Recovered 1999 parameter evidence
=================================

New in this pass
1. Recovered the complete original 50-page SEC2 Version0.6 PDF, printed October1999,
   by following the public CiteSeer URL's redirect to a Wayback PDF wrapper and then
   the playback iframe explicitly present in that wrapper. No guessed archive timestamp
   was used. The saved original bytes have SHA256:
   530dbfd6005b81ad960289dcf20b32824ed2d83160d58ddeefcf1f93d9919f2f
   File: sec2-v06.pdf. Receipts: sec06-fetch-receipt.json and sec06-iframe-receipt.json.
   Original: https://citeseerx.ist.psu.edu/document?doi=2840bfaf7cf8f97deeaddcd4274e82863ec522ee&repid=rep1&type=pdf
   Archive: https://web.archive.org/web/20240414044829if_/https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=2840bfaf7cf8f97deeaddcd4274e82863ec522ee
   Printed cover inspected visually (sec06-front-01.png). Embedded CreationDate is
   D:19991024144018, Title gec1.dvi, dvips(k)5.78, HP-UX Acrobat Distiller3.01.
   These metadata corroborate the cover but are not independent proof of publication date.

2. Recovered exact numeric pages from indexed excerpts of GEC1 Version0.5, printed
   September1999, at ONE primary document URL (both cover and all target pages):
   https://citeseerx.ist.psu.edu/document?doi=fe42ff586c6b5bd314cee6577221a2675319ed5e&repid=rep1&type=pdf
   secp160k1 p9; secp192k1 p11; secp224k1 pp12-13; secp256k1 p14;
   sect113r1 p20; sect113r2 pp20-21; sect131r1 p21; sect131r2 p22;
   sect193r1 p25; sect193r2 pp25-26. See cached-page-manifest.json.
   Breakthrough: requiring the literal phrase GEC1 excluded odd pages because only
   even-page running headers contain that phrase. Querying the document title plus
   exact seed or parameter words exposed the missing odd pages.

3. All10 target curves in each document (20records) passed field-by-field comparison
   against the previously independently verified SEC2v1 parameters. Checks include
   field prime or degree+reduction polynomial, a,b,n,h, both full/compressed G, and
   same-curve seed S for all six binary curves. Public arithmetic rechecks on-curve,
   nG=infinity, and recomputes both cofactor2 halves or the unique prime-order half.
   Every recomputed half equals the prior verified result. Result and reproducible
   parser: historical-parameter-checks.json, verify_historical_parameters.py.

   Thus the shared binary start structure and the four prime k1 half-point pattern
   are already present in the September1999 draft model according to its cached
   primary text; the complete October1999 PDF gives stronger preserved-document
   evidence. This narrows chronology. It does NOT recover the historical recipe.

What generation discussion says
GEC1v0.5 printedp3 and SEC2v0.6 printedp3 both describe selecting curves with a
computable endomorphism until a prime-order curve is found. Their verifiably-random
curve discussion describes SHA1 from seeds and repeated candidate selection/point
counting. GEC1v0.5 printedp17 and SEC2v0.6 printedp20 both specifically name Schoof's
algorithm for the binary curves and say a is random or1. The substantive procedure
text matches; neither inspected discussion supplies a seed-to-G generation recipe.
Do not treat their claims about verifiably random curve coefficients as a claim that
the listed S directly specifies G. Curve definitions label S as the seed for E, then
separately publish G.

SEC2v0.6 references include ANSI X9.62-1998, X9.63 October1999 working draft,
IEEE P1363 July1999 draft, NIST July1999 recommended curves, and SEC1 September1999
working draft. Its only appendices are ASN.1 syntax and references; no hidden
generator-generation appendix appeared in this full document.

Limits / old versus new
The GEC1v0.5 existence/title/frontmatter/TOC was already known from prior work; this
pass adds recovered numeric pages and direct parameter comparisons. SEC2v0.6 full
PDF and this exact CiteSeer document identity are new to this investigation. No
claim of first-ever discovery is made. The mathematical patterns themselves and
our earlier compatibility checks are not newly discovered archival statements.

The GEC1 fullPDF remains unavailable: ordinary exact-URL GET404; Exa CRAWL_NOT_FOUND;
Wayback availability lookup503. The numerical GEC1 evidence is a search-index
extraction, not authenticated original PDF bytes. Its printed date comes from its
cached cover. Search crawler labels like 'Published8monthsago' are unrelated to the
historical date. Source excerpts have hashes and exact-URL provenance in the manifest.
The complete SEC0.6PDF is an archival copy captured in2024, not a1999 server capture.

No GEC1v0.4 numeric pages or original generating program were recovered. The exact
shared starting values' origin is still unresolved. Earlier broad archive inventories,
known1999 citations, and the existing arithmetic are background, not new findings.

Reproduce from this folder using Python with stdlib (math class imported from sibling):
  python verify_historical_parameters.py
PDF extraction helper separately requires pypdf. Its historical subset font has
nonstandard glyph names; the explicit map is saved in the result, and printedp23
was rendered to check it. A separately extracted Poppler text was obtained by root
for an independent crosscheck.
