The ideal of the ‘self-governing citizen’ -- both what it brings into relief and what it casts in shadow -- shapes our vision of democratic life, generally without our awareness. What's brought into relief by this ideal is a prescriptive project, a vision of political freedom rooted in the intuition that an individual or collectivity that is not self-governing is in some way subjugated, subordinated, or shackled by another. What's cast in shadow, to borrow from Robert Dahl, is a series of "half-hidden premises, unexplored assumptions, and unacknowledged antecedents" that form a "vaguely perceived shadow theory that forever dogs the footsteps of explicit, public theories of democracy." This dissertation is about that shadow theory. About how childhood, madness, and criminality complicate claims about the nature of subjectivity and human capabilities in theories of democracy. And about how custody and democracy can inform, deform, and transform one another. In the first two chapters I suggest children, prisoners, and the cognitively disabled present a persistent boundary problem for theorists of aggregative and deliberative democracy. For these schools of thought, custodial populations lack sociality, lack rationality, or lack maturity -- all of which are conceptually necessary to decide, deliberate, or participate in the polity. Wrapped in this 'exclusion thesis' is an assumption that the boundaries of competence can be determined prior to political contest. I demonstrate that this assumption neither stands to reason, nor produces a normatively appealing model of democratic politics. Particular institutional forms, such as the prison, mediate the relationship between the democratic ideal of self-government and competence. Civic competence is shaped by institutional contexts that are, in turn, revisable. The three chapters that follow are based on extensive field work and archival research. The third chapter draws on a unique archive of notes written by civilian observers, newspaper articles, various accounts of outside witnesses, and inmate oral histories to reconstruct the events inside of a maximum security prison in Walpole, Massachusetts during a crisis in the spring of 1973. When prison officers left their posts during a strike, something unexpected took place. Inmates formed a labor union and ran the facility themselves for nearly three months. I situate the events at Walpole in an academic debate about what role, if any, inmates should play in the governance of prisons. The central concern of chapter 4 is the rise and fall of 'self-government' groups in St. Elizabeths Hospital at the mid-century. Using a variety of primary sources, I focus on a violent ward for the criminally insane, Howard Hall, and analyze a series of novel democratic and authoritarian organizational innovations that arose in response to patient collective action. The fifth chapter draws on eight months of fieldwork at a contemporary boarding school for at-risk youth. I use original data from a social network survey instrument, unique institutional merit and disciplinary records, extensive interviews, and months of non-participant observation to call into question the liberal argument for curtailing the freedom of children. Chapter 6 concludes the dissertation. In this short chapter I review key insights from my case studies and reflect on the normative implications of my analysis. The final pages ask how one might imagine custodial arrangements that advance the normative project of democratic self-government without repeating the errors of the exclusion thesis, without succumbing to a "false belief in necessity." As a start, I argue, we could begin to think about responsive custodial institutions. At the highest level of generality, I define a responsive custodial institution as one that builds civic capacity, treats limits to participation as provisional, and experiments with mechanisms for soliciting voice. In turning to the organizational politics of schools, asylums, and prisons, I do not mean to judge one institutional form as reactionary (the Patients' Federation at St. Elizabeths Hospital, for example, that I explore in chapter 4), or another as revolutionary (the organization of the NPRA during the Walpole prison rebellion, sketched in chapter 3). Rather, my central point is that tacit acceptance of the place of custody in democratic societies forecloses an inquiry into the organizational potentials we subsequently recognize as reactionary or revolutionary. Innovations within custodial institutions, as I detail in my case studies, are key places where the possibilities of democracy are reimagined; places where new democratic subjectivities are forged; and places where law is made and remade to anticipate both forms of democracy and forms of authoritarianism to come.