SGA 1-7: English Readers, French Texts, TeX Archives, and SGA7 Source Transcriptions

Start here: The first file is one ZIP containing the current cumulative English reader PDF for each of SGA 1 through SGA 6 together with its complete buildable TeX closure. The same six reader PDFs and master TeX files remain directly accessible, and SGA1 remains the default browser preview. SGA3: The preferred English reader is the clean R29 cumulative, 1,470 A4 pages covering the Introduction, Exposes I-XXVI, the Tome-I subject index, the Tome-III mathematical guide, and the terminal index. It has 13,119 named destinations and 12,337 valid internal GoTo actions. Its complete buildable TeX closure and reference/QA controls remain grouped separately. SGA7 I: This successor provides a complete working source transcription of all six written exposes in the volume: I, II, VI, VII, VIII, and IX. Exposes III-V are not separate written exposes in SGA7 I. The 528 continuous source-page markers compile as a 267-page A4 reader with 162 native diagrams and no raster diagram inputs. The source language is preserved as printed, including the English Expose VI. This is not a complete English translation. Two compact SGA7 visual-evidence archives retain 12 targeted high-detail source crops actually opened during transcription and diagram review, plus provenance metadata for the larger rights-blocked scratch surface. The parent scan and whole-page renders are excluded. English readers for SGA 1 through SGA 6 are presented before available French texts, direct editable TeX masters, and grouped supplementary archives. Earlier Zenodo versions remain immutable history. These are working scholarly translations, editions, or transcriptions, not critical editions, peer review, accessibility certification, rights determinations, or mathematical certification. They do not transfer rights in the underlying works. SGA7 II working French source transcription: this successor extends the direct reader from Exposes X-XIV through Expose XVII: 130 A4 pages covering scan indices 8-260 / book folios 1-253. It also preserves an explicitly incomplete Expose XVIII continuation outside the reader. This is not a complete SGA7 II volume or critical edition. Actual recovered source-image evidence: a compact archive contains four full 600-dpi source pages and six labeled 600-dpi detail crops used to recover scan indices 197, 198, 211, and 212. These are pixels derived from the publicly available Number12 scan, not reader screenshots. Page, folio, dimensions, DPI, bounding box, parent-scan hash, and exact image hashes are recorded. The existing larger high-detail Number12 image archive remains available unchanged. Actual lead-opened SGA7 source crops: this successor adds one compact archive containing 29 scan-derived PNGs used directly during SGA7 I and SGA7 II transcription checks: 23 SGA7 I images and six SGA7 II images. They include page regions and high-detail formula, arrow, symbol, diagram, and wording crops from the publicly available parent scans. They are actual source pixels, not reconstructed-reader screenshots or metadata-only placeholders. The index records parent-scan hash, page and folio, recovered crop box and DPI, dimensions, linked TeX unit, read events, and exact image hash. Four exact pixels already in the public SGA image archives were excluded. SGA7 II Expose XVIII source-work preservation: this successor updates only the compact SGA7 II source/WIP archive. The direct Exposes X-XVII reader and master remain byte-identical. The ZIP now preserves 47 continuous Expose-XVIII source pages, scan indices 262-308 / book folios 255-301; indices 309-334 remain absent, so the Expose is explicitly incomplete and excluded from the direct reader. The bounded WIP compiled in isolation with zero fatal errors, one disclosed overfull box, and is not represented as reader-level release closure. Actual SGA7 I post-cutoff source crops: this successor adds one compact archive containing 260 scan-derived PNG regions used during source transcription and adjudication. These are actual pixels from the publicly available SGA7 I scan, not screenshots of a reconstructed reader and not a metadata-only substitute. The index binds each image to the parent-scan hash, source page and folio, dimensions, linked TeX unit, exact SHA-256, and recovered generation metadata. Nine exact pixels already public were deduplicated; five redundant whole-page or stretch variants were catalogued but omitted.