Modern LaTeX Editions of Public-Domain Mathematics Manuscripts: Ongoing Project Corpus

This record is the public release point for an ongoing project to turn older, mostly public-domain mathematics and physics manuscripts into modern, inspectable LaTeX-based editions, with translations planned after the original-language typesetting layer is stable. The practical aim is a reusable workflow: gather source scans or source witnesses, produce a readable modern LaTeX display PDF, preserve the TeX and build/QC artifacts, keep the source material close enough for checking, and publish the clean results under one continuing DOI so readers and correction volunteers can find the current corpus without chasing scattered files. Status at a glance, 2026-05-26: Per-work artifact packaging: [##########] 100% (31/31) Top-level direct PDFs obey scan policy: [##########] 100% (30/30) -- no loose scan/reference PDFs are exposed as top-level reader PDFs. Full-page render audit coverage of v10 direct PDFs: [##########] 100% (9762/9762) Automatic layout pass after full-page audit: [##--------] 23% (7/30) Needs human layout review but structurally renders: [#######---] 67% (20/30) Repair/demotion candidates from the stricter audit: [#---------] 10% (3/30) -- 02-hecke-vorlesungen, hermite_oeuvres, and poincare_oeuvres. Hecke is readable but tight-margin flagged; Hermite and Poincare need stronger repair/review. Artifact-only work without a direct LaTeX reader PDF: [----------] 3% (1/31) -- darboux_lecons_geometrie. Public translation layer: [----------] 0% (0/31) -- translation is a later layer after original-language TeX stabilizes. Non-Eurocentric/Central Asia source handoff prepared locally: [##########] 100% (50/50) -- 50 PDFs under 100 MB each, not yet part of the Zenodo public corpus. SGA 1-4 coordination/source packet prepared locally: [##########] 100% (4/4) -- downloaded for project coordination, not included in this public-domain corpus release. Where the public corpus stands: the current public record contains artifacts for 31 works. Thirty works currently have standalone reader-facing LaTeX PDFs named pdf__<work>.pdf. Each work has a matching artifact package named artifacts__<work>.zip. The artifact ZIPs contain the generated TeX/source layer, available scans or source witnesses, provenance/QC material, and per-work metadata. A bulk ZIP, full_repo__modern_latex_corpus_v10_latex_only_direct.zip, is also provided for people who want the complete current public corpus at once. Packaging rule: only LaTeX-generated reader PDFs are exposed as top-level pdf__... files. Original scans, scan-backed repairs, and source/reference PDFs are kept inside artifact ZIPs. This version corrects the previous Darboux scan-backed direct PDF: the readable Darboux scan remains available inside artifacts__darboux_lecons_geometrie.zip, but it is no longer a standalone top-level PDF because it is a scan, not a LaTeX-produced edition. Quality/completion status: this is a working machine-assisted corpus, not a final critical edition. v10 fixed the top-level scan problem and all direct PDFs render structurally, but a stricter full-page layout audit run after publication shows that only 7 of the 30 direct PDFs currently auto-pass layout checks. Twenty are usable enough to inspect but need human layout review, and three are repair/demotion candidates before they should be treated as clean reader editions. Future versions should either repair weak PDFs or move them to artifact-only status until their LaTeX display layer is good enough. How to use the files: readers who want a modern typeset PDF should download a top-level pdf__... file, but should treat the corpus as a working draft unless the manifest or later release notes mark the item clean. Readers or editors who want to check, rebuild, correct, or extend a work should download the corresponding artifacts__... ZIP. Users who want the whole current release should use the full-repo ZIP. The manifest, summary, and top-level scan audit describe the included works, file hashes, packaging decisions, and QC status. Current local work beyond this public corpus: local project folders contain web-upload packets for Kimi/web-agent source dumps, a 50-PDF non-Eurocentric/Central Asia source-manuscript handoff, a downloaded SGA 1-4 source/project packet, cleanup-6 partial individual downloads, and continuing quality audits of the current Zenodo PDFs. Those materials are not automatically pushed to the public corpus until they have coherent public-domain provenance, a clean or clearly marked reader-facing LaTeX surface, and artifact packages. Provenance policy: the public record is meant for public-domain source works, scans of those original works, and machine-assisted transcriptions/typeset editions of those works. Scans of original public-domain books or papers are useful reference material, but they belong in artifacts unless and until they are replaced by a clean LaTeX display PDF. Modern collected-work packages are handled cautiously: publisher/editorial matter and exact modern packaging are excluded when necessary, while original public-domain papers are kept as coherent per-paper scan/transcription units.