This paper substitutes a structural hermeneutics for the sequential approach to the Tractatus and its manuscript, which leaves the manuscript completely unexplained. All the pieces come together only if we recognise that Wittgenstein's book is a hierarchical object that was composed following a top-down strategy. Therefore, we realise that the “summary on scattered sheets” represents the perspicuous version of the manuscript, the correlations with the Notebooks become obvious and the “supplements” can be reconstructed. A correct dating of the composition allows (...) us to understand how it was possible, without digital technologies except from an ingenious application of the decimal system, to achieve this formal masterpiece. (shrink)

I argue that the numbering system of the Tractatus lets us see how it was constructed, in two closely related senses of that term. First, it tells us a great deal about the genesis of the book, for the numbering system was used to assemble and rearrange a series of drafts, as recorded in MS 104. Second, it helps us understand the structure of the published book, as cryptically summarized in the opening footnote. I also discuss an unpublished letter from (...) Anscombe to von Wright from 1948 which contains the very first sketch of a tree-structured reading, and what I believe is Stenius’s response to Anscombe’s proposal. The paper critically evaluates previous work on tree-structured readings and contends that we need to read the Tractatus in both the number order used in the published book and the tree order that Wittgenstein used to draft it. It also considers some of the main ways of turning this complex branching structure into a linear, printed text, and so serves as an introduction to the three tree-structured editions of the Tractatus that accompany this paper (the German text, and the translations by Ogden/Ramsey and Pears & McGuinness). (shrink)