SOURCE_MANIFEST — Wattles KB v1

Proof work: «The Science of Getting Rich» (SGR), Wallace D. Wattles, 1910. English is
the ORIGINAL text, so it is the sole evidence pool; the RU on the pages is a reading
layer, marked «перевод наш», and is not gated for verbatimness.

Layer Edition Where fetched PD status
EN (evidence pool) The Science of Getting Rich by W. D. Wattles — Project Gutenberg eBook #59844, release date June 30, 2019. Produced by Craig Kirkwood and the Online Distributed Proofreading Team from images made available by The Internet Archive. sources/gb_59844.txt (raw Gutenberg plain-text) Wattles d. 1911 → work is deep public domain (life+70 long elapsed; first published 1910); Gutenberg US-PD edition.
RU (reading layer) Наш перевод с EN SGR, «перевод наш» на страницах тела заметок в concepts/, principles/, chains/, MOC.md наш текст, local KB, не gated

Not used as evidence: Wattles' other books (The Science of Being Well, Gutenberg
#33917; The Science of Being Great) and the Nautilus magazine articles — verified
PD but not ingested; each gets its own corpus when fetched (see CORPUS_BOUNDARY.md).
Later New-Thought reformulations / abridgements (1976 and modern repackagings) are
meme-distorted with no clean edition trace and stay out of the pool.

Files

  • sources/gb_59844.txt — raw Gutenberg fetch (sha256 adc3fb17…252744).
  • sources/build_corpus.py — deterministic build: slices the body between the Gutenberg
    START/END markers, cuts the publisher back-matter at FURTHER AIDS TOWARD GETTING RICH RIGHT (NAUTILUS advertisement + transcriber notes — excluded per
    CORPUS_BOUNDARY.md), detects PREFACE / CHAPTER I–XVII headings, drops TOC-only marker
    lines, and splits each chapter into whitespace-normalized paragraph units keyed
    sgr.<chap>.<n> (pref, then 1…17).
  • sources/sgr_en_units.json — canonical EN evidence pool: 497 paragraph units,
    preface + chapters I–XVII (sha256 c4a3f538…fc3531). Every verbatim > EN: "…" line in
    the vault must be a substring of the cited unit.
  • sources/check_quotes.py — verbatim gate: each > EN: "…" line must be a literal
    substring (after whitespace/quote/dash normalization) of the pinned unit sgr.N.M.

Reproducibility

cd sources && python3 build_corpus.py   # rebuilds sgr_en_units.json from gb_59844.txt
cd sources && python3 check_quotes.py    # verbatim gate: expects failures: 0

build_corpus.py is deterministic (fixed input file, fixed markers, fixed heading regex),
so re-running on the same gb_59844.txt reproduces the identical 497-unit JSON. The
back-matter cut makes CORPUS_BOUNDARY.md's "back-matter excluded" claim literally true:
the NAUTILUS ad and transcriber notes are no longer present as units.

Corpus boundary note

The last genuine Wattles unit is sgr.17.9 (close of the Summary chapter). Everything the
old build emitted from sgr.17.10 onward (FURTHER AIDS TOWARD GETTING RICH RIGHT, the
NAUTILUS subscription ad, and Transcriber's Notes: … End of Project Gutenberg's…) was
publisher/transcriber back-matter and is now excluded — dropping the unit count from 514 to
497.