# Phase 5 — Skeptic's Brief: The Case AGAINST the Hoax Verdict **Charge:** Mount the strongest possible opposition to the ~99% hoax posterior for Beale Cipher 1. Below is what I found after a genuine effort to break the verdict. --- ## Strongest single counter-argument **Nick Pelling's "double-encipherment / signal-in-the-string" hypothesis, as developed across his Cipher Mysteries posts (2010, 2013, 2018).** Pelling concedes Gillogly's statistic — the `DEFGHIIJKLMMNO`-style run *is* astronomically improbable under a uniform null — but argues this is evidence of a **signal, not noise**. His specific claim: B1 was constructed by feeding plaintext through a homophonic substitution table *first*, and the Declaration was used as a *second* layer; the Gillogly string is what bleeds through when you partially decrypt with the wrong (single-layer) key. He further argues the same homophone table underlies B2 and B1, just used differently, and that "places where the patterns break down" can be exploited to **refine** the cipher table rather than dismiss it. This matters because **the user's Phase 3/4 only tested single-layer book substitution and short Vigenère over book output**. Pelling's hypothesis explicitly predicts that those tests will fail even on a genuine cipher. If correct, the user's null results provide essentially zero evidence for the hoax. Source: [Jim Gillogly's Beale sequence revisited (Pelling, 2010)](https://ciphermysteries.com/2010/06/22/jim-gilloglys-beale-sequence-revisited); [Refining Beale Cipher B1's cipher table (2018)](https://ciphermysteries.com/2018/11/15/refining-beale-cipher-b1s-cipher-table); [The Improbable World of Beale Cipher B1 (2018)](https://ciphermysteries.com/2018/12/22/the-improbable-world-of-beale-cipher-b1). **Assessment:** Pelling is an expert blogger, not peer-reviewed, but he is the most technically literate active dissenter. The hypothesis is unfalsified by what the user has run so far. That is genuine epistemic weight, though not enough to flip the verdict. --- ## Secondary counter-arguments 1. **Carl Hammer (Sperry-UNIVAC, 1970/1971, reported by Kruh and others).** Using UNIVAC 1108 statistical tests (CRYPTA/CRYPTS/CRYPTT), Hammer concluded B1 and B3 "are not random doodles but do contain intelligence and messages of some sort." This is the canonical pro-genuine result. *Assessment:* Hammer's tests pre-date the Gillogly anomaly, were not designed to detect a deliberately-constructed hoax that mimics non-random structure (precisely what an alphabet-ladder construction *would* produce), and have been substantially eroded by Campanelli's 2022 Cryptologia paper showing B1/B3 fail Benford structure that B2 satisfies. Reference: [Wikipedia summary](https://en.wikipedia.org/wiki/Beale_ciphers); [Campanelli, *Cryptologia* 47(5), 2022](https://www.tandfonline.com/doi/abs/10.1080/01611194.2022.2116614). Weight: low — but historically important. 2. **Transcription-artefact hypothesis (the Ward-as-source problem).** The original Beale numbered papers do not survive; everything traces to Ward's 1885 pamphlet. The pamphlet's printed Declaration contains **23 deviations from the actual Declaration**, **126 vs 103 commas**, and at least **6 separate counting errors that affect over 130 code references in B2**. If the printer/transcriber made structured errors (e.g., systematic off-by-one in a numerical range), an alphabet-ladder could arise from a *real* cipher whose intended decryption against an *unmodified* Declaration would be English. *Assessment:* the user should re-test Phase 3 with **un**-modified Declaration text and with small global/local index shifts before treating the alphabet string as load-bearing evidence. Reference: [Cipher Foundation transcription notes](https://cipherfoundation.org/older-ciphers/beale-papers/beale-papers-transcription/); [Wikipedia](https://en.wikipedia.org/wiki/Beale_ciphers). Weight: medium — this is a genuinely under-explored hole. 3. **Composite-cipher schemes the user did not test.** Phase 4 covered only Caesar shifts and short Vigenère keywords *on the book-decoded letters*. Untested: - **Per-position parity key** (different key for odd vs even cipher numbers). - **Hill cipher** over the decoded letters (a 2011 cryptanalysis cited via Wikipedia's "Cross-off list" found 2×2/3×3 Hill matrices with block lengths 3 or 7 were the most plausible hand-computable variants — and have *not* been exhaustively swept). - **Columnar transposition** of the book output. - **Modified Declarations**: starting count from word 2; omitting articles; using only initial-letter mapping for some words and full-word for others (the latter is *exactly* what B2 does). *Assessment:* this is the legitimate residual hypothesis space. The user's 99% claim implicitly conditions on composite schemes being negligible *a priori*; that prior deserves justification. References: [Adelaide Cipher Cross-off List](https://www.eleceng.adelaide.edu.au/personal/dabbott/wiki/index.php/Cipher_Cross-off_List); Fitzgerald's own README notes these were tested by him with null results — see [github.com/david-fitzgerald/beale-ciphers](https://github.com/david-fitzgerald/beale-ciphers). 4. **No peer-reviewed paper has formally overturned Gillogly, but no peer-reviewed paper has *replicated* the exact p-value either.** Gillogly's 1980 *Cryptologia* note ([Semantic Scholar](https://www.semanticscholar.org/paper/The-Beale-Cipher:-a-Dissenting-Opinion-Gillogly/d30667d742b82bbbe5cb5c88cde3aaea0a0103ab)) uses a unigram-frequency null. Under a richer null that allows for *any* hand-encoder's bias (e.g., reaching for nearby numbers under cognitive load, alphabetizing one's wordlist as a working aid for a *real* encipherment), the p-value would be larger. Nobody has, to my knowledge, published the calculation under a defensible "competent-but-tired-clerk" null. *Assessment:* the 1-in-10¹³ figure is almost certainly an over-statement of strength. 5. **Wassmer (IACR ePrint 2024/695) — and what it doesn't prove.** This recent (non-peer-reviewed) analysis claims to show "there is no key which will decipher them into sentences" via non-statistical correlations between cipher numbers and the 1885 pamphlet text. *On the surface this strengthens the hoax case.* But it explicitly assumes a single-layer book-substitution model; its claim is "no simple key" not "no key." It does not rule out Pelling's composite scheme. Reference: [eprint.iacr.org/2024/695](https://eprint.iacr.org/2024/695). --- ## Posterior recommendation **I would defend ~95%, not 99%.** The 99% claim is over-confident for two reasons: (a) Pelling's double-encipherment hypothesis is technically coherent, explicitly predicts the user's null results, and is not yet refuted by anything the user has run; (b) the transcription-artefact channel — Ward's printer made *demonstrable* errors in B2's key — has not been swept for B1. Knocking ~4 percentage points off for "genuine but tested under wrong model" feels right. Conversely, Hammer, Singh's hedging, and Wassmer's residual uncertainty are not strong enough to drop the verdict below ~90%. --- ## Open questions for future work 1. **Run a composite-key sweep**: per-position parity keys, Hill 2×2/3×3 over book-decoded letters, and columnar transpositions of book output, against B1 with the Declaration as outer key. 2. **Re-test Phase 3 against the *actual* (un-modified) Declaration** and against the Ward-pamphlet Declaration shifted by ±1, ±2, ±3 in the numbering. Does the Gillogly string move, scramble, or persist? 3. **Recompute Gillogly's p-value** under a non-uniform null that models a tired-clerk encoder (autocorrelated number choices, alphabetized wordlist as working memory). Does the 1-in-10¹³ figure survive? 4. **Pelling's specific claim** — that the homophone table refined from B2 plus a different sweep generates a coherent B1 cipher table — is testable. Reconstruct his proposed table and search for any short English fragment in B1 under it. 5. **NSA FOIA release on Beale**: the [NSA Beale Papers historical release](https://www.nsa.gov/Helpful-Links/NSA-FOIA/Declassification-Transparency-Initiatives/Historical-Releases/Beale-Papers/) contains Friedman/Kullback-era memos. Do any of them describe untested cipher schemes? --- **Sources (all consulted):** - [Gillogly 1980 — A Dissenting Opinion (Semantic Scholar)](https://www.semanticscholar.org/paper/The-Beale-Cipher:-a-Dissenting-Opinion-Gillogly/d30667d742b82bbbe5cb5c88cde3aaea0a0103ab) - [Pelling — Jim Gillogly's Beale sequence revisited (2010)](https://ciphermysteries.com/2010/06/22/jim-gilloglys-beale-sequence-revisited) - [Pelling — Gillogly Strings revisited (2013)](https://ciphermysteries.com/2013/11/21/beale-ciphers-gillogly-strings-revisited) - [Pelling — Refining B1's cipher table (2018)](https://ciphermysteries.com/2018/11/15/refining-beale-cipher-b1s-cipher-table) - [Pelling — Improbable World of B1 (2018)](https://ciphermysteries.com/2018/12/22/the-improbable-world-of-beale-cipher-b1) - [Campanelli 2022 — *Cryptologia* 47(5)](https://www.tandfonline.com/doi/abs/10.1080/01611194.2022.2116614) - [Wassmer 2024 — IACR ePrint 2024/695](https://eprint.iacr.org/2024/695) - [Fitzgerald — beale-ciphers GitHub (BF ≈ 2×10⁷ hoax)](https://github.com/david-fitzgerald/beale-ciphers) - [Wikipedia — Beale ciphers](https://en.wikipedia.org/wiki/Beale_ciphers) - [Cipher Foundation — Beale Papers transcription](https://cipherfoundation.org/older-ciphers/beale-papers/beale-papers-transcription/) - [Hart Papers (1952)](https://smd173.tripod.com/Beale/HartPapers.htm) - [NSA FOIA — Beale Papers](https://www.nsa.gov/Helpful-Links/NSA-FOIA/Declassification-Transparency-Initiatives/Historical-Releases/Beale-Papers/) - [Adelaide Cipher Cross-off List](https://www.eleceng.adelaide.edu.au/personal/dabbott/wiki/index.php/Cipher_Cross-off_List) - [Singh, *The Code Book* — Beale chapter, review](https://www.acampbell.org.uk/bookreviews/r/singh.html)