Lineage chiefa were considered the socially eldest in the lineag e. High er chiefa were "grandfather" or "older brother" to lesser chiefa, and all the chiefa were descended from the Tu'i Tonga.The power elders had over subordinates wa..:; derived , in part, from their control of prestige-goods.
[ A Tongan's social position was relatively fixed .Although there wa.., some room for movement, the status of his lineage, his parents, and his birth-order, all more-or-le ss detem1ined his status.The potential status of his children, his grandchildren, and, in the long run his lineage, were, however, dependent upon marriages .By continuously "marrying up" in the Tongan hierarchy, an individual's children, grandchildren, and lineage could slowly increa..,e their status .Just a.., a common er wa.., the product of a long line of younger brothers, so a chicf wa.., the product of a long lin e of elder siblings.The goal of Tongan marriag e wa.., to keep one's relati ves marrying elders.
There wa.., tremendous pressure to meet marriage payments and to make them extravaga nt, so that one's descendants would not lose rank throu gh a poor marria ge.Since prestige-goods were needed for these payments, and the payments made possible marriage with a socially elder individual, those who controlled prestige-goods controll ed individuals' abilities to socially reproduce themselves, a.., Kirch (1984: 241) make s clear: Kinship alliances linked the paramount lines with those o.lthe local mling chiefs in the core islands and outliers.Such alliances were confirmed by marriage relations,for which exotic prestige goods were vital.In turn, the outlying islands affirmed their inferior status and loyalty to the hau [Tu 'i Kanokupolu] and the Tu 'i Tonga through the tribute o.lthe 'inasi.Thus within the chiefdom there was a circular.flowo.fgoods, tribute inwards towards the paramounts, prestige goods outwards to the local chiefs.1 \llo nopolizatio n o.l the sources o.fjJrestige goods by the paramounts helped to secure their po wer over the system as a whole.
Tongan social elites controlled pres tige -goods ne eded by their subordinates to socially rep roduc e themselves.A dependency relationship thereby ex isted between elders and subordinates, that served to keep the elders in power and able to control younge r members of socie ty.This rela tionsh ip is diagrammed in Figure 3. discussed earlier, dependent in part on access to prestige -goods) flow to the non-elites, or Tua, on the right side of the diagram.The Tua, in return, support the nobility.The left side offigurc ,, sho,v-s the external trade in prestige-goods and, to some extent, marriage panncrs comb:tetcd by the Tu'is and their representatives.The Tt~is and other nobles comti:tet and manage this trade, as well as the local distribution of traded goods, and by doing so enhance their o,..,n noble status.fu addition, the nobles control ritual knowledge and its disscmination--anothcr arena for status enhancement and rcinlbrccmcnt.
Clearly a weak point in this S;'Stem is the foreign trade in prestige-goods.Ifbrokcn this weak point could easily lead to a legitimation crisis.Jftradc were cu:t ofl'.an important avenue for both status enhancement and, more importantly, the maintenance of mass Jovaltvwould be severed.Jfthc nobilitv could not find another avenue to reinforce mass , , , loyalty, perhaps through new, locally-manufactured prestige-goods, folfi!lingritual or supernatural needs, or some other mechanism, a legitimation crisis could ensue, perhaps causing a collapse of the political stmcturc.l'n the Tongan case this did not occur; however, during a period of civil war among ri-.-,:dTu'is and their followers, tics with Europeans that provided status-enhancing trade objects and supernatural knowledge ( or Christianity) were actively sought by the contending elites (sec, for example, Thompson lfl94: ,H4-Jl8; Vason lfllO: 75-flJ).l'mlccd, Latukcfa (1974: 66-67) suggests that the ultimate rctmi.fication of the Tongan Islands tmdcr Taufa'ahau ( commonly know11 as King George, in power .fromlfl,H-l89J) was achieved through a political strategywhich included his open acceptance of Christianity and support for new sources of statu s and power afforded by Christian missionaries (also see Thompson 1894: 346-352).
There are many cases which end differently ; that is, in legitimation crisis and collapse.One such case, I suggest, is the prehistoric Moundville polity of west-central Alabama.

The Moundville Polity
The Moundville polity represents one of the pinnacles of cultural evolution in eastern North America.Moundville is part of the larger Mississippian cultural system, which evolved beginning about AD. 900 in the Mississippi River valley and its major tributaries.The Mississippians developed large population centers, traded with peoples as distant as the Florida Gulf coast and the Great Plains, and construct ed the larg est prc-Columbian structures north of Mexico.Mississippian influenc e spread across eastern North America in the following century, and by AD. 1100 it was the predominant culture across the Southeast and Journal o.lWorld-Systems Research had fundamentall y influenced peoples in the Ohio River region , in the north ern Mississippi valley, and west into Iowa, Kansas, and Oklahoma.This influence continued even as some of the earliest Mississippian politics, includin g Moundville, declined beginning about AD. 1200.Although altered by time and restricted in their geographical range, Mississippian-like ways of life still existed in parts of the Southca..,t and the lower Mississippi River valley when Europeans first entered the North American continent.
The Moundville polity was located on the Black Warrior River in west-central Alabama, and was inhabit ed from ca.AD. 1000 to ca.AD. 1450.The site it..,cl f consists of 20 mounds covering an area of more than 40 hectares.The mounds surrounded a large plaza that formed the political or religious center for the surrounding region.Residential areas surrounded this plaza-mound complex, and the entire site was surrounded by a defensive palisade.At its peak, 3,000 or more people may have lived within the pali sade at Moundville.Surrounding Moundville arc a series of smaller villages and haml ets linked to the larger polity, and likely provided both material goods and labor to it (W clch 1991 ).
Socio-political organization in the Moundville polity was hierarchical, elites having clearly differential access to food resources and exotic goods (Welch 1991;Powcll 1988), and is most readily classified as a chiefdom (Peebles and Kus 1977).Welch (1991) tested four models of chiefdom political economy (redistribution, mobili zation, tributary, and prestige goods) against the archaeological record of the Moundville polity.He found that the prest ige goods model best fits the archaeological data.This type of political economy is quite similar to that described for Tonga, with lineage-based chiefs maintainin g control over status-displaying or enhanc ing goods obtained through foreign trade .They provide some of these goods to followers in return for their support, both political and material.Hence Moundvillc's elites would have been deeply involved in foreign trade and the production of goods for foreign trade, and in the creation and maintenance of symbol systems and ritual behaviors that required prestige goods (also sec Peregrine 1992Peregrine , 1995)).2 Paukctat (1994) recently examined the Cahokia polity, a contemporary of the Moundville polity located in East St. Louis, Illinois, to explore the ideology which sustained political hierarchy .He argued that Cahokian ideology initially wa<; ba<; cd on a patron-client relationship which developed through time into a divine-secular relati onship.Cahokian chicfa initially legitimated their authority through their generosity with material goods and by supporting craft specialists, but reinforced it through external alliances and claims to esoteric knowledge of distant peoples, places, technologies, and behaviors, and, ultimately, of the supernatural.A similar situation wa<; likely present in the Moundville polity.

Journa I of World-Systems Research
Political power in the Moundville polity, then, appears to hav e been ba<; cd up on the ability to control prestige-goods, legitimated in part throu gh a lin eage structure (Anderson 1994;Knight 1990;Peregrine 1992Peregrine , 1995;;Welch 1991 ), but also throu gh differential access to both exte rnal allies and supernatural power (Pauk ctat 1994).ln this way, the Moundville polity is quite similar to Tonga.Like Tonga, the organi zation of settlement suggests that a hierarchy of chicfa wa<; present in the Moundville polit y, with the pre-eminent chicflocatcd at the major center of Moundville itself, and lower level chicfa locat ed at minor centers and in outlying hamlets (Peebles and Kus 1977;Stcponaitis 1978).The political hierarchy its elf was probably organiz ed lik e a lin eage (Knight 1990), with individua l<; in each level both superior and socially "elder" to individuals in levels below them (DcPrattcr 1983: 100-10) .At the lowest level in the political hierarchy were localiz ed lineages, with elder males a<; their heads.
In this way, the schematic diagram of the Moundville polity in terms ofHabcnna<;'s model presented in Figure 4 is very similar to that for Tonga.Chicfa, in the center, distributed prestige-goods ( and likel y ritual knowledge) to their follow ers, who in return supported the chicfa through their loyalty .Chicfa also engaged in long-distanc e trade for prestige-goods, which they distributed internally, both reinforcin g their status and allowing them the opportun ity to enhanc e the status of follow ers .Collapse of the M.oundvillc Polity By A.D. 1500 Momdvillc ha<.l collapsed as a political entity.Moundville itself was abandoned and the socio-political hierarchy which it sccpportcd had disappeared.Christopher Peebles (1987aPeebles ( . 1987b) argues that Moundvi!Jc's collapse ~-as imtnccd. in part.by a constriction oftratlc in prcstigc-gomls.Data from the Moumlvillc site and the Lubbub Creek locality tlcmonstratc a marked tlcclinc in imp.,rtcd prcstigc-gomls beginning about A.D. 1400 (Peebles 1987a; I 987b).fignrc 5 presents the data this argument is based on.The roman 1mmcrals on the X-axis refer to the archaeological phases at Moumlvillc.The numbers on the Y-axis arc figures for the abundance of these gootls in dated burials.standardized by di .... itlingthc m11I1bcr of gomls by the mllilbcr of dated burials per phase (sec Stcponaitis 1991).

[Page II] Jm1ma l ()( W()rl<l-Syste11i~ R<~wY1rch
Clearly these imponed preciosities arc most ubuml.ant in the late J\:Ioundvillc I (AJ).Paul Welch (l99l: 194 J explains !he effect this decline in access to prcstigc-gootl~ might have had on the Mouml,illc polity: "if a large number oflcgitimizing symbols arc persistently or chronically unavailable, tl1c system of statn~s is likely to brcakdo--wn."Linking Wcklis statement to the diagram presented in figure 4 gives t~~ a clear picture of an ensuing collapse.On tl1c left side of the diagram, tl1c chiefs stecringoftTadc in prcstigc-gooll~ apparently faltered.•n1cir ability to enhance and/or reinJhrcc their o"wn slattts is tl1crcbyadvcrsclyafl'cctcd.More significantly, tl1cir ability to di~u-ilmte prcstige-gooll~ to followers on the right side of the diagram is affected.•nu~, in turn, afl'ccl, the commoner's loyally to the chief.~,leading, ultimately, to a legitimation crisis and political collapse.
\\11at would tlu~ collapse look like?I suggest it would look as if the top level of the political hierarchy was simply cnt ofl'--as if tl1c commoners simply "voted with tl1cir feet" and organiz.cdthemselves at a lower level of political integration.One wonld not expect to sec c,idcncc of conquest, one w,mkl not expect to sec c,idcncc of cn,ironmcn !al or economic calamity, and one w,mld not expect to sec a population decline.

Conclusions
Collapse is a fact of life for complex society, and while collapse may stem from environmental and economic problems, the point I hope I have made here is that an equally Journa I of World-Systems Research compelling source of crisis is the system of prestige which legitimates and supports the elite groups in the society.The prestige system is not epiphcnomcnal to the economic system, but a separate and essential clement of socio-political organization in all complex societies.Unfortunately scholars have tended to downplay this part of society and to cmpha.:;izc the subsistence economy.
In this paper I have presented one ca.:;c of collapse in which subsistence, population, and the environment seem to have had little influence.I suggest the Moundville polity collapsed through a legitimation crisis stemming from a constriction of the inter-regional trade in prestige-good.:;.The Moundville polity functioned a-; a prestige-good system, in which political authority in part rested on controlling access to objects required for social reproduction and status display.When those in positions of authority found it impossible to maintain regular access to these good-;, a crisis ensued which had systemic consequences.While this paper offers no explanation of why trade in these goods became constricted (but see Peebles 1987a, 1987b for some idea-;), this docs not discount the important conclusion that collapse, in this ca.:;e, was apparently not related to the subsistence economy or the environment; rather, it wa.:; a consequence of a failure in the prestige-economy and a transformation of longstanding patterns of int er-regional interact ion.It is the concept oflcgitimation crises a-; a potent force in social change which I hope I have successfully conveyed in this paper, but another, perhaps more subtle point I hope I have made is that crisis and collapse have many sources, and we must be willing to look beyond the subsist ence economy ifw c hope to understand them.
Legitimation crises, too, have potentially many sources, and it would be myopic to focus on transformations of int er-regional interactions a-; the sole one.Some area-; that mi ght prove valuable to investigate include the effect of new ideologies, the decay of existing ideologies, and conflicts stemming from succession to office.The effect of new ideologi es on the legitima tion of an exist ing political order ha-; already been seen in the case of Tonga, where Kin g George purposely used Christianity a-; an alternativ e ideology to separate himself from rival political figurcs--a strategy that ha-; been repeated in many parts of the world (sec, for exam ple, Ekholm (1972) on the Kongo Kingdom of west Africa and Axtell ( 1985) on various Indian politics of ca.:;tcrn North America).
The decay of existing ideologies a-; a potent force of cultural change is an idea that ha-; been put forward most lucidl y by the literary critic Fredric Jameson ( 1981; also sec Dowling 1984).Jameson argues that in every society there arc contradictions and conflic ts "strategies of containment" to ma.:;k these underlying conflicts (Jameson 1981: 10, 193, 269-271 ).These "strategies of containment" are constantly decaying and must be either actively reinforced or, if the decay ha.:; spread too far, transformed (Jameson 198 1: 97).It seems rea.:;onable to a.:;sumc that a failure to succcssfuly transform a decaying "strat egy of containment" could be a source oflcgitimation crisis.
Finally, crises stemming from succession to office have been the stuff of European history for centuries, but they have only recently been explored in non -W estern and prc-Modern societies.One of the best pieces of work carrying the problem.:; of succession to office into the ethnographic and archaeological literature comes from the archaeologist David G. Anderson (1994) who explores chiefdom "cycling," that is, th e cyclical consolidation and collapse of politics, in the Savannah River valley of Georgia.Anderson argues that the kin-ba.:;cdpolitical structure of chiefdoms contains an inherent contradition: a chiefs closest relatives arc generally his stron gest supporters and best allies, but they arc also his greatest potential rivals and successors.Given this contradiction, rivalry, particularly over succession to office, repe atedly builds to crisis and collapse (Anderson 1994: 330).
These , and other, potential sources of legitimation crisis help to explain the apparent instability of centralized political systems, regardless of their environm ent, their subsist enc e, their size.As Taint er (1988 : 1) put it : "civilizations are frag ile, imp erm anent things."I suggest, and it is yet another point I hope I have made in this paper, that centralized politics arc fragile and imperman ent precisely because there arc so many potential sourc es of crisis.Looking only at the cnvrionm cnt or subsistence econom y unnecessarily limits our understanding of these varied sources and we mu st , I argue, look beyond them ifw c hope to understand collapse.

Notes
(l) This paper was originally presented at the 1995 meetings of the American Anthropological Association, and I want to thank Nick Kardulias for inviting me to participate in tho se meetin gs.I also want to thank.Richard Blanton, C hristop her Chase-Dunn, John Clark, Gary Feinman, Stephen Kowalewski, and an anonymous revie we r for their substantive comments on th is paper.Not all of their ideas we re incorporated, but their interest is deeply (2) As defined in two seminal papers by Friedman and Rowlands (1977) and Frankenstein and Rowlands (1978), prestige-good system.., exist when important a..,pccts of political alliance or social reproduction arc tied to the consumption or exchange of specific exotic preciosities that can only be obtained through foreign trade.Frankenstein and Rowlands (1978;76) lucidly explain the economic logic of prestige-good systems; The spec[tic economic characteristics ofa prestige-goods system are dominated by the political advantage gained through exercising control over access to resources that can only be obtained through external trade.However, these are not the resources required for general material well-being or.for the mam!facture of tools and other utilitarian items.Instead, emphasis is placed on controlling the acquisition o.lwealth obje,cts needed in social transactions, and the payment o.lsocial debts.Groups are linked to each other through the competitive exchange o.lwealth objects as g[fts andfeasting in continuous cycles o.lstatus rivalry.Descent groups reproduce themselves in opposition to each other as their leaders compete for dominance through differential access to resources and labour power.
In prestige-good systems political power is ba..,cd on the control and manipula tion of exotic, imported preciosities.While elites in all social system.., disp lay and maintain their status in part through the control of exot ic goods and esoteric knowledg e (Frankenstein and Rowlands 1978;75;Hcl1rn 1979), in prestige-good systems these elite symbols arc needed by all members of the society for social reproduction (Ekholm 1972;Friedman and Rowlands 1977).Pr estige-goods arc used in these societies to fund social debts, such a.., bridcwcalth payment..,, initiation and fun erary fees, and punitive damages, and elites able to control access to these fungible exot ic goods gain political power in direct proportion to the demand for them (Frankenstein and Rowlands 1978: 76) .
Because of compction between elites for access to prestige-goods, these systems tend to be highly unstable (Fried man and Rowlands 1977;228).It seems common for prcstigcgood systems to repeat ed ly centralize and collapse, and henc e, they provide a uniquely valuable social form for examining these processes (Peregrin e 1992).In addition, I have argued that prestige-good systems can be taken a.., a spccial-ca..,c world-system, and arc particularly valuable social forms for exam ining the ris e and demis e of world-systems ( Peregrine 1991Peregrine , 1995Peregrine , 1996) ) .

Figure l is
Figure l is a diagram ofHabcnnas's (1973) conception of the Capitalist system.On th e far left is the economic system, the privately-owned enterprises which produce goods and 4): Tongatabu (100 square miles) ; Haapai (20 square miles); and Vavau (46 square [Page 4] Journa I of World-Systems Research miles), and most of the land area was given over to fields and stands of yam, taro, swe et potato, breadfruit, plantains, and coconuts (Goldman 1970: 281).
figure J. Revised conception of socio-political organization applied to Tonga.

[
Page 14] Journal of World-Systems Research appreciated.
The reproducti on of this system of generati onal ranking throughout the Tongan political hierarchy is obvious in the nature of political relationships : explains:Ranking of individuals within the Tonganfamily ... is the key to the organization o.f'Tongan society in every stratum .From the bottom to top and from top to bottom of the social ladder one general scheme o.tfamily organization prevails.As the Tu 'i Tonga is eiki (chief) to his younger brothers, so in every Tongan.family the older brother is chief to his younger brothers ... Relatively speaking, in every household there are chief, and commoners.
Page 6] Journal ofWorld-Systems ResearchPrestige-goods were a vital part of Tongan marriage alliances, and henc e, were vital to an individual's ability to marry well.Prestige-goods in Tonga were controlled at the highest level, by the Tu'i Tonga and the Tu'i Kanokupolu.They flow ed down the levels of hierarchy, and were a central force in maintaining hierarchical relationships.Kirch (1984:  238)makes this point very clear:If there wa..:; great inequality in the size of the weddi ng gifts, the group making the smaller donation wa..:; shamed and lost social prestige to the other group.In Tongan society marriage with an eldest son or daughter virtu ally determin ed the status of one's children, grandch ildr en, and lineage.AsGifford (1929: 20, 112)explains: a commoner is an individual who by virtue ofdescen t through a series ofyounger brothers has in the course o.lgenerations becomefitrt her andfitrther removed.from the patriarchal head stands unique among the indigenous Poly nesian chiefdoms for its extensive and regular long-distance exchange relations with socie ties beyond its own geographic and political borders.This long-distance exchange had political cons equences which were far greater than any immediate , utilitarian gain due to the importation o_fexotic material items.Long-distance exchange o_fchiefly spouses as well as o_f material items was fimdamentally a political strategy, and played a vital role in binding the core islands and outliers to the central polit y .Marriage payments, and the social prestige that went along with them, were described in detail by Gifford ( l 929: l 92-93):Pr eceding the day o_f the beginning o_f the [wedding] ceremony, the f athers o_f the bride and groom each assembled a large g{fi, including tapa, mats, and oil.The particular f ather not[fled all his relatives and all his w[fe's relatives o_lthe coming ceremony and asked.forcontributions .. .In distributing the presents, the bridegroom's.fatheror other o_ff lcial representative o.lhis people ... had in mind what each person had donated toward the present that had been given to the bride's people, and each got his original g[fi returned in double quantity.In accomplishing this return, the distributor o_fien stripped his own house o.lall its material property .{lhe should.fail to complete the traditional remuneration to all concerned, his unmarried sons and daughters and the proge ny o_fhis married children lost.face and might consequently.failto contract desirable marriages .. .A similar distribution was made o_lthe presents o_lthe bridegroom's people to the bride's people.
Journal of World-Systems Researchwhich are necessary for the society to continue but which would driv e the society apart if they were universally recognized.For this rea.:;on,societies develop what Jameson call.:; [Page 13]