Current theories of the rule of law argue that public officials respect rights if the citizens are coordinated on an equilibrium in which they collectively resist abuse. Constitutional rules are means to coordinate on this equilibrium. In past and present states, however, constitutional rules often have no effect on the rule of law. This paper suggests an alternative view of the origin and development of the rule of law. Instead of considering constitutional rules as coordination devices for citizens at large, history suggests considering them as manifestations of equilibria with rulers constrained by administrators required to implement policy. Analysis of the administrative foundations of self-enforcing constitutions may be the key to a theory and policy that would foster the rule of law in developing countries and those in transition. In particular, constitutional reforms might benefit from focusing on altering the equilibrium distribution of administrative capacity and power, providing incentives to the administratively powerful to check predation by each other and the central authorities, and to align administrators' interests with social welfare.