Abecedarium B of Hernando Colón: AI-assisted transcription and English rendering (unverified working data)

AI-assisted transcription of the Abecedarium B, the alphabetical catalogue of the library of Hernando Colón (1488-1539), kept in the Biblioteca Capitular y Colombina, Seville. The file has 41,707 records, one per catalogue entry. Each record gives a diplomatic reading, a normalised reading, an English rendering of the title, the register and epitome numbers, imprint place and year, annotation signs and uncertainty notes. STATUS: working data, not a critical edition. The pages were read by language models, and human verification has not been performed. The readings are unverified and must not be quoted as the text of the manuscript. Uncertain readings are marked inside the records rather than smoothed over. Independent checking is planned, and its results will be published in a later version, whatever they show. The deposit contains text only. No manuscript images are included. The file is a gzip-compressed JSON Lines file (decompress with gunzip, then read one JSON object per line). This is an independent project, not affiliated with any institutional research project on these manuscripts. It makes no claim to be a critical edition and no claim about the identification of the printed editions or surviving copies recorded in the catalogue. SOURCE: the transcription was made from digital page images of the 1992 facsimile edition: Hernando Colon, Abecedarium B y Supplementum, Madrid, 1992 (Fundacion Mapfre America and Cabildo de la Catedral de Sevilla; title page: Ed. facsimil de los manuscritos conservados en la Biblioteca Colombina de Sevilla). The images were not made by the author and are not included in this deposit. Rights in the 1992 edition (its reproduction and any editorial matter) have not been cleared by the depositor; only the author's own transcription and analysis are offered under the stated licence. KNOWN ISSUE (column field). The field column was read by a language model from handwritten column headings and contains misreadings. Example: the heading of column 436 (f. 104) was read as 439 for nine rows. An automatic sequence check flags about 250 page halves where column numbers do not run in order; some are legitimate gaps in the manuscript, but several are clear misreadings (for example 921 instead of 97, 1007 instead of 107, 1108 instead of 108, 1358 instead of 358). The share of wrong values has not been measured, and the figure of 250 counts flagged places, not confirmed errors. Do not use column as an exact address without checking the image. What was checked: in four spot-checks against the page images (one reader, no second check) the text and the row numbers were read correctly; the English rendering was not examined. This is not a verification of the file. The whole file remains machine-made and has not been checked by a human, and other fields may also contain errors. Numbers on the second line of an entry are stored in other_numbers, not in registrum; to search by number, search the diplomatic text. Planned: a version 1.1 with corrected column numbers, after a page-by-page check of the flagged headings. No date is fixed. This version (v1) stays available under its DOI.