# Machine-Native Reasoning: monograph recovery and release scope

The latest documented prose-refined monograph is 1,155 pages in 97 chapters and 13 parts. Its refinement report survives. Its full editable source archive and full PDF have not yet been recovered. The expected files are `Machine_Native_Reasoning_Prose_Refined_Source.zip` and `Machine_Native_Reasoning_Prose_Refined_Monograph.pdf`.

The complete retained Library inventory (19,938 items), an exact title-only search (32 matches), and the full Core25 formal archive with all nested source ZIPs contain no latest full book. A separate GitHub audit checked all four branch trees, two releases with 24 assets, all seven mainline commits, and reasoning-related search queries; it also found no latest full book. Its evidence and search limits are recorded in `mnr_recovery/github_recovery.json`. The recovered 74-page continuation source and 100-page monograph PDF are predecessors; neither can be presented as the latest full monograph.

The formal selected core and the whole monograph have distinct release scopes. Finishing 27 selected formal counterparts would close the selected-core formal gap, not prove every specialist chapter, validate learned implementations, or close the broader research program. Historical statements of 23/27 or 26/27 must be revised only after final gates pass.

A separately titled reviewed-core reader can be rebuilt faithfully from the authenticated `source/core_v0.2.tex` and its original figures. That deliverable can incorporate current formal correspondence without claiming to rebuild or replace the 1,155-page monograph. The separate selected-core reader is rebuilt from authenticated source. Its formal release status is governed by the final admission record and its pinned proof-gate receipts.

Before a final full-monograph release: recover and hash the exact prose-refined source and full PDF; update only current formal correspondence and release boundaries; rerun its documented exact/reproduction workflow; check proof and display preservation for editorial changes; rebuild and render the full book; confirm navigation, fonts, layout, and all third-party notices. Whole-program research obligations remain described as open where no new evidence closes them.

Source identities and exact recovered archive hashes are in `book_source_recovery.json`.

## Additional local-path check

The prior `/mnt/data` sandbox path is unavailable in this session. Targeted filename discovery under `/workspace` found the older monograph PDF and report, but no `Prose_Refined` source or PDF. This is a search boundary, not a claim that no copy exists elsewhere.

The surviving prose-refinement report names deeper paragraph review of long specialist treatments and reduction of repeated exposition as its next editorial work. Its 334 edit records across 24 active files do not certify a line-by-line review of every paragraph. The older 30-chapter gap register is structurally superseded by the 97-chapter edition and should be reconciled after source recovery. The separate archived Large Signal Model correspondence remains recorded at 39/87 in that report; the selected Machine Theory core's completion does not change that specialist count or turn historical specialist receipts into fresh runs.