Per-claim calibrated confidence (0–1) with explicit open questions
| Claim | Conf. | Basis |
|---|---|---|
| Generator reproduces Reeds's full transcriptions: NISRAM 216/216 · Leo 466/468 (2 diffs = Reeds's catalogued original-errors) · Moon margin+col 24/24 |
0.99 | Cell-for-cell computational check against Reeds's typeset tables. Independent second-reviewer reimplementation confirms. |
| f(W) = V(W) − 1 (mod 23) for all 23 letters as printed in Kupin | 0.97 | Pure arithmetic on primary text. Uniqueness: offset k=1 gives 23/23; every other k gives 0/23. Confirmed by blinded replication and two independent reviews. Single remaining dependency: paleographic check of Section 18 against the manuscript. |
| Two attractor cycles (d→i→z→b and o→y) explain the repetition patterns and letter-frequency skew | 0.97 | The map T(x) = x + f(x) mod 23 has exactly two cycles. O-basin tables (T5, T6, T16, T23, T25, T27, T34) show the yoyo-flood Reeds noted empirically. Provable in five lines; re-verified by second review. |
| Claim | Conf. | Caveat |
|---|---|---|
| o, t = 9 (attested in the printed Kupin edition) | 0.93 | The line "O plenum cum T, comis est, S ter tria portat" is fully legible in the printed edition (second review, 2026-06-09, ~line 8165). IA OCR concurs. Residual uncertainty: manuscript check of this line. |
| 23-letter alphabet (Section 27) + Zadzaczadlin = Adam in the primary source (Section 19) | 0.88 | Both are stated in the book's own text (Kupin lines 22341–22358 and 8682–8686/9061–9065) and confirmed by Whitby. Caveat: the zad/zac/lin segmentation is the book's own reading, not an external analysis. |
| Claim | Conf. | Caveat |
|---|---|---|
| f = V − 1 as the historical origin of Reeds's f (not just arithmetic) | 0.92 (distinct from arithmetic 0.97; raised by blinded replication + book-wide sums) | The arithmetic is 0.97. That this verse is why Reeds's empirical f has those specific values — rather than the arrow pointing the other direction — is an inference, not a proof. The blinded replication (reviewer found the verse independently) and the book-wide arithmetic sums support this strongly. Reverse causality cannot be ruled out from the arithmetic alone. |
| Operative / scrying-mirror reading: the tables are an angel-binding grid and scrying mirror (vicem speculi gerit, §26.1), not a steganographic cipher | 0.85 | Verbatim quotes from §§26.1, 27, 28. The operative reading complements the cryptographic null; both rest on the book's own text. |
| Claim | Conf. | Caveat |
|---|---|---|
| Code words are drawn from the letters of the adjacent Liber Radiorum divine-name triads (23/23 spellable; shuffled-assignment null mean 7.3/23, max 15/23 in 20,000 draws; p ≪ 0.001) | 0.80 | The statistical result is solid under any null tested (uniform or letter-frequency-matched; quote the joint statistic). The link between code words and triads is real. The exact selection rule (see below) is a separate, weaker claim. |
| Soyga → Loagaeth (36 → 49) — material contact proven; direction of influence undecidable | 0.50 | Dee physically bound 8 Soyga tables into Loagaeth (~0.95 for material contact). Whether 36 tables consciously seeded the 49-table structure: Reeds: "I see no way to distinguish." Do not claim more than Reeds did. |
| Exact code-word selection rule (Section 28 ordinals → code-word letters) | 0.45 | 13/23 clean (main ordinals + first-listed spelling + 1-based indexing); 17/23 allowing any printed spelling; 19/23 adding Bodley curly-brace ordinals. Cancer, Pisces, Mercury, Moon underivable from any printed reading without emendation. The book itself notes "ista valde intorta sunt" ("these are very twisted"). Moon row: printed recipe gives UISEUR; unique one-ordinal repair (final xvi→vi) gives UISEUA, awaiting manuscript image. |
| "Soyga alca miketh" = Michael's domain (speculative framing) | 0.35 | The diary's own gloss is judgment / wisdom (Whitby; Laycock and Rowe independently index alca = judgment, miketh = wisdom). Michael is Soyga's appointed interpreter, which is contextual, not lexical. miketh ≠ michael numerically (63 vs 49 in the 23-letter alphabet). Present both the judgment/wisdom gloss and the Michael-domain reading as open. |
State this out loud — it is the armor against overclaim:
We did not decode Soyga alca miketh. The diary’s judgment/wisdom gloss is the safest local reading, but Whitby leaves translation-versus-comment unresolved, and no second primary attestation has surfaced.
We did not fully recover the code-word construction recipe. The letter-pool link (p ≪ 0.001) is solid; the clean selection rule is 13–19/23 depending on lane. Cancer, Pisces, Mercury, and the Moon row are open.
We cannot prove the 36→49 Loagaeth inspiration was conscious or directional.
The Moon row (Tabula 31, UISEUA) awaits manuscript image verification. The unique one-ordinal repair is internally consistent but unconfirmed by the physical document.
The headline rests on one external dependency we welcome being tested: the Latin of Section 18, against the manuscript, by a paleographer.
The book’s “facies” catalogue is the 28 lunar mansions (per Kupin line 12439), not 36 decans. The book never states a 36-table count. Its living count of 36 is the §19 thirty-six-star body. Describing the 36 tables as “the 36 decans” is a misread of the structure.
The arithmetic identity (f = V − 1 for all 23 letters as printed) carries 0.97. The historical-origin claim (that this verse is why Reeds’s f has those values) carries a separate 0.92 — interpretive, but raised from 0.85 by the blinded replication and the book-wide sums. A room of historians will accept 0.97 arithmetic and probe the 0.92 historical origin as an independent matter. Keep them distinct.