7 The Role of Nomads in Core/Periphery Relations Thomas D. Hall OVERVIEW The extension of world-system theory to "pre- capitalist" settings necessarily raises ques- tions about the distinctiveness of the "modern world-system" with respect to "precapitalist" world-systems (See Chapter 1, and Wallerstein 1974a, 1974b, 1979, 1980, 1984, 1989, 1990; Abu-Lughod 1987, 1989, 1990; Chase-Dunn 1988, 1989, 1990a; Gailey 1985; Gledhill 1988; Hall 1986, 1989; Kohl 1985, 1987a, 1987b, 1988). An examination of the historical evolution of the role of nomads in core/periphery hierar- chies is one way to address these questions. Such a study has additional significances. First, it appears that sedentary social orga- nization emerged from an entirely nomadic context (Nissen 1988). Implicitly problematic here is whether it is possible to have a wholly nomadic core/periphery hierarchy. Because of this issue, the following discus- sion will be restricted to core/periphery hierarchies with sedentary components. I will return to wholly nomadic core/periphery hier- archies in the final section. Second, nomadic groups frequently play complex intermediary roles in the interactions among sedentary groups, especially states and core/periphery hierarchies. The significance of these roles in changes in core/periphery hierarchies needs to be examined. Third, each new world-system or core/periphery hierarchy is "restructured from" the remains of its predecessor(s) (Abu- Lughod 1990). This is one way in which spe- cific historical events shape general pat- terns. Fourth, following the work of both Janet Abu-Lughod (1989, 1990) and Thomas Barfield (1989), the rise and fall of the Mongol Empire played a major role in the collapse of the eastern circuits of the thir- teenth century Eurasian world-system. This collapse, in turn, initiated a restructuring from which the European "modern world-system" emerged. Based on Chapter 1, some additional com- ments are in order. First, there is no rea- son, a priori, to require that every core/periphery hierarchy have a semiperiphery, or that it be limited to only one tier between core and periphery. The extent of the hierar- chy should remain an empirical problem. Sec- ond, the precise boundaries of nonstate soci- eties (especially nomadic societies) are both theoretically and empirically problematic. On the one hand, nonstate societies -- based on both contemporary ethnographic studies and ethnohistory -- do not have precise borders, but tend to "fade away" in decreasingly impor- tant kin and other connections (Wolf 1982). On the other hand, just how much and what type of interaction between state and nonstate societies constitutes a significant connection also remains problematic. Trade in vital goods, trade in luxuries, trade in captives, alliances for frontier buffering, recruitment of nomads for armies, or endemic warfare with other nomads probably have different implica- tions for the historical evolution of core/periphery hierarchies. Third, the thing which is evolving is not a self-contained social unit, such as a "tribe," a "state," an "empire," or a "civilization," but a larger unit, which for want of a better term can be labeled a core/periphery hierarchy. This chapter contributes to a general discussion of the historical evolution of core/periphery hierarchies. I do this by examining a few cases of nomad - sedentary relations. The goal is not to produce defini- tive conclusions, but to uncover those aspects of nomad - sedentary relations that are espe- cially salient for the study of core/periphery hierarchies. Because of this goal it is useful to review a few other general issues concerning nomad - sedentary relations. Pristine States, Trade, Nomads, and Technology It seems reasonably clear that states, and hence civilizations, originated in interaction systems. Kohl argues that this is the case for the origin of the state: One purpose of this study of long-dis- tance trade in southwestern Asia was to show that even the earliest "pristine" example of state formation cannot be explained entirely as an internal pro- cess of social differentiation but must be viewed partly as the product of a "world-economy" at different levels of development which stretched at least from the Nile Valley and southeastern Europe in the west to Soviet Central Asia and the Indus Valley in the east" (Kohl 1978, p. 489). The evidence, as is always the case with archaeological materials, is incomplete, and subject to revision based on new discoveries. Nevertheless, the evidence is very persuasive. Trade between lowland Mesopotamia and highland Iranian plateau supplied goods used by state officials to support their claims to authority and legitimacy. The trade induced mutual dependency between the two areas, and consti- tuted a "world-economy." In a discussion of the contributions of archaeology to understanding the origin of the state, Gledhill also focuses on the interac- tion system: A regional focus seems of crucial impor- tance, since the fixation of many theo- ries with agricultural intensifications has tended to obscure the possible im- portance of the fact that the environ- mental settings of 'pristine' state formation cannot be defined purely in terms of aridity and its agricultural implications. The 'nuclear areas' of ancient civilization were also charac- terized by intense interactions between nomadic and sedentary populations, in- teractions which linked farms to fisher- men, gatherer-hunters and pastoralists" (Gledhill 1988, p. 23). Kohl (1978) urges archaeologists to make full use of information available in the historical record, using evidence on such things as: ... qualitatively novel developments in transportation, communication, and mili- tary technology associated with the domestication of the horse; the appear- ance of effective chariots; the spread or "democratization" of metal tools and weapons...(Kohl 1988, p. 30) Anthony (1986) discusses the domestication of horses in the context of state formation and regional interaction systems. A key point in these discussions is that with the domesti- cation of horses, transportation and communi- cation over land are greatly enhanced. Fur- thermore, simple iron making technology is very portable and easily diffused among nomads (Kohl 1987a, p. 22). Thus, relative "depen- dency" between nomad and sedentary groups might be reversed: horses might have enabled nomads to dominate sedentary peoples, and/or might have been the impetus among nomadic groups for technological innovations that subsequently spread to sedentary groups. With these issues and problems in mind it is now appropriate to turn to a brief examina- tion of several different types of nomad - sedentary interactions. NOMAD - CIVILIZATION RELATIONS China and Nomads The importance of nomads in Chinese history is widely acknowledged (Barfield 1989; Eberhard 1965, 1977; Elvin 1973; Kwanten 1979; Latti- more 1951, 1962c, 1980; Szynkiewicz 1989). This account, however, stresses only certain processes, drawing heavily from the work of Owen Lattimore and Thomas Barfield. As Chinese agriculturists spread to the steppe they adopted other styles of making a living (e.g., herding) that were better suited to the local ecology. Thus, pastoralism was not a case of "devolution," but an alternative survival strategy (Lattimore 1980). Those remaining in agricultural areas eventually developed a cavalry to fight nomads. When "China" con- sisted of separate kingdoms, many built walls as much to mark their own limits as to keep out the nomads. Since nomads produced little of interest or value to settled Chinese, nomads used threats of force to induce trade: trading and raiding were alternative means to the same ends. Indeed, some studies show that raiding corre- lated with changing conditions of trade (Szyn- kiewicz 1989, p. 154) and changing state stability (Barfield 1989). Chinese officials acquiesced to this trade as a way of control- ling nomads. The trade was primarily in luxury goods used by nomadic leaders to shore up and symbolize their power. This interac- tion fueled changes both in China and among nomads. It helped in incorporating new lands. In times of state decline, nomadic leaders sometimes served as protectors of beleaguered areas. In times of state ascendance, unified Chinese response promoted wider unity among nomads. Nomads were as often a source of change as a receiver (Lattimore 1980, Barfield 1989). Barfield (1989) analyzes the interconnec- tions between the rise and fall of the Chinese empire and various steppe empires (see espe- cially his chart p. 13). The two are inti- mately connected via the organizational system of the Chinese empire and the varieties of steppe politics and organization. A key feature in Barfield's analysis is the distinc- tion between inner and outer frontier strate- gies. The outer frontier strategy is the more familiar. In this strategy a dominant steppe leader uses violent attacks to terrify the court of the sedentary empire, alternates war and peace to successively raise tribute pay- ments, and assiduously avoids taking over Chinese lands and the necessarily intendant entanglements in Chinese court politics (p. 49). The inner strategy is more subtle, and one that develops when a steppe confederation begins to disintegrate. Some contending steppe faction leader, typically of a weaker faction, seeks alliance with Chinese officials against his rivals. The Chinese officials acquiesce, since typically they favor using "barbarians against barbarians." The steppe faction sometimes uses the Chinese military to aid in the defeat of his rivals, and uses favor at Chinese court to sever tribute flow between the Chinese and his rivals. Typical- ly, this leads to success on the part of the Chinese allied faction. Once dominant, the steppe leader can either use the new power base to unify the steppe and return to an outer frontier strategy or leave the steppe politically fragmented and seek to dominate a local region, monopolizing tribute flow (p. 63). This oscillation explains the rather long cycles of nomad - Chinese relations (Barfield 1989, p. 13). In particular, it explains the correlation of strong steppe polities with a strong Chinese empire. Only when the empire is strong can it be steadily "milked" via an outer frontier strategy. When the empire is weak, steppe leaders tend to favor an inner frontier strategy, making alliances with local "war lords." The Mongols used both strate- gies, but with their own peculiar twists. The Mongol Empire Analysis of the relations between Mongols and sedentary states is doubly difficult. First, the entire Central Asian field of action must be examined (Morgan 1986; Allsen 1987; Bar- field 1989; Kwanten 1979). This is a diffi- cult task because of the way records are made and preserved. The interpretation of docu- ments -- nearly always written by and from the point of view of sedentary state officials -- requires detailed knowledge of many local histories. Second, the Mongols were peculiar in many ways as a steppe empire (Barfield 1989). According to Abu-Lughod (1989, 1990) they played a pivotal role in the thirteenth century collapse of the Eurasian world-system and in the consequent rise of the European world-system. Mongol success can be attributed to factors and processes occurring at different levels simultaneously (Saunders 1971; Morgan 1986; Lindner 1981, 1982). First, the states in western Asia were weak and thus vulnerable to attack and conquest: This was the first time that a major nomadic power direct from the Chinese frontier had invaded the sedentary states of the west. The outer frontier strategy of devastation and terror wreaked havoc with the more fragile ecology of the region. China might replace large population losses within a relatively short period, but here the damage was more long lasting. Cities whose populations numbered in the hun- dreds of thousands were completely de- stroyed. Irrigation systems were ruined, severely hampering economic recovery. ... Nomads who had previously entered southwestern Asia from the steppe had always attempted, usually successfully, to found new dynasties and become rul- ers. The Mongols with their heritage from the Chinese frontier refused to take administrative responsibilities (Barfield 1989, pp. 201-202). This latter, of course, is the familiar outer frontier strategy, but in this case applied against states which were politically weaker and ecologically more precariously positioned than was typical for China. Second, continual warfare made a client relation with the Mongols an attractive "bar- gain" given the alternatives: Those areas which accepted the new situ- ation (Manchuria, Korea, Uighar oa- ses)... avoided destructive campaigns by the Mongols and retained their own lead- ers. Those areas which rejected the Mongol peace terms or reneged on previ- ous agreements (Chin China, western Turkestan, and the Tangut kingdom) be- came the scenes of numerous campaigns that wiped out much of their population and productivity. In Chinggis' lifetime wars of destruction were aimed at lead- ers who violated previously agreed peace terms. These campaigns were so devas- tating that they led to the overthrow of the ruling dynasties and, by default, their direct incorporation into the Mongol Empire (p. 200). One of the peculiarities of the Mongol empire was a direct result of this policy (rooted in Chinggis' fierce demands for loyalty), namely the conquest of China and the founding of a dynasty due to overly vigorous pursuit of an outer frontier strategy of terror. This is one of the ways in which the Mongols differed from all previous Central Asian nomadic groups. They over-pursued the outer frontier strategy. In the west they destroyed some states or were forced to directly incorporate and administer others. In the east they were ultimately trapped by their own vigor and success into founding a new Chinese dynasty. A third factor in Mongol success was con- tinued presence of pastoral nomads who main- tained a flexible tribal/kin social structure which allowed them to recruit first other nomads, later even sedentary groups, into a larger and larger machine for conquest. Fourth was the presence of several leaders who were able to manage astutely the drive for conquest and plunder, with the needs of admin- istration. Specifically, Mongol leaders were able, for some time, to maintain a sufficient volume of plunder and tribute to insure loyal- ty of tribes that might otherwise be inclined to leave the confederation. In short they perfected the outer frontier strategy of "milking" sedentary states (with some excep- tions noted above). Fifth, superior logistic ability of pastoral nomads in communication, transportation, and mobility was key to con- quest of large territories, and even of seden- tary states. This superiority was rooted in the pastoral way of life: availability of horses, intimate knowledge of geography, and ability to move their entire society (families and their resource base, their herds) with them. This same superiority was also a key to the collapse of the system. All the great Khans -- if only temporarily -- have been able to put these processes in operation (e.g., Allsen 1987). First, capital- izing on the inclusive nature of pastoral tribes, conquered groups were given an honor- able option of joining the group (the alterna- tive was being put to the sword). This worked well with other pastoralist and poorly with sedentary peoples (unless they wanted to become pastoralists). Second, as in modern pyramid schemes, as long as the system kept expanding, new recruits (groups) could gain both status and wealth. Continued expansion also alleviated, if it did not solve complete- ly, the problem of revenue by a constant inflow of booty. As long as this was success- ful, it distracted conservative elements from social changes which flowed from this strate- gy, and minimized factional rivalry. Superior communications and mobility were used to amass troops and overcome enemies. However, this strategy was inherently unstable, that is, temporary. The instability stems from two closely intertwined sources: political and technolog- ical. The political problem is the orderly succession of rulers. A "big man" comes to power on the basis of his personal skills, not the least of which is alliance-building. Among Central Asian pastoralists, this is typically based on military prowess. Hence, succession of rulership necessarily entails armed conflict. When there was only one son who had distinguished himself in battle, conflict was merely postponed for a generation until either there was no such son, or there were several. For the Mongols these problems were exacerbated by the competing, and at time conflicting, principles of lateral (older to younger brothers) and lineal (father to son) succession. The lack of clear priorities inevitably led to justification of succession by arms. Institutionalization of succession would have undermined the very basis of lead- ership. Thus, it is not only that the Mongol Empire did not institutionalize political control as Eisenstadt (1963) argues, but also that Mongols could not institutionalize lead- ership and remain Mongols. The same problems inhere in political control and revenue gar- nering. A key to Mongol success was communication and mobility of men and resources. These technological factors also contributed to instability because they made it impossible for any central leader to monopolize control of strategic resources as a means of coercing compliance. Hence, there was no way to insure that revenue collectors would forward revenue to the leader. The same features that make tribes inclu- sive also make them divisive; what can be built quickly can equally quickly disinte- grate. The material basis of this situation is the adaptation to plains/steppe environ- ment. This is the underlying limit of pasto- ralist expansion. They cannot effectively control territory beyond the plains/steppe without giving up their lifestyle, thus the edge of the steppe remains a permanent fron- tier (Lindner 1982; McNeill 1964). Conversely, sedentary states could not conquer nomads -- except by sedentarizing them. They could control them by a combina- tion of constructing barriers and employing highly mobile troops, who could essentially beat the pastoralists at their own game -- decisive hit-and-run victories (Lattimore 1962a, p. 485). Thus, Central Asian pastora- lists, especially the Mongols, could build huge empires, but could not maintain them. Conversely, the Chinese could manipulate, but never conquer, their nomadic adversaries. According to several writers (Allsen 1987; Barfield 1989; Lattimore 1951, 1962d; Morgan 1986), this accounts for the convoluted quali- ty of Chinese histories of these events. Chroniclers had to warn princes and emperors of the inherent impossibility of conquest, while never admitting that the "son of heaven" was not all powerful -- a task that makes contemporary American "spin doctors" appear to be rank amateurs. The Ottoman Empire The formation of the Ottoman empire is of interest because it was built by the transfor- mation of nomadic pastoralists into sedentary farmer-soldiers. According to Lindner (- 1983), nomads, particularly of the "tribe" of Osman, played a vital role in the founding of the Ottoman Empire. It was precisely the fluid, multi-cultural aspects of tribal orga- nization that made nomadic "tribalism" an effective model for building a state. Once built, the needs of the new state led to the oppression and destruction of nomads by con- version into sedentary peasants. The erstwhile nomads, now rulers of a large state, were compelled to sedentarize the remaining nomads. This was done by first shifting obligations of support of the state from contributions of manpower in fighting (nomad tradition) to taxes in kind and/or money, and later by levying taxes in early spring immediately after lambing time, count- ing all animals as adult sheep, and therefore taxable. This contrasts with the Mongol custom which exempted small herds so that they could continue to function (p. 57). In other words, the goal of Ottoman tax policy was to undermine nomad economy. Two sets of state needs impelled this policy. First, the state rulers sought to avoid the "state within state" arrangement implied in tribal loyalty, in which the local chief administered his tribe with considerable autonomy. This required clear tribal bound- aries, but tribal boundaries are highly perme- able, membership changing with shifting eco- nomic and political conditions. The second set of reasons behind sedentari- zation were military. As the nature of war- fare changed, especially as gunpowder came into use, the demand for mounted archers decreased, while the demand for infantry increased. Thus, nomads per se had less to offer the state in compensation for their political administrative liabilities. Addi- tionally, more people -- and therefore more fighting men -- could be supported on the same amount of territory in a sedentary, agricul- tural adaptation than by nomadic pastoralism. The volatility of pastoral production due to disease and weather also makes sedentary production more certain. While the Ottoman Empire had clear nomadic origins, it became and remained a sedentary state which eventually destroyed its own nomadic foundations. In contrast, the Mongol Empire never made a successful transition to a sedentary state. The tribe of Osman did succeed, but only by ceasing to be nomads. Spain, America and Los Indios B rbaros The comparison of Spanish and American treat- ments of various nomadic groups inhabiting what is now the American Southwest, what was long the northwest of New Spain, is useful in several ways. First, the region is the same in both cases. Second, the region was a frontier for both states -- a "periphery of a periphery" (Weber 1982). Third, America and Spain contrast a rising capitalist state with the earliest phases of the capitalist world- system, more akin to an empire (Cipollo 1970, Doyle 1986, Eisenstadt 1963, 1967). Spanish explorers first entered the region sometime in the 1530s, slightly over a decade after Cort‚z conquered Mexico. The region was not formally colonized until 1598. That colony collapsed, due to the Pueblo Revolt, in 1680, and was re-established in the early 1690s. Thereafter it remained a tenuous, but relatively thriving colonial outpost whose fortunes waxed and waned with those of New Spain and the Spanish Empire in general. Both before the Pueblo Revolt, and for the half century or so after the reconquest, the region was marked by a state of endemic warfare with surrounding nomadic groups. In the early eighteenth century this warfare became so intense that the sedentary Pueblos, and erst- while ousters of the Spaniards, formed a symbiotic alliance with them. The late eigh- teenth century was marked by considerable local population growth and relative prosperi- ty. In 1786 a lasting peace was established with several closely linked bands of Shoshoni speakers who became widely known as Coman- ches. This peace between Comanches and Spaniards lasted well into the American era. The Mexican interregnum disrupted this pattern and renewed the pattern of conflict with nomads common in the early phases of the colony. Throughout the Spanish era, warfare with nomadic groups rose and fell with changes in the trade in Indian captives, with the need for local governors to impress the viceroy with their success in subduing "los indios b rbaros" or their desire for more money and troops, and with viceregal and crown concerns for protection of the borders of New Spain from rival European powers. It is the latter concerns that were the driving force behind first maintenance, then re-founding of the colony (Bolton 1929). These concerns likewise shaped policies toward nomadic groups, at once a nuisance along the frontiers, yet simulta- neously -- especially after the Comanche peace -- a singularly effective "border patrol" for scouting and controlling movement of European rivals. The Bourbon reforms, instituted in New Spain in the late eighteenth century, were intended to increase state efficiency. These policies led to a general increase in prosper- ity throughout the Empire (Lang 1975) and in New Mexico. Subsequent Indian policy aimed at: (1) pursuit of peace in order to increase revenues; (2) use of frontier bands as buffers against foreign intrusion; and (3) lowering the cost of administration and defense. These goals gave rise to four strategies to control nomadic groups. First was the use of "gifts" to engender dependency upon Spaniards. Sec- ond, divide-and-conquer strategies were used to pit one group against other hostile groups. Third, nomadic groups were pressed to form more centralized political structures, which gave rise to the "tribes" we know today. Fourth, the frontier provinces were reorga- nized, a line of forts was constructed, and highly mobile "flying companies" were used to control hostile nomads (Moorhead 1968, 1975; Griffen 1983a, 1983b, 1984, 1985; Thrapp 1967). American influence in the Southwest dates to the opening of the Santa Fe Trail in 1821. As trade increased, New Mexico became more strongly oriented toward the United States. The Mexican - American war (1846-1848) result- ed in the annexation of the northern half of Mexico to the United States in 1848. Califor- nia and Texas were the major goals of the conquest. New Mexico (which then included modern Arizona) was primarily a "land bridge" between California and Texas. Fighting with nomadic groups increased until the American Civil War (1860-64), spurred by increasing traffic through nomad lands and increased trade or encroachment on hunting territories. After the Civil War the American state began a major effort to control nomadic groups throug- hout the west (Utley 1984). The Comanche bands became a major internal nuisance instead of a buffer-border patrol. A major effort was mounted to force them, and subsequently the Apache bands, onto very limited reservation territories. Comanches declined from "Lords of the South Plains" (Wallace and Hoebel 1952) to a handful of reservation dwellers (from between 20,000 and 30,000 in the early nineteenth century to between 1,000 and 2,000 in the late nineteenth century [Hall 1986, 1989]). Their territory shrank from the western half of Texas to a small reservation in Oklahoma (Indian Territo- ry). They had become a barrier to internal trade in the U.S., and their major resource, the buffalo, had become very scarce. Apache groups fared better. Centuries of a "raiding mode of production" had perfected their survival techniques. Low interest in New Mexico and Arizona led to considerably lower pressure on Apaches. A combination of eastern sentiments for the "vanishing red man," and lucrative contracts to be had for supplying first the army and later reserva- tions, prevented complete genocide. Thus, the American state succeeded in accomplishing in less than fifty years what Spanish administrators had not been able to accomplish in nearly two hundred and fifty years -- total sedentarization of nomadic groups. DISCUSSION, CONCLUSIONS, AND SPECULATIONS What, then do these brief cases suggest about the roles of nomads in the historical evolu- tion of core/periphery hierarchies? I begin with some general remarks, then turn to some more specific conclusions, and end with dis- cussion and speculation about further re- search. Nomads, States, and Core/Periphery Hierarchies While it is clear that I have not sampled the entire range of nomad - sedentary relations (for instance, there is no examination of wholly nomadic settings as a limiting case, nor of very early, pre-horse, nomad - seden- tary relations), still, a tentative, schematic account of the role of nomads in the histori- cal evolution of core/periphery hierarchies is discernible. This sketch is not intended as a definitive statement, but as an elaborate working hypothesis useful for guiding further research. Once states domesticated horses and mas- tered the production and use of iron, they more commonly took the core role in core/periphery hierarchies, giving rise to other states, and unleashing interstate compe- tition (Chapters 1, 4; Chase-Dunn 1989, 1990a; Wilkinson 1983, 1987, 1988a, 1988b). States also altered the other elements in core/periphery hierarchies. As core states sought to increase their wealth -- not infre- quently to maintain and enhance their internal control over subjugated portions of their populations -- they began to expand territori- ally. Sometimes this expansion was in the direct pursuit of wealth through conquest, other times through enlarging the quantity and types of trade goods they could acquire. Expansion necessarily led to contact with, and frequently incorporation of, new groups into a core/periphery hierarchy. The incorporation process varies in both speed and intensity, can be reversed to some degree (Hall 1983, 1986, 1989), usually pro- motes social change among incorporated groups, and typically elicits resistance to incorpora- tion (Hall 1989; Gailey and Patterson 1987, 1988). This type of change is precisely what is meant by the statement that "civilization gave birth to barbarism" (Lattimore 1962d, p. 99, cited in Wallerstein 1974b, p. 98). Typically, but not exclusively, such "barbari- ans" occupied peripheral or semiperipheral positions in core/periphery hierarchies. For China, they were nomadic pastoralists; for the early Sumerian states they were other states (see Chase-Dunn 1988). Even the pre- ceding brief account of Central Asian nomads indicates that, as with early marcher states, such "barbarians", once engendered by adjacent states, could become relatively autonomous sources of social change. When the degree of technological difference between nomad and sedentary populations was low, as it was in the early stages of the agrarian era (3,000 to 1,000 B.C., +/- 1,000 years), which group would dominate a core/periphery hierarchy was an open issue. Stationary targets are easier to attack than mobile ones, thus nomads could readily defeat sedentary groups in battle. Furthermore, sedentary agriculturists could either be forced to pay tribute or be moved and used as slaves elsewhere. Nomads, however, are more difficult to defeat. Additionally, their wealth -- usually in the form of animals -- is mobile, and hence they can avoid paying trib- ute. Conversely, they are inimical to seden- tary life and difficult to make into slaves. However, nomads can be induced into trade relations. Sedentary peoples typically have some surplus agricultural products, but a shortage of meat and/or horses. Nomads typi- cally have a surplus of meat in the form of animals, but a shortage of vegetable resourc- es. Thus, there is opportunity for trade. Where the terms of trade were not suitable, or where sedentary people were unwilling to trade, raiding became an alternative form of exchange. The Mongols perfected the threat (and use) of violence as a tool in trade relations. Where large distances separated different states or core/periphery hierarchies, nomads could become middlemen in trade based on their superior mobility, and superior knowledge of the "uncivilized" territory, as was the case for Central Asian nomads and for Comanches under the Spanish empire. Where state traders were sufficiently strong they could conduct trade directly, but were subject to raids by nomads, or they could engage nomads as armed protectors. Nomads were strategically and tactically well positioned to extract "protec- tion rent" (Lane 1973; Chase-Dunn 1989). Even in the case of strong states, nomads could maintain a considerable degree of autonomy -- or even domination -- if they could play one state against the other (Kohl 1987b). The many ways in which state systems deal with nomads, in turn, shape their own administra- tive, political, and trade systems, as was clearly the case for both China and the Span- ish empire. Conclusions Several conclusions emerge from this discus- sion. First, formal systems (bureaucracies, in short, states) have a very difficult time dealing with informal, or acephalous, societ- ies. This is true whether it is the Aztecs dealing with the "Chichimecas" (McGuire 1980, 1986; Mathien and McGuire 1986), the Romans dealing with Gallic and Germanic "tribes" (Luttwak 1976; Dyson 1985), the Chinese deal- ing with Mongols or Turks, the Byzantine empire dealing with Turks and others (Lindner 1983), the Spanish empire dealing with "los Indios b rbaros," or the United States dealing with various Native American "tribes" (Hall 1989; Utley 1984). Second, nomads occasionally conquer states, but cannot rule them (for long) without becom- ing sedentarized. Lattimore (1962b, p.508) has described the problem cogently: As the Chinese pithily expressed it long ago, an empire could be conquered on horseback, but not ruled from horseback; civil servants more sophisticated than barbarian warriors were needed to ex- tract a regular flow of taxes and trib- ute from the civilized part of the em- pire, they could be recruited only among the upper classes of the conquered civi- lized people, and they and their fami- lies had to be protected and allowed to perpetuate themselves. It was therefore impossible to fuse completely the bar- barian and the civilized parts of the structure of empire, and impossible also to make the barbarian conquerors as a whole a new upper class imposed on the conquered society. Comparison of the Chinese, Mongol, and Ottoman empires underscores this argument: empires require a sedentary base. Clearly, pastoral empires or core/periphery hierarchies are ephemeral. The Mongol empire is distinctive precisely because of its success in thwarting this general tendency. Yet, even the Mongol empire built a capital city, Karakorum. I must add a caution based on Barfield's account. Under special circumstances nomads can extract some tribute from sedentary states. The outer frontier strategy of Cen- tral Asian nomads is an example of this. The outer frontier strategy was not a monopoly of the Mongols. Nomadic Indians in the American Southwest, especially Comanches, were also adept at extracting tribute from the Spanish Empire. There is a significant difference, however, between extraction of a minimal "tribute" which constitutes a nuisance and large scale extraction familiar in tributary empires. Assessment as to whether "gifts" from sedentary rulers to nomadic clients consti- tutes tribute or "trinkets" is a complicated matter. Relative worth is a significant component of the assessment. What may have been "mere trinkets" to the state, may have been vital prestige goods to the nomads. Evaluation of relative worth is not facilitat- ed by the nearly universal tendency of contem- porary writers (almost all from sedentary societies themselves) to gloss such gifts as trinkets, even when the cost of those gifts was bankrupting the state treasury. A third, yet abstract, conclusion is that comparisons of seemingly similar social forms, such as "nomads," "tribes," or "chiefdoms" that span long periods (centuries or millenia) must be executed with extreme care. The qualities of these types of social groups have shifted through time as core/periphery hierar- chical systems have evolved. Frequently, such social structures cannot be understood apart from their place in larger social systems and social processes. Similar manifest forms may belie very different processes of formation, some of which may have a significant role in shaping further changes. Still, ethnohistori- cal materials, if used with due caution, can shed considerable light on the processes of change, as Kohl (1978, p. 475) suggests. Fourth, the qualitative aspects of nomad - sedentary relations have shifted through time. The effect of nomads on states has lessened through time, while the effect of states on nomads has strengthened through time. This trend is due, at least in part, to a growing technological gulf between nomadic and seden- tary populations. As Kohl (1987b, p. 22-23) puts it: Economic development and dependency were not linked phenomena during the Bronze Age in the manner postulated by contem- porary critical theory for -- to para- phrase their terminology -- the develop- ment of underdevelopment in the Bronze Age was sharply constrained or itself underdeveloped. Critical technologies, such as metal working, could diffuse relatively easily and new means of tran- sportation and sources of power, such as horses, could be raised in peripheral zones and radically restructure this ancient world- system. Thus, while relative dependency may be prob- lematic for the Bronze Age, it becomes less problematic after the appearance of mounted pastoralists in Central Asia, and entirely clear by the seventeenth century of the Chris- tian era. Fifth, the preceding conclusions underscore the argument that world-system theory requires considerable elaboration and modification to provide a framework for examining precapital- ist core/periphery hierarchies. In doing so, several supposed distinctive features of the modern world-economy must be seen as somewhat different repetitions of older patterns and processes. The perennial problem of changes in intensity shading into qualitative changes remains. One aspect, however, of the "modern world-system" that is clearly distinctive is the capability, through twentieth century technological means, to eliminate nomads as a source of social change. These conclusions easily give rise to as many questions as they answer, and suggest a number of continuing research problems. Speculations First, the preceding evidence and analysis sheds precious little light on the role and significance of wholly nomadic core/periphery hierarchies -- if indeed, such can even be said to exist. Further research, in this case, highly speculative based on archaeologi- cal evidence, will be required to discover if anything like a core/periphery hierarchy existed in a wholly nomadic setting. Analysis of such a core/periphery hierarchy would compound and conflate the problems of perme- able group boundaries and fluid group member- ships with ephemeral geographic boundaries. Since, by definition, the wealth of nomads is portable, wholly nomadic core/periphery hier- archies probably would have been very fragile, and marked by a very shallow gradient of hierarchy. Hence, they would leave little direct physical evidence of their existence. Since the formation of states so dramati- cally altered the entire social field, great caution must be used in using any post-state nomadic societies as bases for speculation about pre-state nomadic societies. This is not to say that the task is insurmountable, but that it must be approached with great caution. The evidence for, or against, the existence of wholly nomadic core/periphery hierarchies will be quite thin. Methodological and evidential problems notwithstanding, this suggests several inter- esting research questions. Were the forms of states that we know of from the historical record the only solution to these problems? Were there others? Could there have been others if the extant ones had not preempted the field? Do different solutions to the ephemeral quality of wholly nomadic settings give rise to different types of core/periphery hierarchies? Second, the profound social changes that accompany changes in transportation technology suggest parallels between the nomad - seden- tary distinction and the overland - maritime distinction elaborated by Fox (1971), Hochberg (1985), Genovese and Hochberg (1989), Tilly (1989), and Fox (1989). Horses, like sea transport, enhance communication over broad, trackless areas, and as with maritime powers, do not readily facilitate amassing permanent, large armies. Horses and ships do, however, permit sizable temporary amassings for rapid attacks. A key feature of nomadic empires, like maritime states, is their relative fra- gility and instability as compared with agrar- ian states and empires, and the typical depen- dence of the former on the latter -- albeit with many exceptions. The major differences seem to be that horses can be produced on the trackless area, where ships are built on land, while ships facilitate bulk trade, horses (and camels and mules) are better used for (rela- tive) luxury goods. How important were nomads in the diffusion of ideas and technologies? Clearly, they played a vital role in the diffusion of diseases, transmitting the plagues to both Europe and China (McNeill 1976). Third, the inner and outer frontier strate- gies can be seen as precapitalist analogues of strategies for advancement used by semi- peripheral states in the "modern world-system" -- right-wing, authoritarian regimes in alli- ance with capitalists and left-wing, class- based anti-systemic rebellions (Chase-Dunn 1990b). In both settings, the semiperipheral players can be either sources of stability, or instability. This parallel warrants further examination. Is the semiperipheral social position always a locus of change? Does the role, or set of possible roles, of the inter- mediate tiers of a core/periphery hierarchy change systematically with the type of core/periphery hierarchy? Do the roles change with the type of social unit (sedentary, nomad, state, nonstate) occupying the posi- tion? Whatever the answers to these questions, the evidence appears to be compelling that the study of social change must focus on core/periphery hierarchies, or at least the intersocietal context, and should not focus exclusively on various individual components (states, tribes, etc.) of the hierarchy. Within the system or hierarchy, peripheral and especially semiperipheral units play highly variable roles, which are at times crucial to processes of change. Specifically, nomadic groups have had major influences on the course of social change. The habit of pushing bar- barians beyond the pale -- a military and political goal that was seldom achieved by any empire for any significant period of time -- is for intellectual pursuits at best mislead- ing, and potentially disastrous for under- standing the processes and variations of historical evolution. NOTES