<p>The paper introduces a new non-factive theory of knowledge, employing the recent theory of verisimilitude developed by Jakob Süskind to model the relative adequacy of representations as their approximation to truth. It makes the relation of representation one of the necessary conditions of knowledge without requiring the truth condition to be satisfied. However, the relation of knowledge to truth is built at the meta-level into the notion of the adequacy of representations, which enables the proposed theory to cover any thinkable case of research yielding real knowledge, although of a non-factive kind. Despite its universality, the theory relativizes knowledge to a set of research questions, while rejecting alethic relativism, which — it is argued — does not provide a satisfactory explanation of truth in reasoning about the growth and extinction of scientific knowledge.</p>