# The January properties table, and what its extra columns say Found 16 September 2026 while executing Michael's dispositions on the Act 1 data audit. It came out of finding n7 — "show me what #272's Arweave link points at". ## How it was reached Piece **#272** is the one piece whose on-chain description carries an extra closing paragraph with its own Arweave link. Every other piece that carries an archive link points at `LvwCpQp6ctt-wRCG5eU4KR_j5-jHDjpAj_ugwoPgc5o`. #272 points at `Tn-2Uclzan8-VqcitD9pfnYpRPRCjX8RoE4h3wKnCng`. Both are captures of the same page, `bitgans.com/journal/page015`. The difference is the date. The common link is the **31 January 2022** capture. #272's is the **1 January 2022** capture, taken at 22:12 UTC that day. #272 is also the only piece in the whole collection whose first transfer out of Pindar's wallet happened in January 2022 before the sheet — on 4 January. It left three days after the 1 January capture and so never received the 31 January link. The anomaly and the transfer record explain each other. ## What the January capture contains The 31 January page is a short index: eight Arweave links and nothing else, 1,526 characters of text. The 1 January page is 29,478 characters, because Pindar pasted an entire properties table into it as inline CSV — 300 rows, pieces 1 to 300. Its header is not the header of the published sheet: index,,Number,Title,Decimal,Hidden Type,Name,Strokes,GANg,SubType,Type,Eyes, Dimensions,Essence,Drop,,Level,bip-0039 Four columns exist here and nowhere else: **Decimal**, **Hidden Type**, **Strokes**, **bip-0039**. On every column the two versions share — Title, Name, GANg, SubType, Type, Eyes, Dimensions, Essence, Level — the January table and the published sheet agree on all 300 pieces, with no exceptions. The January table is not an earlier draft that was later corrected. It is the same data with four columns still attached. Both files are kept here: `page015_capture_2022-01-01.html` and the extracted `page015_jan1_table.csv`. The columns are also machine-readable in `bip39_index.json`. ## Column 1 — Hidden Type, and the two questions from journal 002 The column publishes the hidden-type letter in plaintext, for pieces 1 to 208, four months before anyone had to read it off the chain. Against the chain trait it disagrees on exactly two pieces, and they are exactly the two Pindar flagged in journal page 002 on 31 August 2021 as "typo or important?": | piece | chain | Pindar's own sheet | the class the number actually is | |---|---|---|---| | #32 | `32` | `e` | even (remaining) — `e` | | #119 | `e` | `d` | odd (remaining) — `d` | His private sheet holds the **correct** letter for both. So both are chain errors, not codes. That answers the question he asked in journal 002 and left open. The other three hidden-type misfiles are **identical in both sources**: #161 also says the literal `32`, #187 also says `C`, #193 also says `F`. Those three he carried forward deliberately; the two above he had already fixed in his own data. This splits the five misfiles 2 / 3 on evidence rather than on taste, which is what Michael asked for on claim 18. ## Column 2 — Decimal The piece's position inside its own drop: 1-based for skulls (1..128), 0-based for everything after (robo 0..63, spooky 0..15, solstice 0..91). It confirms the drop-relative numbering the site already uses. ## Column 3 — Strokes Zero for all 208 skull, robo and spooky rows. For the 92 solstice rows it holds a date — `Day`, `January 4`, `May 1`, … `November 30`, `Night`. It is the solstice `Date` trait under a column heading left over from another sheet. ## Column 4 — bip-0039, and what it settles An explicit 1-based index into the BIP-39 wordlist, one per piece. Three rows disagree with the index of their own Essence, and all three are already-known anomalies: - **#191** essence `err0r`, index **615**. Word 615 is `error`. His own column decodes the leetspeak. - **#259** essence `bless`, index **181**. Word 181 is `bird`. - **#268** essence `bird`, index **190**. Word 190 is `bless`. The index column runs 180, 181, 182 … 189, 190, 192 — strictly ascending through that stretch. The **words** at #259 and #268 are transposed against it. The two essences were swapped, and the index column preserves the un-swapped order. This matters for the `essence_out_of_sequence` tag, which currently files four pieces together. Two of them are a different thing from the other two: - **#21 `bean`** and **#57 `mistake`** carry indices that match their own words (157 and 1136). They are deliberately outside the run. - **#259** and **#268** are a transposition. Nothing is outside the run; two labels changed places. ## The shape of the whole allocation Taking Pindar's column for 1–300 and deriving the rest from the published essences, 510 of the 512 pieces map to a BIP-39 index. (#363 `TRUE` and #473 `FALSE` are not wordlist words.) **No index is used by two pieces.** The essences are an injective map into the wordlist. Indices **1 to 224 are one solid block**, allocated in piece order: - **skulls take 1–128**, index equal to piece number, for 126 of the 128 - **solstice takes 129–224**, in piece order, for 88 of the 92 with exactly six displacements and one hole: | index | word | belongs to the run at | actually held by | |---|---|---|---| | 21 | `action` | skull #21 | **nobody** | | 57 | `alpha` | skull #57 | robo #129, RoboboR | | 135 | `awkward` | solstice | robo #141, Dr Awkward | | 157 | `bean` | solstice | skull #21 | | 191 | `blind` | solstice | robo #131 | | 195 | `blue` | solstice | robo #189 | Skull #21 and #57 gave up their slots; #21 took `bean` from the solstice run and #57 took `mistake` (1136) from far outside it. Two of the four robots that reach into the low block are palindrome nagas, and **Dr Awkward holds the word `awkward`**. `action` at index 21 is the only index in 1–224 that no piece holds. Above 224 the pattern changes rather than continuing: - **spookies** hold a clean consecutive block, **800–815**, `glue` through `grape` - **ghosts** begin at 225 but only 8 of 144 stay in the low block; the other 135 scatter - **review** likewise: 7 of 67 low, 60 scattered - **robos** scatter from 57 to 2046 So Michael's claim-16 question has an answer. The drops do not all continue one logic. Skull and solstice are one continuous run of the wordlist. Spooky is its own block. Robo, ghost and review are not runs at all. ## What is NOT claimed here No sweep was run and none should be until a selection rule, a word-derivation rule and an ordering rule are written down and agreed (standing rule D60). What this file records is structure, not a solve. The obvious questions it raises — why `action` is empty, why the six displacements land where they do, whether the displaced set is a selection — are hypotheses to state and argue, not to enumerate.