Trendelenburg argued that Kant's arguments in support of transcendental idealism ignored the possibility that space and time are both ideal and real. Recently, Graham Bird has claimed that Trendelenburg (unlike his contemporary Kuno Fischer) misrepresented Kant, confusing two senses of . I defend Trendelenburg's : the ideas of space and time, as a priori and necessary, are ideal, but this does not exclude their validity in the noumenal realm. This undermines transcendental idealism. Bird's attempt to show that the Analytic considers, (...) but rejects, the alternative fails: an epistemological reading makes Kant accept the alternative, while an ontological reading makes him incoherent. As I demonstrate, Trendelenburg acknowledged the ambiguity of , focusing on the transcendental, not the empirical sense. Unlike Fischer, Bird denies Kant's commitment to things-in-themselves in favour of a descriptivist, non-ontological reading of transcendental idealism as an inventory of . But neither Bird's descriptivism, nor Fischer's commitment to things-in-themselves, answers Trendelenburg's sceptical worry about transcendental idealism. (shrink)