In 1727 Isaac Newton died as one of the world's most famous natural philosophers. Remarkably, of the many draft manuscripts he left, only a few concern what we now call physics or mathematics. Instead, they contain thousands of pages on alchemy, theology and chronology or world history. Newton compiled experimental reports of his alchemical experiments, copied rare alchemical tracts, wrote lengthy treatises defending a heretical form of Christianity and rewrote the history of ancient empires. And not just once: he painstakingly composed sentences, paragraphs, and chapters, deleting, adding, starting all over, to the extent that we sometimes have over 50 versions of a single paragraph. Ordering these manuscripts, or even reliably dating them, has hitherto been an impossible task. And apart from a few selected short documents, no one has yet dared to edit them.
Anno 2014 the majority of these draft manuscripts, which amount to over 6 million words, have been transcribed by the Newton Project using TEI P5 and MathML. For the first time in history we can actually start to analyse the logical structure of the many projects that Newton worked on, and use his manuscripts to catch a glimpse of his working practices. Classical notions of editing, which have so far served digital editors rather well, are difficult to apply to this vast and complex Corpus Newtonicum. Moreover, they wouldn't do justice to the richness and depth of Newton's research. Using the concept of Peter Shillingsburg's knowledge site, this paper explores some of the challenges and possibilities of digitally editing Newton's draft manuscripts, the fading borders of various types of editions in the digital world and the changing role of the editor.