In 1938 a number of sheets of notes were sent by Paul Léon, then acting as Joyce’s secretary, to Harriet Shaw Weaver, and ultimately deposited in the British Museum. These sheets were transcribed by Phillip Herring in James Joyce’s Notesheets in the British Museum (1972). They comprise nearly 3000 notes, of which approximately 2000 entries contain examples of period diction, and the remainder relate to embryology, the history of the English language or detail from previous episodes. Joyce then used the information he amassed to write ‘Oxen’, striking through entries on the notesheets as he incorporated them in successive drafts.
The "notesheets" do not seem to be notecards, but of "double (that is folded) sheets." The notesheets for the "Oxen of the Sun" are sheets of paper that contain about 2000 entries that Davison describes as "localized chaos." He is also said to have used little scraps of paper from writing blocks made for the waistcoat pockets on which he noted things he heard or overheard. These materials were then transferred to the larger notesheets. Material used from the notesheets were struck out (using colored pencils, usually in red, blue and green.
Joyce also kept some sixty Notebooks (as in the case of Finnegan's Wake, for instance). He appears to have used some of them over a long period of time. Some of them are alphabetical, containing quotes, remarkable sentences, descriptions of people, places and episodes.
His "workflow" has been described as consisting of eight steps: 1. raw notes, 2. notesheets, 3. raw drafts, 4. final drafts, 5. typescripts, 6. copy-edited typescripts, 7. page proofs and 8. final proofs.
1. See also Thomas E. Connolly, The Personal Library of James Joyce (Buffalo: Norwood Editions, 1978), a book I have to get.