preexisting centralized bureaucracies were needed to plan and construct such systems. More than 30 years ago, Robert McC. Adams (1974:5) had decided that "truly large-scale irrigation works depended on state initiatives not only for their construction but for their subsequent maintenance." Eva Hunt and Robert Hunt (1974:153) agreed with Adams, adding that "the power elites are crucial for conflict resolution over water. The persons occupying water-control roles are intimately connected to the distribution of power in the society." Hunt and Hunt (1974:154) went on to argue that without unified and centralized decision making, high levels of social disruption and conflict ensue, and such conflict would tend to paralyze the socioeconomic system.The argument that state control is needed to suppress conflict and ensure the smooth running of the system is a recurrent theme of some importance. Many anthropologists have looked at systems that operated over long spans of time. Those studying small-scale irrigation systems have emphasized the role of consensus and voluntarism in decision making at the local level (e.g., Lees 1970;Mabry 1996 Mabry , 2000. It is mainly those who have studied large-scale systems who stress the need for prior centralized authority (Adams 1974).does scale maTTer?Scale matters, but scale must be defined. What is described as "large-scale" in one region may be "small-scale" in another. And scale is clearly only one of several key variables. Wittfogel focused on "top-down" state-managed systems, but researchers like Clark Erickson (2000, this volume) and Gray Graffam (1992) have emphasized the need for a balanced approach, devoting just as much time to "bottom-up" and "locally managed" decentralized systems.Bottom-up approaches are usually based on modern ethnographic or ethnohistoric data, while top-down approaches are often based on prehistoric data. This distinction is due in part to the fine-grained and temporally restricted analysis that ethnographers can achieve by observing and interviewing living informants and contemporary farmers. Ethnohistorians have eyewitness accounts of similar behaviors. In contrast, archaeologists focus on multigenerational units of time and work with a body of data that generally precludes observation of individual decision making. A good example of an ethnohistoric bottom-up study is that of Patricia Netherly (1984:229) on Peru's north coast. Although she expected to find evidence for state control of the irrigation systems in the Chimu and Chimu-Inca areas, "it soon became clear that political and economic institutions had not been managed by state bureaucracies and that, in general, hydraulic management was carried out on a number of different levels that tended to be lower rather than higher on a hierarchic scale."