Modern LaTeX Editions of Public-Domain Mathematics Manuscripts

This record is the public front page for an ongoing project to produce modern, inspectable LaTeX editions of older mathematics and physics manuscripts, especially works that are public-domain or otherwise suitable for public scholarly transcription. The immediate goal is a clean original-language typesetting layer: readable LaTeX PDFs, corresponding TeX/source files, source-witness scans or references, provenance notes, and quality-control manifests. Translation is a later layer once the source-language editions are stable, but this version also adds a practical SGA translation/source handoff packet because that material is useful for parallel translation work. The project is deliberately a working corpus, not a final critical edition. Machine-assisted transcription, OCR cleanup, layout repair, source comparison, and human review are all in progress. Files are published here when they are useful for inspection and correction, but the manifests should be consulted before treating any individual item as polished. At-a-glance status for this version: Current published release surface: [##########] 100% carried forward from v17. No whole works were removed in this SGA addition pass. Represented legacy work packages: [##########] 31/31 current per-work artifact packages from v17 are present. Reader-facing legacy LaTeX PDFs: [#########-] 28/31 legacy packages have standalone top-level PDFs, unchanged from v17. Legacy artifact ZIP coverage: [##########] 31/31 represented legacy packages have per-work artifact ZIPs with TeX/source material, provenance/QC files, and available reference material. Top-level reader PDF configured hard-marker check after v17: [##########] 28/28 legacy reader PDFs reported zero configured hard publisher-marker hits after the v17 cleanup. SGA 1-4 source/translation starter coverage: [#########-] present for 1-4, uneven by volume. SGA 1-2 have modern French arXiv PDF/TeX baselines; SGA 3 has Polo/Gille re-edition/reference PDFs and English LLM-derived Markdown from a public repo snapshot; SGA 4 has Orgogozo/Laszlo French PDF/TeX sources plus new cumulative English translation batches for Expose I through sections 6-7. Public English translation snapshots are included where found. SGA 5, 6, 7-I, 7-II working LaTeX coverage: [######----] usable working source packet. Four combined TeX files, page-sliced TeX, working combined PDFs, and reference scans are included for checking and continuation. These are not proofed editions. SGA handoff ZIP integrity/PDF audit: [##########] 138/138 PDFs in the base local SGA handoff packet passed pdfinfo before staging; the v20 added SGA translation ZIPs also passed ZIP integrity checks before inclusion. Transcription/layout polish: [###-------] early working-draft stage. Some PDFs are readable draft editions, some TeX still needs compile/layout repair, and all mathematical content remains open to proofreading against witnesses. Non-European / Central Asian source-manuscript release layer: [####------] first cleaned release layer. v21 adds 22 cleaned non-European combined reader PDFs as individual top-level files plus a cleaned artifact ZIP containing 85 locally audited PDFs, TeX files, manifests, and cleanup notes from the KIMI5/web-session batch. Overall long-term corpus goal: early proof-of-concept subset. The percentages above describe this Zenodo release surface, not completion of every work by every relevant author. What changed in v18 while preserving availability: the v17 corpus was copied forward in full. This version adds top-level SGA reader PDFs for SGA 1, SGA 2, SGA 4, and working SGA 5-7 compilations, plus 10_artifacts__sga_translation_handoff_1_7.zip containing SGA 1-4 baselines, public translation/source repository snapshots, SGA 5-7 working LaTeX, reference scans, provenance pages, and manifests. The full-repo ZIP was rebuilt from the current v18 payload. What changed in v19 while preserving availability: the v18 corpus was copied forward in full. This version adds 00_pdf__sga3_schemas_en_groupes_polo_gille_modern_french.pdf, a top-level reader PDF assembled from the Polo/Gille SGA3 re-edition component PDFs exposed at the SGA3 page. This closes the top-level SGA 1-4 reader gap: SGA 1, 2, 3, and 4 now all have reader-facing PDFs while the SGA artifact ZIP retains component PDFs, scans/reference material, translation snapshots, and provenance. What changed in v20 while preserving availability: the v19 corpus was copied forward in full. This version replaces 10_artifacts__sga_translation_handoff_1_7.zip with an updated cumulative SGA handoff artifact. The updated artifact adds 06_NEW_SGA_TRANSLATION_BATCHES/, including the initial SGA translation plan packet and the cumulative batch 003 packet with SGA1-3 English Markdown/TeX snapshots, SGA4 Expose I English LaTeX through sections 6-7, rendered PDFs, source extracts, worklogs, validation notes, and render-check images. The top-level file count remains stable and the full-repo ZIP was rebuilt from the current v20 payload. What changed in v21 while preserving availability: the v20 corpus was copied forward with one reader-surface demotion: 00_pdf__gauss_werke.pdf is no longer presented as a clean top-level PDF because it is a rough 1,793-page OCR/stitched working draft, but the material remains preserved inside 10_artifacts__gauss_werke.zip and inside the release-history ZIP. The former individual 80_* audit/report files through v20 were packed into 80_metadata__release_history_and_audits_through_v20.zip to reduce front-page clutter without losing history. This version adds 22 cleaned non-European combined reader PDFs as 00_pdf__non_eu__... files, adds one clearly marked partial Cayley incremental reader PDF, and adds 10_artifacts__cleanup9_kimi5_non_eu_cayley_redo_cleaned.zip. Local audit of the cleaned cleanup 9 tree reported [##########] 85/85 PDFs passing pdfinfo and text extraction after removing placeholder/bad PDFs and rebuilding the Cayley all-in-one PDF. How the files are organized: 00_pdf__... files are reader-facing PDFs. They are placed first so individual PDFs can be found quickly. 10_artifacts__... files are per-work or per-project artifact ZIPs. These contain TeX/source material, provenance/QC files, and available reference material for checking or rebuilding the work. 80_... files are current manifests, audit summaries, replacement reports, packaging notes, and cleanup queues. Older v20-and-before 80_* files are preserved together in 80_metadata__release_history_and_audits_through_v20.zip. 99_full_repo__... is a bulk ZIP for people who want the whole current release at once. Quality and cleanup policy: top-level PDFs should be generated reader PDFs or modern typeset re-editions rather than loose scans. Scans and reference PDFs belong inside artifact ZIPs. Publisher or collected-edition front matter, title pages, prefaces, ISBN/copyright pages, and running headers are being stripped or replaced as packages are rebuilt. Availability remains the higher priority: cleanup should produce replacement files rather than dropping whole works from the current public corpus. Use guidance: readers who only want a typeset draft should download a 00_pdf__... file. Editors, proofreaders, or downstream agents should download the corresponding 10_artifacts__... ZIP. For SGA continuation, use 10_artifacts__sga_translation_handoff_1_7.zip; it is the consolidated packet with TeX, scans/reference PDFs, public translation snapshots, and source URLs. Forward plan: continue consolidating Kimi/web-session outputs, repair compile/layout failures, cut collected works into clean per-paper or per-work units, strip remaining publisher apparatus, add non-European and Central Asian materials after transcription, and then add translation layers once the original-language TeX is stable.