This thesis compares the English decommissioned ships used as confinement of convicts known as prison hulks to the Swedish fortress prisons in the late eighteenth- to the mid-nineteenth century. Both prisons were of an improvised nature in the manner that they were not originally built to confine offenders. Prisoners were consequently kept in prison rooms under horrible conditions either below the deck of the prison hulks or within the vaults of the fortresses. Despite the inefficiency of both prisons being evident to an overwhelming majority of their contemporaries, they persisted for many years. This has provoked the question of how they managed to continue their operation despite continuously and relentlessly being considered antitheses to modern forms of confinement. Historical research on prisons has not satisfactorily pursued this question and instead chiefly focused on the emergence and development of penitentiaries and cellular prisons. This thesis instead seeks to explore how older forms of confinement not only persisted but were developed and refined in an era of prison reform. By using the theory of path dependency, this thesis will argue that both prisons managed to persist because of numerous instances of institutional reproduction. These range from, drawn-out discussions and debates on new forms of confinement, overcrowding, disease, strong prison subcultures, and mismanagement. To ensure a sharp focus, the prison hulks around Portsmouth and the fortress prisons around Gothenburg have been selected for comparison based on their general equivalence. However, references to other geographical locations will be prominent throughout.