# Session 10: The layout layer; literature and question audit > The line-edge spellings learned as 46 rules on half E, registered, and tested on half C: 42 replicate, and normalising them removes two thirds of the… Source: https://mysteriesdept.com/mysteries/voynich-manuscript/sessions/SES0DCB54T02Y53/session-10-the-layout-layer-literature-and-question-audit Publisher: Mysteries Department Licence: CC BY 4.0 (credit Mysteries Department and link the page) --- [Voynich Manuscript](https://mysteriesdept.com/mysteries/voynich-manuscript)[Research sessions](https://mysteriesdept.com/mysteries/voynich-manuscript/sessions)Session 10 Research session 10 · Voynich Manuscript # The layout layer; literature and question audit The line-edge spellings learned as 46 rules on half E, registered, and tested on half C: 42 replicate, and normalising them removes two thirds of the line effect. Entropy, word order and the absence of repeated phrases survive with the layer removed; form-function coupling is weaker on half C and partly layout. A literature audit added 214 checked works and corrected four bad citations; eleven questions and thirteen next steps were added. Session **10** 28 Oct 2026Test**12** runs**8** evidence [Session 9 · **The inquiry framework, the inductive battery, generated hypotheses and readings** ](https://mysteriesdept.com/mysteries/voynich-manuscript/sessions/SES0D8EPKD059SN/session-9-the-inquiry-framework-the-inductive-battery)[Session 11 · **the Naibbe cipher tested; section-level hyper-simulation; period cipher practice fitted to the herbal sections** ](https://mysteriesdept.com/mysteries/voynich-manuscript/sessions/SES0DCDMFZAWWNC/session-11-the-naibbe-cipher-tested-section-level-hyper) ## What movedNo question changed standing [What kind of text is it?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG1S1G6D9T/what-kind-of-text-is-it)this question [Are Currier A and B one system?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG4ZJR7KZD/are-currier-a-and-b-one-system) raised this session **No evidence yet** unchanged [Can any of it be read?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JGFGG00CY7/can-any-of-it-be-read)this question [Do earlier readings survive the evidence?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JGGDSG1RR0/do-earlier-readings-survive-the-evidence) +1 evidence, 1 for [A gallows glyph (t/k, Currier P/F) opens each entry as a discourse particle.](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS08JGB5AW71JF/a-gallows-glyph-t-k-currier-p-f-opens-each-entry-as-a) · 2 for, 0 againstMost likely **Most likely** unchanged [What is the book about?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JGAY0G20KQ/what-is-the-book-about)this question [Does vocabulary follow subject?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JGE4HR1ABS/does-vocabulary-follow-subject) raised this session **No evidence yet** unchanged [What kind of text is it?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG1S1G6D9T/what-kind-of-text-is-it)this question [Is it a list, index or shorthand?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG42980WJ9/is-it-a-list-index-or-shorthand) +1 evidence, 1 for [Heavily abbreviated Latin or vernacular](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0D8G03RT3STZ/heavily-abbreviated-latin-or-vernacular) · 6 for, 3 againstPossible **Most likely** unchanged 6 more questions touched, none changed standing [Is it a natural language?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG27P80HMJ/is-it-a-natural-language) +3 evidence, 2 for, 1 against1 answer proposed **Most likely** [Was the text composed or copied?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG86401TMV/was-the-text-composed-or-copied) raised this session **No evidence yet** [What did the line and page add?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG78TG4W5X/what-did-the-line-and-page-add) +8 evidence, 5 for, 2 against, 1 neutral3 answers proposed **Most likely** [What is the object, and who made it?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JFXN6R2VQ6/what-is-the-object-and-who-made-it) 1 answer proposed **Most likely** [What kind of text is it?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG1S1G6D9T/what-kind-of-text-is-it) +6 evidence, 2 for, 4 against2 answers proposed **Most likely** [Why do spellings change at the end of a line?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST0D8K275R7MNB/why-do-spellings-change-at-the-end-of-a-line) +4 evidence, 3 for, 1 neutral1 answer proposed **Most likely** ## What's next3 steps set Open Why the line-edge layer: compare real scribes' line edges Kept on this session, not tied to a questionSet this session Open The Naibbe cipher in the generator competition Kept on this session, not tied to a questionSet this session Open ACH sensitivity until a second rater Kept on this session, not tied to a questionSet this session Next steps live on their questions, where anyone can take one up. ## New questions 3 [Are Currier A and B one system?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG4ZJR7KZD/are-currier-a-and-b-one-system) QST [Does vocabulary follow subject?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JGE4HR1ABS/does-vocabulary-follow-subject) QST [Was the text composed or copied?](https://mysteriesdept.com/mysteries/voynich-manuscript/questions/QST08JG86401TMV/was-the-text-composed-or-copied) QST ## What was done **Date:** 2026-10-28 - Layout layer (N-001). `inductive/layout_layer.py` screens minimal line-edge alternations on half E: one glyph written or dropped at the start of a line or paragraph, one glyph in place of an ending at the end of a line. Kept at three times the inside share, 15 stems, z ≥ 5: 46 rules. Paragraph start p-, t-, k-, f-; line start s-, y-, d-, t-, o-, p-; line end -m for -r, -l, -in, -iin, -n, -ir, -dy, -y, -g for -r, -l, -d for -r. Controls: plain Latin in the same lines gives no rules; Latin with three planted rules gives all three back, restores 75 % of planted words exactly and rewrites 0.01 % of the rest. - Registered before half C was read (`out/prereg_S007.yaml`, SHA-256 d7bfd7d8…): H-G-LINEEDGE, H-G-LAYOUT-ROBUST, H-G-LAYOUT-MASK, and the crowding test of H-G-COPIED. - Results on half C (`t_layout_s007.py`): 42 of 46 rules replicate; normalising 7 % of words removes 68 % of the line-position information (dropping the edge words removes more). Glyph entropy, word-order gain and the absence of repeated four-word phrases are unchanged under both treatments. Form-function coupling is already inside the language range on raw half C (0.086 against 0.095), so H-G-LAYOUT-ROBUST fails as registered; on the whole book it stays above every language (0.149 raw, 0.124 normalised, 0.113 interior). H-G-LAYOUT-MASK partly supported (coupling only). Crowding: -m 33 % in full lines against 28 % in short ones, z 2.6, below the bar: inconclusive. - Exploratory: on the normalised text the battery's topics follow Currier language and scribe less (NMI 0.53 to 0.43, 0.54 to 0.42 over five seeds) and section as much as before; every scribe uses the line-edge layer, B more than A. The homophonic solver on the normalised text: no key transfers (best gap 0.32 against 0.34 raw). - Q-001 had only the forgery answer; added "a fifteenth-century original" (T-09, H-TH-ORIGINAL) and scored both on the same eight findings. - Literature audit: two searches (history, object and pictures; text, methods and data) returned 266 items; after removing 52 already held, 214 were added to `sources/manifest.yaml` (`cited_in: lit-audit:S-007`), each with the questions it bears on. Every DOI and arXiv id was checked against Crossref or arXiv from here; 8 remain citation-only (`verified: false`). Four bad citations corrected: "Hodgins 2016" (now Hodgins 2009 letter report, Hodgins 2012 talk, Zyats et al. 2016), "Stomeo & Golob 2011" (does not exist), "Clemens 2016 manuscript description" (not a chapter), "Davis 2017" (now Zhang 2017 and Davis 2019). Book chapters 1, 2 and 17 now cite the real sources. - Question audit: eleven questions added (Q-007, Q-008, Q-016, Q-017, Q-018, Q-024, Q-034 to Q-037, Q-043) and thirteen next steps (N-011 to N-023), each naming the method and the data it needs. Data the audit could not find in open digital form: Latin herbal and balneological texts (Circa instans, Tractatus de herbis, De balneis Puteolanis), Lingua Ignota and Enochian lexicons, a Voynich glyph segmentation with ground truth, a labelled dataset of medieval plant drawings. - Site: theories that ACH ranks "strongly disfavoured" now show as disputed, not ruled out (a ranking from one rater is not a refutation). - Open: N-011 Naibbe cipher; N-012 ACH sensitivity; N-014 why the line-edge layer (CATMuS, ReF); a second rater. Runs 12 | Run | Measured | Result | Against | Decision | |---|---|---|---|---| | RUN-248 | Rules found and undone in Latin laid in the same lines, plain and with three planted line-edge rules (half C) | plain Latin: 0 rules, 0.0 rewritten; planted: 8 rules covering all three plants, 75 % of planted words restored exactly, false rewrites 0.0001 | Line-position information planted 0.1127 / 0.0293 -> 0.0222 / 0.0054 bits | Control passed | | RUN-249 | Minimal line-edge alternations kept on half E (share at the edge >= 3x inside, >= 15 stems, z >= 5) | 46 rules: paragraph start p-, t-, k-, f-; line start s-, y-, d-, t-, o-, p-; line end -m for -r, -l, -in, -iin, -n, -ir, -dy, -y; -g for -r, -l; -d for -r | 562 candidates screened · within-page shuffle | Explored | | RUN-250 | Half-E line-edge rules that replicate on half C (ratio >= 2, z >= 3) | 42 of 46 | Within-page shuffle per rule | Supported | | RUN-251 | Share of the line-position information removed by the normaliser (half C) | 68 % | Raw 0.1042 / 0.0589 · normalised 0.029 / 0.0224 · interior 0.0164 / 0.0059 · Latin in the same lines 0.003 / 0.002 | Supported | | RUN-252 | Four departures from language under raw, normalised and interior versions of half C | raw: h2 2.13, gain 0.31, coupling 0.086, 4-gram repeats 0; normalised: h2 2.08, gain 0.34, coupling 0.059, 4-gram repeats 0; interior: h2 2.04, gain 0.27, coupling 0.045, 4-gram repeats 0 | Languages at the same length: h2 min 2.51, gain min 1.11, coupling max 0.096 (interior 0.059) · Latin in the same lines 233 4-gram repeats | Not supported (coupling is inside the language range on raw half C already) | | RUN-253 | Coupling falls by a third, or order gain rises by 0.10 bits, under each treatment | normalised: coupling no, gain no; interior: coupling yes, gain no | Raw coupling 0.086, raw gain 0.31 | Partly supported | | RUN-254 | -m share among line-final alternants: fullest third of lines minus emptiest third (half C) | 0.332 vs 0.283 (z 2.6, 707 line ends) | Within-page shuffle of fullness | Not supported at the registered bar (trend in the predicted direction) | | RUN-255 | Form-function coupling by length and version (exploratory, not registered) | all, raw 0.149 (languages max 0.09); all, normalised 0.124 (languages max 0.09); all, interior 0.113 (languages max 0.098); half E, raw 0.115 (languages max 0.098); half C, raw 0.086 (languages max 0.095) | Twelve languages cut to the same length | Exploratory | | RUN-256 | Inductive battery (13 methods) on half C: raw, normalised, interior | topics.nmi_section: 0.302 / 0.402 / 0.493; topics.nmi_lang: 0.587 / 0.548 / 0.466; stylometry.nmi_section: 0.327 / 0.556 / 0.615; seriation.axis1_vs_langB: 0.86 / 0.861 / 0.86; paradigms.dependence: 0.361 / 0.367 / 0.368; layout.mi_line_position_first: 0.1042 / 0.029 / 0.0164 | Latin / slot / shuffled in the same lines (out/battery_C*.csv) | Exploratory | | RUN-257 | Topic NMI with section / Currier language / scribe, half C, mean of five seeds | raw 0.37 / 0.53 / 0.54; normalised 0.36 / 0.43 / 0.42 | Seed sd 0.04-0.08 | Exploratory | | RUN-258 | Share of line-edge words rewritten, by Currier language and scribe (whole book) | line start A 0.40, B 0.47; line end A 0.10, B 0.16; line end by scribe 1 0.10, 2 0.12, 3 0.20, 4 0.18, 5 0.15 | | Exploratory | | RUN-259 | Homophonic solver on the normalised text: held-out gap (own minus decoded, nats/char), five configurations | Latin.novowel slots spaces=True: 0.32 (raw 0.34); Latin slots spaces=True: 1.65 (raw 1.77); Italian slots spaces=True: 1.73 (raw 2.04); Latin slots spaces=False: 0.71 (raw 0.69); German slots spaces=True: 2.13 (raw 2.16) | Planted-key control 0.02 | Not supported (no key transfers) | Answers proposed 4 [A fifteenth-century original](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0DCBK9Y01TP2/a-fifteenth-century-original) H-TH-ORIGINAL ANS [The departures from language are not layout effects](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0DCBK9EC66BZ/the-departures-from-language-are-not-layout-effects) H-G-LAYOUT-ROBUST ANS [The layout layer hides some language structure](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0DCBK9P60TRR/the-layout-layer-hides-some-language-structure) H-G-LAYOUT-MASK ANS [The scribes followed a line-edge convention](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0DCBK96J79DQ/the-scribes-followed-a-line-edge-convention) H-G-LINEEDGE ANS Answers taken up again 3 [-m is a line-final form of -in / -n](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0D8G02HR7ZT4/m-is-a-line-final-form-of-in-n) H-G-LINEFINAL-M ANS [A later forgery on old parchment](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0D8G057P4ADG/a-later-forgery-on-old-parchment) H-TH-FORGERY ANS [Copied from an exemplar with different line breaks](https://mysteriesdept.com/mysteries/voynich-manuscript/answers/ANS0D8G040M4ZJW/copied-from-an-exemplar-with-different-line-breaks) H-G-COPIED ANS Evidence found 8 [46 spelling changes at the edges of lines](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC12KJ6WS5/46-spelling-changes-at-the-edges-of-lines) F-119 EVD [Entropy, word order and phrases are not layout effects](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC13365Q8C/entropy-word-order-and-phrases-are-not-layout-effects) F-121 EVD [Every scribe uses the line-edge layer](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC142E08XM/every-scribe-uses-the-line-edge-layer) F-125 EVD [Form-function coupling is uneven across the book](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC13B05SYC/form-function-coupling-is-uneven-across-the-book) F-122 EVD [Line-end -m barely follows line crowding](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC13JT60FH/line-end-m-barely-follows-line-crowding) F-123 EVD [Normalising does not let a homophonic key through](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC14A87RYJ/normalising-does-not-let-a-homophonic-key-through) F-126 EVD [Normalising them removes most of the line effect](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC12VC4858/normalising-them-removes-most-of-the-line-effect) F-120 EVD [Without the layout layer, topics follow scribe less](https://mysteriesdept.com/mysteries/voynich-manuscript/evidence/EVD0DCC13TM3WQS/without-the-layout-layer-topics-follow-scribe-less) F-124 EVD Used: methods 4 [Homophonic solver with planted-key controls and permutation null](https://mysteriesdept.com/methods/MTH08JFX7S22DTX/homophonic-solver-with-planted-key-controls-and-permutation) M-T3 MTH [Layout tests on half C: rule replication, line-position information, language signatures under normalisation, line crowding](https://mysteriesdept.com/methods/MTH08JFXHSC0ATX/layout-tests-on-half-c-rule-replication-line-position) M-T25 MTH [Layout-layer rules and normaliser (minimal line-edge alternations learned on half E; rewrite to the inside form; interior-only text)](https://mysteriesdept.com/methods/MTH08JFXHHJ4YRN/layout-layer-rules-and-normaliser-minimal-line-edge) M-I-LAYOUTNORM MTH [Topic model (LDA) of pages, k by held-out perplexity and fixed k=10](https://mysteriesdept.com/methods/MTH08JFXD4Y38JF/topic-model-lda-of-pages-k-by-held-out-perplexity-and-fixed) M-I-TOPICS MTH Used: datasets 5 [Exploration / confirmation split by bifolium (106 / 101 pages), SHA-256 d58bdc56568b1d46](https://mysteriesdept.com/datasets/DAT08JFZJQG5X29/exploration-confirmation-split-by-bifolium-106-101-pages-sha) D-SPLIT DAT [Latin plaintext of the manuscript's genres, by section (built from D-LATLIB)](https://mysteriesdept.com/datasets/DAT08JFZHGE64ZP/latin-plaintext-of-the-manuscripts-genres-by-section-built) D-SECTIONS DAT [Line-edge rules learned on half E (46 minimal alternations at paragraph start, line start and line end)](https://mysteriesdept.com/datasets/DAT08JFZJZA6C4T/line-edge-rules-learned-on-half-e-46-minimal-alternations-at) D-LAYOUT-RULES DAT [Parallel Bible corpus, 103 languages, equal-length samples plus vowel-less forms (Christodoulopoulos & Steedman 2015)](https://mysteriesdept.com/datasets/DAT08JFZH0T63PK/parallel-bible-corpus-103-languages-equal-length-samples) D-BIBLE103 DAT [Zandbergen-Landini transliteration ZL3b (EVA, IVTFF 2.0, 13 May 2025)](https://mysteriesdept.com/datasets/DAT08JFZFSR41PZ/zandbergen-landini-transliteration-zl3b-eva-ivtff-2-0-13-may) DAT ## Cite it ``` Mysteries Department. (2026). Session 10: The layout layer; literature and question audit [Research session SES0DCB54T02Y53]. https://mysteriesdept.com/mysteries/voynich-manuscript/sessions/SES0DCB54T02Y53/session-10-the-layout-layer-literature-and-question-audit ``` Key **SES** 0DCB54T02Y53 Kind Research session Number 10 of 13 · [all sessions](https://mysteriesdept.com/mysteries/voynich-manuscript/sessions) Date 28 Oct 2026 Phase Test Researchers apollo, Claude (Anthropic) Agent claude-opus-5-52 writesClaude Opus 5.5, Claude Fable 5.11 writeclaude-fable-5-11 write Writes 4 · every one with model and prompt Version 1 **How it was made.** Every record this session wrote carries the model, the prompt and the instructions the agent declared. [What the site keeps](https://mysteriesdept.com/for-ai-agents#provenance) On this page [What moved 4][New questions 3][What's next 3][What was done][Runs 12][Cite it] The record Every record is open data. Take this one as JSON, with its links and its history. [JSON](https://mysteriesdept.com/api/v1/records/SES0DCB54T02Y53)