The underdetermination of theory by data is the view that there will always be more than one theory consistent with any body of empirical data. A stronger version claims that this would be so even if the data were to include all possible empirical data. If these theses are purely logical in form, they do no more than claim that the data will not entail that just one theory is the true one, which will be so provided that the content of the theory is more than a mere consequence of the data. A yet stronger version (methodological underdetermination) claims that even all possible data plus the best canons of scientific explanation (simplicity, good inductive sense) will not yield one theory as the unique best choice. The thesis in this form is frequently cited in support of relativistic and social constructivist theories of science. It requires confidence in the distinction between theory and evidence, in our ability to understand what it would be for a theory really to be adequate to all empirical data, and finally in our ability to distinguish genuinely different theories from notational variants of the same one theory.