What is constant in logic?

The thesis opens with an account of the origins of the modern problem of the demarcation of the logical constants in Alfred Tarski's 1936 lecture "On the Concept of Logical Consequence". Focus then shifts to the logical notions, to free the investigation from parochialism. John Etchemendy's recent challenge to the demarcation project is discussed and rejected. The thesis then surveys various methodological issues, conceptions of the aims and purposes of logic, and needed technical concepts. This provides the requisite tools to carry out the evaluation of existing demarcation proposals. Three groups of demarcation proposals are elaborated and evaluated.First are model-theoretic proposals, which attempt to define the logical notions as those whose extensions are held fixed under various classes of morphisms on the domains on models for logic. Next are the proof-theoretic proposals which attempt to define the logical notions as those whose meanings can be fixed by inference rules. Last, various other proposals which do not fall into any general grouping are discussed. In all cases, the proposed demarcations are rejected, either due to internal technical considerations, or because the demarcation offered will not suffice to meet the aims and purposes of logic.Finally, the thesis turns to arguments that there could not be a unique demarcation of the logical notions which serves all of the central purposes of logic. Further, it is argued that disputes over the nature of logic are subject to a particular intractability, as a dispute over logic is, at least in part, a dispute over the norms under which disputes are to be carried out; thus we have strong reason to doubt that there could be a demarcation that would resolve the debates. I conclude that logic is, at base, an irreducibly interest-relative and pragmatic affair