Arguments from underdetermination take two forms, those from global sceptical underdetermination, global scientific underdetermination and local underdetermination. Arguments from global sceptical underdetermination bring into question all knowledge, they develop sceptical scenarios that purport to show that we cannot trust any knowledge that we obtain within the world. Arguments from local underdetermination aim to bring into question the nature of our knowledge and are geared against scientific realism. This thesis is an evaluation of the arguments that claim to do the latter, however it shows that these arguments are not arguments from local underdetermination but are from a type of global underdetermination that I call global scientific underdetermination. Based on this evaluation a new argument from local underdetermination is developed that attempts to show that nevertheless local underdetermination is indeed a problem for scientific realism. However, I argue that this argument also fails to undermine scientific realism. Recently Kyle Stanford has reintroduced an historical argument from underdetermination that he calls the argument from unconceived alternatives. Stanfords argument from unconceived alternatives is an inductive historical argument. It maintains that scientific theories are chosen from a non-exhaustive set of theories; claiming there is always at least one unconceived alternative that would better explain the empirical evidence. Stanfords new induction attempts to undermine scientific realism by arguing that our most successful theories will eventually be shown to be false. Various arguments against this induction will be considered. It will be shown that traditional scientific realism fails to address the argument from unconceived alternatives and the only form of scientific realism that can overcome this problem is structural realism.

Arguments from underdetermination take two forms, those from global sceptical underdetermination, global scientific underdetermination and local underdetermination. Arguments from global sceptical underdetermination bring into question all knowledge, they develop sceptical scenarios that purport to show that we cannot trust any knowledge that we obtain within the world. Arguments from local underdetermination aim to bring into question the nature of our knowledge and are geared against scientific realism. This thesis is an evaluation of the arguments that claim to do the latter, however it shows that these arguments are not arguments from local underdetermination but are from a type of global underdetermination that I call global scientific underdetermination. Based on this evaluation a new argument from local underdetermination is developed that attempts to show that nevertheless local underdetermination is indeed a problem for scientific realism. However, I argue that this argument also fails to undermine scientific realism. Recently Kyle Stanford has reintroduced an historical argument from underdetermination that he calls the argument from unconceived alternatives. Stanfords argument from unconceived alternatives is an inductive historical argument. It maintains that scientific theories are chosen from a non-exhaustive set of theories; claiming there is always at least one unconceived alternative that would better explain the empirical evidence. Stanfords new induction attempts to undermine scientific realism by arguing that our most successful theories will eventually be shown to be false. Various arguments against this induction will be considered. It will be shown that traditional scientific realism fails to address the argument from unconceived alternatives and the only form of scientific realism that can overcome this problem is structural realism.

Entre les archives, telles que définies par l’archivistique, et l’archive comme concept, tel qu’utilisé en dehors de la discipline, il existe un écart sémantique, conceptuel et théorique. D’un côté, les archives représentent des rassemblements documentaires issus des activités d’une personne ou d’un organisme ; de l’autre, il est question d’un outil conceptuel permettant d’exprimer une variété d’idées liées à l’histoire ou à la mémoire. Toutefois, cet intervalle entre archives et archive est particulièrement fertile. C’est dans cet espace que les utilisateurs non traditionnels des archives, comme les cinéastes de réemploi (dont les œuvres sont constituées d’images en mouvement préexistantes), trouvent leur inspiration. À travers leurs mises en récit, ces derniers montrent ce qui n’est pas visible dans les archives. De la décomposition des matières filmiques, aux dynamiques de pouvoir derrière le geste d’archivage et leurs récits tacites, en passant par les émotions et l’affect véhiculés par les documents, les œuvres nous confrontent à une double dimension inarchivée (ce qui n’est pas archivé) et inarchivable (ce qui ne peut pas être archivé), qui est constitutive de ce que sont les archives et de comment elles se construisent. En étudiant les archives qui constituent les œuvres de réemploi à partir de leur exploitation, c’est-à-dire leurs diverses utilisations et l’ensemble de leurs utilisations potentielles, il est possible de catégoriser ce qui se trouve dans l’inarchivé et l’inarchivable. Ainsi, trois principales modalités émergent des analyses : l’absence, qui relève de la lacune, du fragment et de l’incomplétude ; l’interdit qui se manifeste dans les archives comme traces matérielles ; et l’invisible qui participe de ce qui ne se montre pas. Ces trois catégories relèvent d’un impensé archivistique, c’est-à-dire d’un état de la discipline qui reflète l’inconcevabilité ou l’omission, volontaire ou non, de certains de ses aspects théoriques ou pratiques. C’est en investissant l’impensé, en étudiant l’archivistique à partir des pratiques en marges, qu’il est possible renouveler les discours sur la discipline. / Between archives, as defined by archival science, and the archive as a concept used outside of the discipline, there is a semantic, conceptual and theoretical gap. On one side archives represent the documentary by-product of human activity retained for their long-term value. On the other the archive has become a conceptual and critical tool to address a variety of ideas linked to memory and history. However, this interval between archives and archive is particularly fertile. In this space, nontraditional archives users, such as found footage filmmakers (whose works consist in reusing pre-existing footage) find inspiration. Through the narratives of their work, they show what is not always visible in archives. From the decomposition of film stock to the power dynamics behind archiving and its tacit narratives, through emotions and affect conveyed by records, the artworks confront us with a double unarchived and unarchivable dimension (what is not archived and what cannot be archived), constituent of how archives are created. Studying the archives that are part of found footage works through their usage (exploitation), namely their uses and potentials uses, it is possible to categorize the composition of the unarchived and the unarchivable. Three main divisions emerge from this analysis: the absence, linked to gaps, fragments and incompleteness; the forbidden that manifests in archives as material traces; the invisible that takes part in what is not shown. These three categories have to do with an unconceived (impensé): a state of the archival field reflecting the intentional or unintentional inconceivability or omission of some of its theoretical or practical aspects. In investing in the unconceived, in other words studying archival science from practices on the margins, it is possible to renew ideas and discourses inside the discipline.