In this chapter we do several things. We begin by summarizing the conceptual apparatus we have developed for a theoretical research program that compares world-systems. A more extended treatment may be found in our Rise and Demise: Comparing World-Systems (1997). We describe our approach to the problem of spatially bounding world-systems, summarizing what we think is one of our more important contributions: conceiving of world-systems as having four, typically nested, networks of interaction. We also outline our approach to core/periphery relations. Then we discuss three recurrent processes: incorporation of new areas or peoples; merging of formerly autonomous world-systems; and the phenomenon of the upward mobility and transformational innovations by semiperipheral actors. Then we briefly sketch our explanation for the evolution of small-scale egalitarian world-systems into the single hierarchical global system of today.

The remainder of the chapter is largely devoted to analyzing the cyclical processes of world-systems: the pulsation of interaction networks, and the rise and fall of central polities. We report on a puzzling empirical finding: the synchronicity of urban and empire growth and decline phases at the two ends of the Eurasian landmass over the last two millennia. We conclude with a reprise on the issue of transformation of world-systems, a summary of the findings in Rise and Demise, raise more unanswered questions, and speculate about the future of the world-system.

Our main conclusion is that world-systems, properly conceptualized and bounded, are the fundamental unit of analysis of social change. Put simply, it is not possible to understand how humans got, in a mere 12,000 years, from living in small groups of 50 to 100 individuals to today's global system composed of states of up to a billion (1,000,000,000) individuals without attending closely to inter-societal interactions. Studies that look only at single groups, societies, or states are doomed to misunderstand social change, because much of it originates in the interactive and structured relations among these units.

Such structural arguments often raise questions about where are the actors? What about agency in this structure? To paraphrase Marx's old chestnut, humans make their own history, but not any way they please. In short, they are constrained by existing structures. We also note that macro structures and processes have emergent relations to more localized processes and structures. We argue that it is

[86]

only by understanding how and why structures change, that we can determine where human action can have the greatest effect in transforming those structures. Bluntly, structural analysis does not obviate human action. Rather, it makes it possible. We return to this issue in our concluding speculations about the future of the world-system. Thus, our analysis does not ignore human action. Rather, we seek to understand structural processes in order to know better precisely where, when and how human action can most fruitfully be employed to build a more humane world (see Boswell and Chase-Dunn 2000; Chase-Dunn 1996).

We have described elsewhere (Chase-Dunn 1992; Hall and Chase-Dunn 1993, 1994) how those who study precapitalist world-systems fall along two continua. First is the transformalionist-continualionist continuum. At one pole are those who see essentially the same processes repeating through time with no major qualitative breaks. At the other pole, where we place ourselves, are those who see major transformations in the way world-systems function. Many scholars have argued that the world is currently experiencing another major transformation. To save the suspense, we do not think so, although we do see the possibility of a major transformation in the next century or two.

The second continuum is one we inelegantly call lumpers vs. splitters — those who see large systems with similar features versus those who see small systems that are qualitatively different. Lumpers are often also continuationists, while splitters are more likely to be transformationists, but these categories are by no means totally overlapping. We are, not surprisingly, closer to the splitter pole than the lumper pole.

Because we wish to study transformations, we maximize the range of possible cases by including all sedentary human groups that have existed on Earth over the last twelve thousand years.' To facilitate broad comparisons we define world-systems as intersocietal networks in which the interactions (e.g. trade, warfare, intermarriage, information) are important for the reproduction of the internal structures of the composite units and importantly affect changes that occur in these local structures. Because the boundaries of non-state social groups (e.g. 'bands' or 'tribes') are often empirically fuzzy, and because the term 'society' can too easily imply a clearly bounded social group, we use the term 'composite units' in our definition.

Because the ability to produce more than is needed for immediate consumption plays important roles in social reproduction and social change² in all human groups, we take the 'mode of accumulation' as a fundamental characteristic of any world-system. Even so-called egalitarian (classless) groups organized accumulation by storing foodstuff's and socially regulating the use of resources. We define mode of accumulation as the deep structural logic of production, distribution, exchange, and accumulation. That is, the reproduction of social structures and cyclical processes occurs by means of certain typical forms of integration and control for any specific mode of accumulation. These constitute a logic of development. While our

[87]

position is avowedly materialist, we do leave open the possibility that non-material factors may sometimes initiate social change. We prefer 'mode of accumulation' to 'mode of production' because we do not want to restrict our focus solely to the analysis of production. Instead, we want to focus on the institutional mechanisms by which labor is mobilized and social reproduction is accomplished.

We begin with a heuristic typology drawn from the works of Amin (1980, 1991) and Wolf (1982), supplemented by Polanyi (1944, 1977).³ We distinguish among four classes of systemic logics:

1 Kin-based modes a/accumulation, in which social labor, distribution, and collective accumulation is mobilized by means of normative integration based on consensual definitions of value, obligations, affective ties, kinship networks, and rules of conduct - a moral order.

2 Tributary modes, in which accumulation of surplus product is mobilized by means of politically institutionalized coercion based on formally organized military power and codified law.

3 Capitalist modes, in which land, labor, wealth, and goods are commodified and strongly exposed to the forces of price-setting markets and accumulation occurs primarily through the production of commodities using commodified labor.

4 Socialist modes, an hypothetical class of logics in which major policy, investment and allocation decisions are controlled democratically by the people they affect according to a logic of collective rationality.

The main features of modes of accumulation that can be used as empirical indicators are forms of exchange (gift-giving, state-administered exchange, market trade) and forms of control that are employed to mobilize social labor and/or to extract surplus product (normative regulation, serfdom, slavery, taxation, tribute, wage-labor). Different modes of accumulation are often present within the same system. Furthermore, some forms of exchange and control have elements of more than one mode.

We do not claim that modes of accumulation are features that permeate entire world-systems. Rather, they may exist at any level of a system (see Chase-Dunn 1998:335-7). For instance, the broad category of tributary mode includes both centralized and decentralized political forms that rely on coercion to mobilize labor and extract taxation, tribute, or rent. Thus, feudalism is a sub-type of the tributary mode, one of its most decentralized forms. Similarly, the so-called 'Asiatic' form, in which the state owns the land, is one of the most centralized forms of the tributary mode.

Different modes may coexist within the same system. Some forms of organization are best understood as transitional or mixed. For example, class-stratified but stateless systems in which kinship metaphors are used to legitimate the exploitation of commoners by a noble class (e.g. precontact complex chiefdoms in Hawaii or early Thai 'chiefdoms' in Southeast Asia) constitute a mix of kin-based and state-based (coercive) systems. The 'Germanic mode of production' or

[88]

'decentralized stratified society' (Kristiansen 1991:19), various forms of slavery, and 'market socialism,' are other mixed modes.

While transitional and mixed forms complicate the analysis of transformations,, we contend that there have been qualitatively distinct logics of accumulation. We do not assume a theory of unilinear evolution, but seek to discover empirically the patterns, possibilities, probabilities of past and future transformations. We further argue that it is possible, if difficult, to use knowledge of past transitions to help humans choose among more desirable future alternatives. We emphasize transformations of modes of accumulation more strongly than most of the other scholars who study precapitalist world-systems.

Since we are interested in connections, we propose to study interaction networks rather than various types of trait distributions. The types of interactions that are important should be studied empirically, rather than assumed. This allows us to investigate any variations in the relative importance of different types of interaction.

We note that different kinds of interaction often have distinct spatial characteristics and degrees of importance in different sorts of systems. We hold that the question of the nature and degree of systemic interaction between two locales is prior to the question of core/periphery relations. Indeed we make the existence of core/periphery relations an empirical question rather than an assumed characteristic of all world-systems.

Spatially bounding world-systems necessarily must proceed from a locale-centric beginning rather than from a whole-system focus. This is because all human societies, even nomadic hunter-gatherers, interact importantly with neighboring societies. Thus if we consider all indirect interactions to be of systemic importance (even very indirect ones) then there has been a single global world-system since humankind spread to all the continents. But we note that interaction networks, while they were always intersocietal, have not always been global in the sense that actions in one region had major and relatively quick effects on distant regions. When transportation and communications were over short distances the world-systems that affected people were small. Obviously, the spatial range of consequences of all kinds of action increases as transport and communications costs decrease.

We use the notion of 'fall-off of effects over space to bound the networks of interaction that importantly impinge upon any focal locale. The world-system of which any locality is a part includes those peoples whose actions in production, communication, warfare, alliance and trade have a large and interactive impact on that locality. It is also important to distinguish between endogenous systemic interaction processes and exogenous impacts that may importantly change a system but are not part of that system. Maize diffused from Mesoamerica to eastern North America, but that need not mean the two areas were part of the same world-system. A virulent microparasite might contact a population with no

[89]

developed immunity and ravage it. Such an event does not necessarily mean that the region from which the microparasite came and the region it penetrated are parts of a single interactive system. Interactions must be two-way and regularised to be systemic. One shot deals do not a system make.

Clearly, economic forms of interaction are important in all world-systems. Of these, bulk-goods exchanges are constitutive forms of interconnection (Wallerstein 1974a, 1974b, 1979a). However, we also agree with Jane Schneider (1977) that luxury goods, especially when they are used in a prestige-goods economy (Friedman and Rowlands 1977; Peregrine 1996), are very important for the reproduction of power structures. Since there is considerable ethnographic and archaeological evidence that even nomadic foragers can pass goods over great distances, we expect that the prestige goods net may be several orders of magnitude larger than the other nets.

Intermarriage networks are also central institutions of interconnectedness in many systems, but especially in kin-based systems where they are a fundamental basis of geopolitics and geoeconomics (Collins 1992). Furthermore, marriage exchanges in kin-based systems are almost always associated with exchanges of bulk and prestige goods.

For political interconnections we use regularized political/military conflict interaction (Wilkinson 1987a, 1987b; Tilly 1984:62). Typically, this network will differ from the bulk or prestige goods networks. Finally; we note that networks of information including, but not limited to, ideology, religion, technical information, and culture must also be included as a bounding mechanism. We do not expect the information network to spatially coincide with any of the other networks.⁴

Thus, we propose four sets of bounding criteria:

• bulk-goods exchange network (BGN)

• prestige-goods exchange network (PGN)

• political/military exchange network (PMN)

• information exchange network (IN).

All regularized material and social exchanges should be included as criteria for bounding systems. Often these networks will define a set of nested boundaries. Generally, bulk goods will compose the smallest regional interaction net. Political/ military interaction will compose a larger net which may include more than one bulk-goods net, and prestige-goods exchanges will link even larger regions which may contain one or more political/military nets. We expect the information net to be larger than the prestige goods net, sometimes far larger. Nonetheless it may also be smaller (see Figure 4.1).

At first it may seem counterintuitive to have the information boundary inside the prestige goods exchange, since exchange of goods typically implies some exchange of information. There are, however, well known mechanisms by which goods can be exchanged beyond the range of information. When trade goes from partner to partner the physical objects may travel much further than information.

[90]

Figure 4.1 Boundaries of the four world system networks.

When warfare is severe the political/military boundary may cut the flow of information even while prestige goods cross the boundary via circuitous down-the-line exchanges.

A few other comments on Figure 4.1 are in order. First, other nets are shown as regular geometric shapes for clarity of presentation. Second, the numbers of nested nets are illustrative only. Any actual world-system could be either simpler, or more complex than Figure 4.1 suggests. Third, we do not expect each net to be of the same scale. Rather, we seek only to convey relative sizes. Rather than a priori privileging one of these networks, we propose that world-systems may be constituted by any or all these linkages.

We do not claim that the networks will always be nested in the fashion described. Occasionally, as in both the modern global world-system and some earlier geographically isolated systems (e.g. the Hawaiian Islands), these four networks converge. Such convergence may be an important characteristic which differentiates some world-systems from others.

We do, however, expect that the relative sizes of the nested networks may change through time. Figure 4.2 shows one possible sequence. Reading Figure 4.2 from left to right as an historical sequence, we note that the bulk goods and political/military nets increase in size relative to the prestige goods and information nets. Second, the information net shifts through time. These shifts suggest significant changes in intersocietal relations and in the conditions and costs of travel and transportation. Again, the figure is suggestive of the types of changes we might find. Once we have bounded a system we can discuss how it might change.

The first question for any focal locale is about the nature and spatial characteristics of its links with the above four interaction nets. This is prior to any consideration of core/periphery position because one region must be linked to another by systemic interaction in order for consideration of core/periphery relations to be

[91]

Figure 4.2 Hypothetical evolutionary sequence of world system boundaries.

Key: BGN: Bulk Goods Net. P/MN: Political/Military Net. PGN: Prestige Goods Net. INF: Information Net.

relevant. We divide the conceptualization of core/periphery relations into two analytically separate aspects: core/periphery differentiation, and core/periphery hierarchy.

Core/periphery differentiation exists when two societies systemically interact and one has higher population density and/or greater complexity than the other. The second aspect, core/periphery hierarchy, exists when one society dominates or exploits another. These two aspects often go together because a society with greater population density/complexity usually has more power, and so can effectively dominate/ exploit the less powerful neighbor. But there are important instances of reversal (e.g. the less dense, less complex Central Asian steppe nomads exploited agrarian China) and so we want to make this analytical separation so that the actual relations can be determined in each case. We also note that the question of core/periphery relations needs to be asked at each level of interaction designated above. It is more difficult to project power over long distances and so we should not expect to find strong core/periphery hierarchies at the level of Information or Prestige Goods Networks.

To test hypotheses about how core/periphery relations shape intersocietal relations and processes of social change we distinguish between two types of interaction (following Myrdal 1971). First are spread effects, akin to diffusion, in which core/periphery interactions cause peripheral areas to become more core-like. Second are backwash effects in which core-periphery interactions cause peripheral areas to become less core-like. This is familiarly called the 'development of underdevelopment.'⁵ The circumstances that promote one over the other, their relative frequencies and, most importantly, their roles in the degree of intersocietal inequality and the stability of core/periphery relations are fundamental issues for empirical investigation.

We intentionally omit consideration of the nature of what is produced and traded between cores and peripheries in our definition of these concepts. The idea that core areas specialize in manufacturing and peripheral areas in raw materials remains controversial even for the modern world-system (Chase-Dunn 1998:ch. 10; Martin 1994a). Phil Kohl (1987a, 1987b) argues that in the ancient Mesopotamian world-system steatite (soapstone) bowls were manufactured on the

[92]

peripheral Iranian plateau and traded to the core cities of the Mesopotamian lowland in exchange for food, reversing the division of labor typical of the modern world-system. This reversal is readily explained: in the absence of cheap bulk transportation it is easier to move already manufactured goods to the core than to move the raw materials there.

Core/periphery relations can be even more complex. Mitchell Alien (1997:ch. 1) has developed the concept of a 'contested periphery,' for which one or more core regions compete. Based on a study of Philistia and its relations to the Neo-Assyrian and Egyptian world-systems, he finds that once an area has been incorporated into one world-system it can more easily be moved into another world-system than if it were being incorporated for the first time. Not surprisingly, contested peripheries have more leverage in responding to core demands. Furthermore, what is a periphery in one world-system can become a semiperiphery in another. If such a region provides access to valuable resources or other core regions, it can often leverage this control into semiperipheral relationships.

Alien's analysis is one impetus to rethinking our conceptualization of the semi-periphery. We argue that semiperipheral regions:

1 may mix both core and peripheral forms of organization;

2 may be spatially located between core and peripheral regions;

3 may be spatially located between two or more competing core regions;⁶

4 may be the locus of activities between core and peripheral areas; and

5 may be one in which institutional features are in some ways intermediate between those forms found in core and periphery.

Without more-detailed empirical studies, it is premature to define the semi-periphery more narrowly. Indeed, we should not assume a priori that all world-systems have semiperipheries.⁷

Finally, we see historical social change as open-ended and path dependent. That is, it occurs in any existing social structure within the context of a specific historical legacy and specific current conditions. Important bifurcations and discontinuities of development, rapid transformations, and instances of devolution are normal characteristics of social change (see Sanderson 1990, 1999).

Our argument that world-systems are the primary unit of analysis for understanding these processes does not vitiate the importance of processes that operate within societies or other social groups. A world-system is composed, not only of intersocietal interactions, but of the totality of interactions that constitute die whole social, economic, and political system. Good world-systems analysis in modern or precapitalist settings always attends to the complex dialectic between social change within any of its composite units and the entire system.

If our hypothesized typology of modes of accumulation is correct, we have only two transformations to study: the transition from kin-based to state-based or

[93]

tributary logics and the transition from tributary to capitalist logic. A transition to whatever follows capitalism would constitute a third transformation.

We propose the following four research strategies:

1 Hierarchy should be approached as an empirical issue, not as a theoretical assumption. Degrees of hierarchy, the units which are hierarchically related, the forms of exploitation and domination, rates of mobility, rates of expansion, peripheralization vs. more egalitarian interaction, the development of underdevelopment vs. coevolution, are matters to be investigated in each case.

2 Commodification should be conceptualized as a variable process. It should be broken down into the sub-components of land, labor, wealth, and goods. Forms of commodification should be analyzed. The extent and importance of commodification should be determined in comparative perspective. Types of goods, forms of production, nature of payment, importance of price-setting, competitive market forces, and the timing of spread of these should be determined by careful studies.

3 World-system interaction networks should be studied empirically in terms of densities, types of contact, and relationships between the four interaction nets. Future studies may require the refinement of these categories or the development of new ones.

4 The concrete study of transformation should attend to the historical particularities of the systems being studied. While patterns and general principles may emerge from comparative analysis, they must be grounded in historical details.

We also suggest inventorying of large numbers of world-systems to facilitate formal comparative studies that can address issues regarding what is typical and what is exceptional. The assembly of a data set containing large numbers of world-systems will enable us to separate some of the more conjunctural aspects of the transformation problem from its more systemic aspects.

World-systems tend to grow. Societies have gotten larger, in terms of population size, population density, territorial extent, absolute and relative productivity. Growth entails absorption of formerly external areas, the incorporation of new peoples and territories and/or the merger of formerly autonomous world-systems. Throughout these processes no core area remains a core area indefinitely. Development is uneven. Old cores are replaced, often by formerly semiperipheral societies.

Incorporation involves the absorption of territory and population into a larger world-system. While mergers occur when systems with comparable levels of

[94]

complexity unite, we use the term incorporation to refer to cases in which a large world-system engulfs a small one.

We argue that incorporations or mergers can occur at each network level. Two separate systems are first likely to interact through the information network. If one or both are expanding they will later become part of a single prestige goods network, and still later they will join into a single political/military network, and then a single bulk goods network. This sequence is likely to be the same regardless of the relative levels of complexity of the two systems. But their relative levels of complexity will often be an important element in the nature of the interactions that occur.

We conceptualize incorporation as a continuum that ranges from weak to strong. We argue that to label the entire continuum of incorporated areas 'peripheral' masks important variations and makes it difficult to understand the boundary processes of world-systems (see Figure 4.3 and Hall 1986, 1989, 1999; Chase-Dunn and Hall 1997:ch. 4).

At the weak pole are areas external to a world-system; next are areas where contact has been slight. We call these external arenas and contact peripheries respectively. In the middle range are 'marginal peripheries,' or 'regions of refuge* At the strong pole are 'full-blown,' or dependent peripheries.

Key points of Hall's critiques (1986, 1989) of conventional world-system analyses of incorporation are: (1) that even at very weak levels, incorporation can

Figure 4.3 Continuum of incorporation.

Source: From Chase-Dunn and Hall 1997.

[95]

have dramatic effects; (2) that while somewhat reversible, incorporation is 'sticky' or 'grainy,' tending in general toward stronger levels and very seldom returns to the status quo ante (contra Arrighi 1979); (3) that even at very weak levels of incorporation, peoples in newly peripheralized areas attempt to control and shape the process, often with a modicum of success; and (4) that conventional world-systems analysis has examined only strong levels of incorporation where backwash effects have been very strong, and as a result they misunderstand the overall process.

For instance, North American furs were not vital to European economies, yet the fur trade produced major social and economic changes in indigenous world-systems (Wolf 1982:ch. 6; Kardulias 1990; Abler 1992; Dunaway 1994, 1996a, 1996b).

Similarly, plunder is a form of incorporation within the PMN. Thus, West Africa was incorporated into the PMN of the Central system when slave raiding became regularized, rather than, as Wallerstein (1989:ch. 3) would have it, after the development of colonial agriculture.

Even weak incorporation can lead to major transformations of incorporated groups. When Cherokees became extensively involved in 'a putting-out system financed by foreign merchant-entrepreneurs' (Dunaway 1994:237), not only was their culture transformed, but techniques of production changed. The spread of horses from New Mexico to foraging groups living on the fringes of the Great Plains transformed erstwhile sedentary (or sometimes semisedentary) horti-culturalists into full-time nomadic hunters (Secoy 1953; Mishkin 1940).

Wolf (1982) and Ferguson and Whitehead (1992) present other examples, all of which point to the dramatic consequences and limited reversibility of the effects of incorporation of nonstate peoples into state systems: precapitalist, mercantile, or capitalist. Because the effects often spread far beyond the zone of contact, many so-called 'pristine' ethnographic examples are actually the products of dramatic transformations.

Sherratt described quite similar effects in Bronze Age Europe (1993a, b, c; Sherratt and Sherratt 1993): 'The characteristic of the margin is that it is dominated by time-lag phenomena - 'escapes' - rather than structural interdependence with the core' (Sherratt 1993a:43). His 'margin' corresponds approximately to our contact or marginal peripheries.⁹

Thus, incorporation creates frontier zones where 'mixed' and hybrid social forms are found. Our main point is that 'frontiers' are formed and transformed by world-systemic processes, and cannot be fully understood by examining only the frontier itself (see Hall 2000 for more details).

The process of incorporation also varies with differences both in the type of world-system doing the incorporating and the type of group being incorporated. Our heuristic typology suggests that there are many possible combinations. As we have noted, large differences are likely to lead to incorporation, whereas smaller differences led to merger.

One very important difference is between state and nonstate societies. While all stateless societies are kin-ordered, tributary and capitalist systems are state-based.

[96]

Their organizing principles are fundamentally different. State-based world-systems typically try to impose some degree of political centralization on nonstate peoples.

Sometimes the strategy can backfire, and a powerful enemy can be created. This almost happened with Comanches in eighteenth century New Mexico (Hall 1989). It definitely happened with the Mongols (Barfield 1989).

Analogously, severance of core domination, typically due to collapse of a core state, can produce opposite effects on weak versus moderately incorporated peripheries. If the core is extracting some local resource, such as human captives, the loosening or severing of that connection would typically allow a return of local prosperity. If, however, the core supplied some resource for which there was no local substitute, such as guns, any consequential prosperity would typically collapse with its loss.

Mergers are somewhat different. First, the incorporation or merger of states is a costly process because states are capable of powerful resistance. Second, following the 'no intervening heartland' rule (Collins 1978, 1981, 1986), expansion that entails passing through the heartland of another state is likely to fail. This is why most successful expansion is typically into contiguous territories. Third, tributary empires more often engage in pure plunder, whereas capitalist states are more likely to follow initial plunder with an effort to set up commodity production based on coerced labor.

Mergers of world-systems are sufficiently different that they should not be subsumed under the topic of incorporation.' A merger of world-systems seldom results in the peripheralization of one by the other, whereas incorporation most often does. Rather, it constitutes a new, larger world-system in which the two formerly separate systems play more or less equal parts. The two most well-known instances are the merger of the Mesopotamian and Egyptian systems and the merger ofAfro-Eurasia.¹⁰ Obviously, throughout human history many small systems must have merged to make larger ones. These processes need to be studied comparatively in order to discover: (1) whether world-system mergers differ across types of world-systems; and (2) the differences in merger or incorporation at the information, prestige goods, political/military, and bulk goods nets levels.

We argue (1997:ch. 5) that the semiperiphery is fertile ground for social, organizational, and technical innovation and is an advantageous location for the establishment of new centers of power. In particular, secondary state-formation on the marches of empires has frequently been recognized as a semiperipheral phenomenon that is related to the rise and fall of empires and the shift of hegemony within interstate systems (e.g. Mann 1986). A broadly similar

[97]

phenomenon occurs among chiefdoms (e.g. Kirch 1984:204). Semiperipheral capitalist city states in the tributary world-systems and some semiperipheral national states in the modern world-system have been upwardly mobile and played transformative roles.

Sargon at Akkad, the first unifier of the Mesopotamian city states, combined elements of a peripheral kin-based mode with those of the core tributary mode to conquer the Sumerian core and establish a more centralized, more exploitative, purer form of the tributary mode of production than had ever existed before."

The ability to generate new and effective institutional forms also occurred very often in capitalist city-states, which typically existed in the semiperipheral interstices of empires dominated by the tributary mode of accumulation. The city-states of antiquity were semiperipheral because they were on the edges of, or the boundaries between, large territorial empires. They were often located such that they could easily mediate trade between the core empires and peripheral regions. They could sometimes manipulate this position to maintain considerable political and economic autonomy, although they were not infrequently swallowed up by imperial expansion (Frankenstein 1979). The important cases were formally sovereign: e.g. Dilmun, Byblos, Tyre, Sidon, Carthage, Malacca, Venice, Florence, Genoa, Antwerp, and the cities of the Hanseatic League.

Most of these cities specialized in maritime trade. Coastal or island locations made them defensible with naval forces from would-be conquerors. Easy access to water lowers transport costs. These cities often engaged in manufacturing goods that facilitated their trade-based strategy of accumulation. They were powerful agents of commodification and commercialization in the still predominantly tributary world-systems. We argue that the 'rise of the West' is best understood as another instance of semiperipheral development, an upwardly mobile semi-peripheral region within the larger Afro-Eurasian system that eventually succeeded in dominating the entire globe.

Our explanation of world-systems evolution is composed of three elements:

1 semiperipheral development;

2 an 'iteration' model that involves demographic and ecological variables as causes of hierarchy formation and economic intensification; and

3 transformations of modes of accumulation.

We have already outlined the phenomenon of semiperipheral development above. The iteration model (Figure 4.4), a synthesis of the approaches developed by anthropologists Harris (1977, 1979), Cohen (1977) and Carneiro (1970), shows what we think are the main sources of causation in the development of more hierarchical and complex social structures. It is an 'iteration model' because the variables both cause and are caused by the main processes.¹² This is a positive

[98]

Figure 4.4 The iteration model with temporary direct effects.

feedback model in which systemic expansion, hierarchy formation, and technological development are explained as consequences of population pressure. In turn these changes intensify population growth. The process repeats cyclically. With each 'iteration' the process repeats on an expanding scale. This is a world-systems model becauserit necessarily involves processes that occur among societies as well as within them. The main Tortes are population growth, ecological degradation and population pressure. Population pressure results when scarcities cause people to increase the effort necessary to meet their needs. This often results in emigration, if new regions are available. Circumscription occurs when new locations do not geographically exist or are already filled with other populations who resist. Thus population pressure causes competition among societies for land and other resources and this is necessarily an intersocietal phenomenon. One possible outcome is increased conflict, especially warfare, as groups contend for scarce resources. In some systems, endemic warfare functions as a demographic regulator by reducing the population density and alleviating (temporarily) population pressure. But in other cases new hierarchies or larger polities emerge to regulate the use of resources, and/or new technologies of production develop that allow larger numbers of people to live within a given area. The world-system insight here is that the newly emergent elites often come from regions that have been semiperipheral. This is because semiperipheral actors are usually able to assemble effective campaigns for erecting new levels of hierarchy.

The institutional inventions made and spread by semiperipheral actors often led to qualitative transformations in the logic of accumulation, and thus significant alterations in the operation of variables in the iteration model. Still, these

[99]

qualitative changes are themselves the consequences of people trying to solve the basic problems produced by those forces and constraints depicted in the model. Figure 4.5 illustrates in a general way what happens with the emergence of new modes of accumulation, especially states and capitalism. The new modes allow some of the effects of population pressure to have more direct effects on changes in hierarchies and technologies of production. This shortcuts the path that leads through migration, circumscription, and conflict. How can the emergence of states allow population pressure to have a more direct effect on hierarchy formation and technological change? Population pressure in outlying semiperipheral areas combines with the threats and opportunities presented by interaction with the existing states to promote the formation of new states. Thus, secondary state formation becomes common. This is the main way state formation short-cuts the processes. This does not mean that conflict disappears. Rather that it does not need to reach the same levels of intensity in order to provoke the formation of new states once states are already present in a region.

State formation also articulates the rising costs of intensification with changes in technology. The specialized organizations that states create (bureaucracies and armies) sometimes use their powers and organizational capabilities to invent new kinds of productive efficiency and to implement new kinds of production. Governing elites sometimes mobilize resources and labor for irrigation projects, clearing new land for agriculture, developing transportation facilities, and so forth (e.g. the oft noted superiority of Roman roads). In this scheme semiperipheral marcher states and semiperipheral capitalist city states were the most important transformational actors in the rise of larger and larger empires, the

Figure 4.5 Population pressure/intensification hierarchy formation model.

[100]

increasing scale of markets and the eventual emergence of predominant capitalism.

Here we construct a spatio-temporal chronograph based on our approach to spatially bounding world-systems. The sequence runs from small networks of nomadic foraging bands, to larger systems of mesolithic sedentary foragers, to even larger systems of sedentary horticulturalists, to still larger systems in which core regions included the first cities and early states, to yet larger systems of agrarian empires, and eventually to today's single global capitalist political economy. Please note Figure 3.1 on p.58. Here Wilkinson used the PMN (political/ military network) to bound major systems, following his earlier work (1987b). Figure 3.1 shows how a 'Central' PMN composed of the merging of the Mesopotamian and Egyptian PMNs in about 1500 BCE eventually incorporated all the odier PMNs into itself. Figure 3.1 shows only those PMNs with cities larger than 10,000. Many cityless PMNs were also incorporated.

Figure 3.1 does not show what happened to the other interaction nets. If we were to consider the expanding boundaries of PGNs, the result would look much the same except that the time scale on the left margin would be shifted. For example, die Mesopotamian and Egyptian PGNs became linked probably as early as 3000 BCE, while the PMNs did not merge until 1500 BCE.

We have found that all hierarchical world-systems exhibit cyclical processes of:

• pulsadons in the spatial extent of interaction networks;

• sequences of rise and fall of large polities; and

• oscillations between state-based and private forms of accumulation suggested by Ekholm and Friedman (1982).

We have found that all systems we have studied, including even very small and egalitarian ones, exhibit cyclical expansions and contractions in the spatial extent of interaction networks. Based on this, we posit that all four networks (BGNs, PMNs, PGNs, and INs) 'pulsate.' By 'pulsate' we mean that die spatial scale and intensity of interaction increases and then decreases at each of these network levels. When interaction increases, there are more exchanges with consequences over a greater distance.

We call cycles of centralization/decentralization of political/military power among a set of polities 'rise and/all.' Again, we observe that for all hierarchical world-systems (whether they are composed of chiefdoms, states, empires, or capitalist states) die larger polities experience cycles of growth and decline in size and power." However, very egalitarian and small-scale systems such as the sedentary foragers of northern California (Chase-Dunn and Hall 1997:ch. 7) do not display this kind of cycle. According to Ekholm and Friedman (1982), Frank (1993a), and Frank and Gills (1993) all world-systems for die last 5,000 years also oscillate between private and state-based forms of capital accumulation.

[101]

Figure 4.6 The expansion of Central and East Asian networks.

That these cyclical processes - pulsation, rise-and-fall, and oscillation - occur in very different kinds of systems raises several questions. Are the underlying mechanisms that generate these sequences the same or similar in all systems? What are die temporal and causal relations among these different cycles? What is the relationship between the rise and fall of large polities and changes in the degree of inequality within polities? Are these relationships similar across different kinds of world-systems? What is the relationship between the rise and fall of large polities and oscillations between state-based and private forms of accumulation? How are political rise-and-fall and network pulsations related to the general 200 year phases of expansion and contraction posited by Gills and Frank (1992; Frank 1993a)? Are these cycles really synchronous in regions connected only by very long-distance trade in prestige goods? We can only begin to address some of these questions here.

Some cyclical processes have different characteristics and different causes in distinct types of world-systems. Even though die rise and fall of chiefdoms is analytically similar to the rise and fall of empires and of hegemonic core powers, and

[102]

even though all are related to the stability of institutions for extracting resources from distant regions, there are tremendous differences in scale and process.

Anderson (1994) summarizes the anthropological and sociological literature on 'cycling' in his study of the rise and fall of Mississippian chiefdoms. These 'cyclings' are processes by which a chiefly polity extended control over adjacent chiefdoms and developed a two-tiered hierarchy of administration above local communities. Later these regionally-centralized chiefly polities disintegrated back toward a system of smaller and less hierarchical polities.

Chiefs typically relied more completely on hierarchical kinship relations, control of ritual hierarchies, and control of prestige goods imports than have the rulers of states. States have specialized organizations for extracting resources that chiefdoms lack. In turn, states and empires have been more dependent on the projection of armed force over great distances than modern hegemonic core states have been. The development of commodity production and mechanisms of financial control, as well as further development of bureaucratic 'techniques of power' (Mann 1986) have allowed modern hegemons to extract resources from faraway places at less cost.

The development of techniques of power have made core/periphery relations ever more important in competition among core powers and have altered the ways in which the rise-and-fall process works in other respects. One of these is the degree of centralization achieved within the core areas of the modern capitalist world-system as compared to tributary world-systems. Tributary systems alternate between a structure of multiple, competing core states and core-wide (or nearly core-wide) empires. The modern interstate system experiences the rise and fall of hegemons, but the hegemonic core state never even attempts to conquer the other core states to form a core-wide empire. This is because modern hegemons are pursuing a capitalist, rather than a territorialist, form of accumulation. Thus, even omitting non-state, chiefdom world-systems, there is a significant difference between capitalist and tributary systems.

The simplest hypothesis regarding the temporal relationships between rise and fall and pulsation is that they occur in tandem. Whether or not this is so, and how it might differ in distinct types of world-systems, are questions amenable to empirical research. The spatial relationship between PMNs and PGNs expand and contract synchronically across Eurasia over the past 6000 years (Chase-Dunn and Hall 1997:ch. 10; see Figure 4.6).

Figure 4.6 is a hypothetical depiction of the temporal relationship between PMN and PGN pulsation in the Central and East Asian systems as they expanded, intermittently touched each other, and eventually merged to form the global system. This is the old issue about whether the flag follows trade or trade follows the flag. In our version, prestige goods trade leads the flag, but both expand more or less concurrently. While this portrayal is hypothetical, it would be possible to use actual temporal changes in the spatial extent of PMNs and PGNs to examine the synchronicity of expansion and contraction.

What about the relationship between rise-and-fall and PMN expansion and contraction? It would seem that this also would be temporally synchronized.

[103]

When core regions are more centralized (during 'rise') they contain larger polities with presumably greater spatial reach. This would extend the boundaries of the PMN. Thus, we should be able to use a measure of the centralization of political power within a core as a proxy for the expansion and contraction of the PMN.

At this point we do not have a direct measure of trade expansion and contraction. What we do have are estimates of the population sizes of cities. It may be reasonable to assume that cities grow larger during periods of economic growth and the expansion of trade, and that they decrease in size (or grow more slowly) during periods of decline.

Based on this tenuous assumption we have examined the relationships between changes in the population size of the largest city and changes in territorial size of the largest empire within several different PMNs. The correlations of the relationships between these measures for the Central and Far Eastern PMNs over the last 4,000 years are positive, but neither large, nor statistically significant (Chase-Dunn and Hall 1995:130-1). This weak support for simultaneous trade and political/military pulsations might be due to poor measures. Clearly, further research is needed.

The same study found little support for the hypothesized expansion and contraction phases specified by Gills and Frank (1992; also Frank 1993a). Neither urban growth nor empire size correspond very well with the Gills/Frank phases in either the Central or the East Asian PMNs (see Figure 4.7).

While examining the relationships within PMNs of urban and empire growth/ decline we discovered that city growth and empire growth seem to occur synchronously in the Central (West Asian and Mediterranean) and the East Asian PMNs (Chase-Dunn and Willard 1993). Some relevant evidence is contained in Figure 4.7 which shows the territorial sizes of the largest empires in the Central and East Asian PMNs from 1500 BCE to 1750 CE.

Figure 4.7 presents strong evidence that growth and decline phases are synchronized in PMNs that are linked within a larger PGN. However, we also found that the intermethate Indie PMN did not experience a similar sequence of growth and decline phases (Chase-Dunn and Hall 1997: fig. 10.11).¹⁴

What are the causes of this synchronization? Does this kind of relationship hold in other PGNs? To answer the second question we have examined the relationships between urban and empire growth in the Mesopotamian and Egyptian PMNs during the period in which they were linked into a larger PGN. Figures 4.8 and 4.9 show these results. Figure 4.8 displays die population sizes of the largest cities in the Mesopotamian and Egyptian PMNs between 2250 and 600 BCE. In 1500 BCE the Mesopotamian and Egyptian PMNs merged to become a single larger PMN that we call the Central PMN. Nevertheless, changes in the sizes of the largest cities in these regions do not seem to be synchronous.

Figure 4.9 shows the changes in the territorial sizes of the largest empires in Egypt and Mesopotamia from 3000 BCE to 1450 BCE. Clearly, the empire sizes

[104]

Figure 4.7 Central and East Asian empire sizes, 1500 BCE-1800 CE.

Note: Pearson's 4 = .90, n = 60, p < .0001.

are uncorrelated across fourteen time points (Pearson's r correlation coefficient is -.17 [n=14]). The Egyptian/Mesopotamian comparison does not support the idea that all PMNs within larger PGNs have synchronous processes of growth.

The synchronicity of the growth of cities and empires in the Central and East Asian PMNs remains a puzzle. One possibility is northern Eurasian-wide climatic fluctuations. India, at a more equatorial latitude across the Himalayas, may have experienced very different climatic fluctuations. Climate change can affect urban growth and empire-formation through its affects on agricultural productivity (Nix 1985). Periods of flooding may disrupt irrigation systems, and drought may negatively affect agriculture. Recent evidence indicates that the collapse of Mayan states may have been caused by a period of extended drought. Weiss et al. (1993) contend that both the expansion and collapse of the Akkadian empire were spurred by climate changes.

If we found significant relationships between indicators of climate change and the urban and empire growth/decline sequences we would want to examine the direction of causality. Does climate change cause urban change, or does the expansion of agriculture associated with urban growth cause climate change? It is possible that expanded agricultural activity, and/or deforestation due to human exploitation, may have affected local and regional rainfall patterns and ground water levels. (See Chew in this volume.) Thus, population density, methated by intense agriculture and forest exploitation, and thus urbanization, may have affected climactic fluctuations. There is a developing literature on the anthropogenic causes of climate change. It is well known that the intensification of

[105]

Figure 4.8 Largest cities in Egypt and Mesopotamia. Source-. Chandler 1987; Wilkinson l992a.

productive activities causes em-ironmental degradation. This, in turn, has affected the development of human societies for millennia. If urban growth episodes precede climate change or changes in water levels then causality in the direction of human effects on climate will be supported. But as it stands research on climate change in the relevant areas has not been combined with our measures of urban and empire growth to see if these are empirically related.

Microparasites, mediated through trade networks, could also affect city and empire growth and decline. As trade increases in density and volume, formerly isolated disease pools come into contact, unleashing 'virgin soil epidemics' (Crosby 1972, 1986). These epidemics can produce massive disruptions, and following Goldstone's (1991) argument, can unleash social, economic, and political changes. As pathogens and hosts adapt to each other, diseases become less lethal and populations recover. Trade resumes, and the cycle can repeat as other, formerly isolated disease pools come into contact, or as new diseases spread along trade networks. This might account for some regularity in such cycles since humans have only been able to intervene in the biological processes of mutual adaptation in the late twentieth century.

A more interesting explanation from the world-systems perspective is Frank's (1992) hypothesis of the 'centrality of Central Asia' as a peripheral region linking both ends of the Eurasian continent. The Mongol Empire briefly linked Western Asia and China into a single polity in the thirteenth century CE. Barfield (1989) built on Lattimore's (1940) observations to trace the long-term linkage of the rise and fall of steppe empires with the rise and fall of agrarian empires in China. Citing this and other evidence, Frank (1992) contends that processes of peripheral

[106]

Figure 4.9 Egyptian and Mesopotamian empire sizes.

Source: Taagepera (1978a,b). Note. Pearson's r correlation = -. 17, n = 14.

migration and"steppe-9mpire formation and their affects on the long-distance trade carried along the Silk Roads are the explanation of the linkages between Rome and China first reported by Teggart (1939). Figure 4.7 (and others in Chase-Dunn and Hall 1997:ch. 10) show similar synchronization.

Perhaps it is Frank's Central Asian linkage that accounts for this. But in order to accept his explanation we need to rule out the climatic fluctuations hypothesis and explain why India was not affected in the same way. The same caveat holds for disease linkages (McNeill 1976). One explanation for the South Asian exception -but a none too satisfactory one - might be that the tropical and semi-tropical climates there were subject to a different disease regime. It is also conceivable, given the Himalayan barrier, that climatic cycles in South Asia differed significantly from northern Eurasia. The monsoons certainly follow a different rhythm from the weather in northern regions.

Since warfare affects both urban growth and the territorial size of empires, steppe-empire formation and the attendant fighting and migration of pastoral nomads may well be the cause of the simultaneous rise and fall of empires at both ends of the Eurasian landmass. If so, warfare between steppe nomads and agrarian states in both western Asia and East Asia should be correlated. Barfield's (1989) Perilous Frontier provides information for the East Asian region. For western Asian, data on warfare obtained by the LORANOW project (Cioffi-Revilla 1991; 1996) could be used.

[107]

The way in which South Asia was connected into Afro-Eurasian trading patterns may account for the South Asian exception. India had multiple connections overland. Maritime links go back at least two millennia. At first they consisted of coastal trade. Later, as sailors mastered the monsoons, they crossed the Arabian Sea and the Bay of Bengal (see maps in Chaudhuri 1985 and Abu-Lughod 1989:172-3, 202, 252). Thus, at any given time. South Asian states had multiple routes of access into trade, so disruption of any one route - for whatever reason - could be compensated for by means of alternate routes.

Beyond the multiple routes, South Asia had extensive relations with Southeast Asia which opened up even further routes.'⁵ At times when conditions through the straits of Malacca and/or Sunda made sea trade very risky, portages across the Malay Peninsula or overland through northern Southeast Asia (what today is northern Myanmar (Burma), Thailand, Laos, and Vietnam, were possible. Thus, while a large state (e.g. Funan, Khmer Empire, Srivijaya [Java], and later Siam or the city-state Malacca) could block one or more paths, no one of these states could control all the paths from India to China.

Indian connections to, and trade with, various Southeast Asian states itself buffered India from blockages on other routes. Southeast Asia supplied aromatic woods, spices, and especially gold. When access to northern sources of gold were severed by Bactria in Central Asia, and the Romans sought to curb the export of gold to the east in the first centuries of the current era, India turned to Southeast Asia to fill the gap. As Coedes notes (1968:19ff) the region was known as 'the land of gold.' However, unlike the Central Asia steppe federations which rose and fell with Chinese empires (Barfield 1989, 1991), the states of Southeast Asia seemed to wax and wane counterpuntally to Chinese and Indian empires.

Thus, for the Mongol era at least, an elaborated version of Frank's Central Asian thesis seems to make more sense of the coordination of events in West and East Asia, and the exceptionalism of South Asia than any other explanation. While not ruling out a role of climatic fluctuations and/or spread of pathogens, it suggests that they were key factors.

On some issues the differences between ourselves (as probably the most extreme transformationists) and the most extreme continuationists (Frank and Gills) are of a rather mild order. Frank and Gills, like Ekholm-Friedman and Friedman, see a continuous coexistence of capitalism and geopolitics for five thousand years since the emergence of states in Mesopotamia. We see a slow and uneven rise of commodification and capitalism over the same period, culminating in a watershed of capitalist predominance that occurred first in the European regional system in the seventeenth century. We emphasize comparison of different systems while Frank and Gills concentrate on the Eurasian World System. We agree that there are important similarities between different systems and that there are many structural and processual continuities that link the Eurasian world-system with the contemporary global system. While we see the value of very long-term studies of

[108]

single large systems, we argue that a comparative approach is important because it allows us to study both differences and transformations as well as similarities and continuities. Our case studies support the notion that there are important differences and qualitative transformations.

Our first conclusion is that there have indeed been transformations in the basic logic of accumulation. Our study of the Wintu and their neighbors revealed a small world-system in which social labor was primarily mobilized by means of kinship structures based on cultural agreements about obligations. Accumulation was not accomplished by means of market exchanges or political taxation or tribute. These forms of accumulation did not exist. We contend that this very small world-system of sedentary foragers was similar to the first sedentary systems of mesolithic western Asia and the Levant twelve thousand years ago. If true, major qualitative transformations must have occurred to produce world-systems in which tributary and capitalist accumulation became predominant. We also identify important qualitative changes in the operation of cycles of rise and fall as they occur in tributary and capitalist world-systems. This supports the transformationist approach.

Regarding our multilevel approach to spatially bounding world-systems (the four nets) our survey leads us to conclude that the degree of systemness is typically greatest in the bulk goods net and lessens as we move to the larger nets. This does not mean that the larger nets are unsystemic. Political/military and prestige goods nets are frequently important for both reproduction and structural change. We do not assert that this is always the case, but it is a plausible generalization. This conclusion provides a new slant on the debate about which are the most important types of interaction. Wallerstein's emphasis on bulk goods networks is the most conservative, because bulk goods are always important.

World-system evolution, involves three interlinked processes: semiperipheral development, iterations of-population pressure and hierarchy-formation, and transformations of modes of accumulation. These three processes account for the evolution of human societies from a hundred thousand or so nomadic foraging bands to the single complex global system of today. Semiperipheral development linked core/periphery structures to institutional innovations that expanded and transformed social networks. Iterations of population pressure, intensification, and hierarchy-formation provided the engines of development and the dynamics of political rise and fall that are visible in all systems. Transformations of modes of accumulation altered the nature and dynamics of production, distribution, and accumulation. This, in turn, changed the way the processes of rise and fall and expansion operated.

World-systems composed of unhierarchical groups do not have stable core/periphery structures. Core/periphery relations become more unequal as societies develop the 'techniques of power' that allow distant groups to be dominated and exploited. Core/periphery relations are very important for world-systems evolution both because of the role of semiperipheral development and because competition among core societies revolves around how well they can exploit and dominate peripheral regions. The modern world-system is an extension of this

[109]

trend toward the increasing stability and importance of core/periphery relations. Geopolitically structured capitalist commodity production is the most efficient and stable exploiter of peripheral regions. But semiperipheral development continues to operate in the global world-system, and so new transformations are possible.

We have seen that throughout the process of world-system evolution new areas and new peoples have continually been incorporated or merged into expanding systems. Incorporation and merger often drastically transform individual groups, underscoring the need to study social change in a context of intergroup interaction. Pristine conditions seldom exist, although we argue we have found one such case (Chase-Dunn and Hall 1997:ch.7). The study of incorporation further emphasizes a point McNeill made in Polyethnicity (1986), that multiethnic states and ethnogenesis are ancient social processes (see also Hall 1998).

Answering the many questions we have raised will require a combination of detailed case studies of single systems and formal comparisons of large numbers of systems. No one has yet proposed a chronograph of the expansion of the Central information net or the Central bulk goods net. Both of these face the problem of fall-off. This does not mean that systemic consequences of changes (e.g. shortages or surpluses) extend indefinitely. The range of systemic interaction is a function of transportation and communications costs, and these change greatly across systems. Studies of these relations arc sorely needed. We also need studies of the boundaries of kin-based systems.

The existence of pulsation, rise and fall, and oscillation need much more precise documentation. Are the cycles of pulsation of each of the four kinds of interaction temporally related to each other in regular ways in most systems? What are the relations among the three types of cycles? What is driving these cyclical processes and how are their causes related?

Part of the answer may be embedded in our model of demographic intensification and hierarchical iterations. The pulsation and rise and fall cycles interact with demographic and epidcmiological processes to shape long-term sociocultural evolution. The ways in which these processes interact appear to vary across different types of world-systems, and yet the hypothesized regularities suggest some underlying basic mechanisms that operate in all systems.

Many authors contend that the contemporary system has undergone fundamental changes in the last decade. We argue that most of the visible changes in technology, globalization of investment and trade, the demise of the Soviet Union, and the appearance of 'world culture' are not fundamental. Rather, the systemic trends and cycles that have been characteristic features of the modern system for hundreds of years are continuing. Fundamental change has not yet occurred. However, fundamental change (systemic transformation) is likely to occur within

[110]

the next two hundred years as global capitalism runs into planetary boundaries (Meadows et al. 1992).

There are several possible scenarios for future transformation. World-systemic , cycles and trends could continue much as they have in the past and the whole system could be destroyed in the 2020s by a global war among core powers. Another possibility is the emergence of a world state. This would involve another round of the iteration cycle in which a global state emerges to deal with global problems. However, the iteration model has usually operated through long phases of conflict and centralization by conquest. These mechanisms would likely be fatal given current military technology. The transformation of the rise and fall sequence from empire-formation to hegemonic rise and fall involves an increase in the importance of economic competition over political/military competition, though it did not eliminate warfare. But this long-run evolutionary trend may presage the eventual elimination of warfare as a method of choosing global 'leadership.' The problem is to find an alternative mechanism to conquest for allowing global political integration. It is obvious that democratic political institutions at the global level could serve several purposes - the legitimation of a global collective security provider; a mechanism to address global environmental problems; and an opening for democratizing and balancing the processes of economic development.

We also speculate on a possible future transformation to democratic socialism. We suggest that further experiments with socialism are likely to emerge in the contemporary semiperiphery. Any rise to predominance of socialism as a mode of accumulation will also require global-level organizations.

While our main purpose has been to construct a scaffolding for the comparative study of world-systems, our eventual goal is to use this apparatus to address the real problems and possibilities of the contemporary system. This requires further research and more refined theorizing. Our speculations are meant to prod others to join the effort by demonstrating these are not mere idle, if interesting, inquiries into our past, but vital questions to understanding, and we hope, shaping a humane future.

We thank the editors of this volume and the participants in the Lund conference for many suggestions, corrections, and stimulating ideas that contributed to the development of our thinking. Over the intervening years many others have contributed to sharpening our ideas at many conferences where we have presented these and other ideas derivative of them. Figures 4.3 through 4.6 are copyright © 1997 Westview Press, and are reprinted here by permission.

1 The earliest known sedantary foragers are the Natufians of the Levant circa 10,000 BCE.

[111]

2 In sociology 'social change' is a 'covering term' which subsumes cultural, political, and economic change.

3 Typologies developed by most evolutionary thinkers are broadly convergent, even while differing in the details. Moseley and Wallerstein (1978) present a useful concordance. These typologies are rooted in considerable empirical investigation.

4 We added the information network to take into account the important insights of Schortman and Urban (1987), and Bentley's (1993) masterful survey of cross-cultural encounters in Eurasia. Discussions with William McNeill at the 1995 International Studies Association meetings and the Lund Conference prompted us to include the systemic aspects of information flows.

5 The phrase, 'the development of underdevelopment' was coined by Gunder Frank (1966) to describe the process in which core countries extract resources from the periphery. This core/periphery exploitation not only blocks development in the periphery, but also systematically distorts development in harmful ways, producing 'underdevelopment.'

6 Berquist (1995) brought this to our attention in his discussion of the Achaemenid

empire's treatment of its various western colonies. Alien (1997) discusses this in detail.

7 Woolf(1990) argues that there was no semiperiphery in the Roman world-system.

8 The term 'region of refuge' (Aguirre Beltran 1979) describes areas only partially incorporated into a state system. Partial incorporation has the consequence of freezing social change. This 'preserves' the area by setting it aside for future development. Hence it is a 'refuge' for older, 'traditional' social forms that have been destroyed elsewhere in the system.

9 The concept of 'hinterland' used by Frank and Gills (1993), GUIs and Frank (1991) and Collins (1978, 1981) also corresponds to the weaker ranges of incorporation.

10 For discussion of the former see Chase-Dunn and Hall (1997:ch.5); for the latter (ch. 8). Also note Amin (1991).

11 Collins (1978) contends that the advantages of states in the marchlands is primarily geomilitary. Because they are near the edge of the core 'heartland' they do not need to defend several borders at once. They can pursue a strategy of conquest that adds territory sequentially without threats from the rear. The disadvantage of older core powers is that they must defend themselves from many sides and their resources are spread thinly.

12 This discussion goes beyond that in Rise and Demise (1997), but draws heavily on it. \\'e have respecified our model in order to clarify the distinction between intensification and technological change.

13 For the modern world-system, rise and fall cycles involve national states that perform the role of hegemons (e.g. Britain, the US) but do not conquer the whole core. This phenomenon is termed the hegemonic sequence (Chase-Dunn 1998:ch.9).

14 Further testing with somewhat refined data further strengthens these findings (Chase-Dunn, Manning and Hall forthcoming).

15 See Cady 1966; Coedes 1966, 1968; Glover et al. 1992; Man and Milner, 1986; and Wyatt 1984, 1994.