Gian-Carlo Rota’s conception of mathematical identity is rooted in his reading of Husserl’s transcendental and eidetic phenomenology, and especially in the practice of eidetic variation. Beginning with Rota’s 1964 “conversion” to phenomenology, the paper reconstructs his account of identity as eidos: not a fixed form, but the contextual pole of an open-ended activity of identification, distinct from its changing facticities and disclosed only when one attends to the “how” of givenness rather than to the “what” alone. Carried over to mathematics, this account yields a strong thesis: a mathematical object such as the real line retains one and the same identity across radically different axiomatic presentations, none of which can exhaust it. Mathematical identities are consistent yet never complete, and every formal presentation presupposes a “pre-axiomatic grasp” of the object. Rota’s notion of cryptomorphism and his analyses of the Reynolds (averaging) operator, of quantum logic, of multisets and of probability theory are read as so many tests of a ternary relation between syntax, semantics and identity. The paper argues that this relation, and the eidetic structure it presupposes, are required to render the inner historicity of mathematics intelligible, and it situates Rota’s project within Husserl’s call for a reform of logic, against reductionism and objectivism.