From the time of Nietzsche at least we can identify the emergence of an emphatic scepticism concerning the fundamental idea of transcendental philosophy, namely that there is a single world to be known. Rorty (1978) criticises the notion that there is one epistemically accessible world in the context of a brief history of modern philosophy narrated in terms of two broad phases. In the first phase, that of re-transcendentalisation, we find several very different attempts to renew transcendental philosophy by identifying conditions of knowledge that are independent of experience, as in Peirce's pragmatism, Husserl's phenomenology, Heidegger's Being and Time, Witggenstein's Tractatus, or the earlier thought of Russell. But this eventually gave way to a second phase of de-transcendentalisation. In the development of pragmatism from Peirce to Dewey, in the movement within phenomenology towards a philosophy of "the Other", in the developments within analytical philosophy that led to Quine and Sellars, and then to Davidson and Putnam, in the path that led Wittgenstein from the Tractatus to the Philosophical Investigations, and in Heidegger's gradual turn from fundamental ontology to a philosophy of "commemorative thinking", we can recognise a common thread. In every case these thinkers came to renounce the quest for an a priori conceptual structure or an Archimedean point or foundation for all knowledge. According to Rorty's account, Kant's hope of finally bringing philosophy onto "the sure path of a science" has effectively yielded instead to a kind of epistemological behaviourism. Thus Quine (cf. 1960 already, and 1981: 2) presents three arguments that contest the assumption of a priori foundations. For he claims that truth and knowledge are essentially issues to be decided by scientific rather than philosophical considerations, that the conceptual structure in terms of which we in fact know reality is only one of several possible such structures, and finally that philosophical questions are entirely dependent upon context.