# Bibliography for the sheet finding, with read marks

Compiled 2026-09-17 for the write-up, from the literature check in sheet_finding_prior_work.md (Parts 1 to 5). Entries are in the order a reader should meet them.

Marks. READ: a person on this search read the work itself, in full, at the link given (entries 1, 3, 4, 6, 8 were read or re-read on 17 September 2026 in the passes recorded in the report; the rest on the first pass the same evening). DESCRIBED: we have only an abstract, an index entry or another author's account. NOT READ: we tried to reach it and could not; the reason follows. OURS: a figure or inference that is our own arithmetic or reading, not printed by the work; where a "Claims" line carries one, it is marked in place. Everything not marked OURS is quoted or read from the work.

## 1. The paper

Colin Layfield and Lisa Fagin Davis, "Singulion Structure and the Voynich Manuscript", Digital Medievalist 19 (2026), online 10 July 2026, DOI 10.4000/16k0a, https://journals.openedition.org/digitalmedievalist/2331
Mark: READ (full text, twice; the second time end to end, every paragraph, note, table and appendix).
Claims: Latent Semantic Analysis cosine over whole pages, ZL 2a transliteration, stop words and hapax removed. Overall (Table 10) confoliate 0.383, facing 0.298, conjoint 0.433; t-tests (Table 11) confoliate/facing p 0.005, confoliate/conjoint p 0.168, facing/conjoint p 0.000; per quire in Table 8 (quire 20: 0.568, 0.438, 0.600). Inference (paragraph 56): the sheets "are meant to be read as single bifolia of four consecutive pages". Reordering of quire 20 (Table 14) to 105|114, 104|115, 106|113, 107|112, 108|111, 103|116, score 0.526 against 0.438 as bound. Their only reading of Table 8 (paragraph 39): facing "tends to be the lowest of the three".
OURS: every difference between their three columns; the sheet-pair counts (38 conjoint, 76 confoliate, about 67 facing); the reproduction of 0.433 as the pair-weighted mean of Table 8; the page-level writing-out of their quire 20 order. They print none of these.

Companion: Colin Layfield and Lisa Fagin Davis, "The Application of Latent Semantic Analysis to the Voynich Manuscript", Digital Humanities Quarterly 20.1 (2026), DOI 10.63744/2ezxpskcezq4, https://dhq.digitalhumanities.org/vol/20/1/000857/000857.html
Mark: READ (at DHQ; the ResearchGate copy returned 403).
Claims: coherence of adjacent text documents within sections against shuffled baselines; quires 13 and 20 the most coherent sections, which is why the singulion paper reorders those two; no bifolium or page-pair result.

## 2. The review

Author not identified, "The Singulion and the Sequence: A Methodological Review of Layfield and Davis 2026 on the Voynich Manuscript", academia.edu, 2026; listed under https://www.academia.edu/Documents/in/Codicology_of_medieval_manuscripts
Mark: NOT READ. academia.edu returned 403 to every fetch and the document's own page was not located. DESCRIBED from its indexed abstract only: "the empirical core of the first hypothesis, an elevated textual similarity within physical bifolia, is real and independently confirmed. The reading-order interpretation built upon it, however, is underdetermined by the data and contradicted by the authors' own explanation of the herbal section."

## 3. Currier

Prescott H. Currier, "Some Important New Statistical Findings", in M. E. D'Imperio (ed.), New Research on the Voynich Manuscript: Proceedings of a Seminar, 30 November 1976, Washington D.C., privately circulated typescript; pp. 20-26 as cited by Layfield and Davis (the typescript itself not seen).
Mark: READ in full in the transcription by Jacques Guy and Jim Reeds (January 1992) on René Zandbergen's site, "Papers on the Voynich Manuscript", https://www.voynich.nu/extra/curr_main.html (HTML dated 01/08/2004, 05/06/2015, 29/05/2025), including the seminar discussion and Appendix A. The NSA-hosted PDF of the proceedings, https://media.defense.gov/2021/Jul/13/2002761428/-1/-1/0/PROCEEDINGS-OF-A-SEMINAR-30-NOVEMBER-1976.PDF, returned 403 (Access Denied): NOT READ in that form.
Claims: "The first twenty-five folios in the herbal section are obviously in one hand and one language, which I called A ... the text of this second portion of the herbal section ... is in two languages, and each language is in its own hand." In the discussion: "The writing on folios 1 to 25 was done by one man. On folios 25 to 65, it was done by two men". Currier speaks of folios and pages throughout; the words bifolio, bifolium and sheet do not occur in the transcription (OURS: a count over the transcribed text, confirmed on a full read). His own production scenario is leaf by leaf: "some sixty-five folios were prepared ahead of time with drawings on them. They were placed on a table so. The first twenty-five folios were taken, one at a time, off the top and filled in with writing by one individual." The sheet-level reading of his finding is Zandbergen's and Pelling's (entries 4 and 6), not his.

## 4. Zandbergen

René Zandbergen, "Currier A and B: two different languages?", voynich.nu, written 11 January 1997, last updated 29 May 2025 (dates as printed on the page), http://www.voynich.nu/extra/lang.html
Mark: READ.
Claims: word-token overlap between pages, normalised by the square root of the product of the page lengths, on the FSG transliteration. On the recipe quire: "The pages that also appear to be in Bio-B 'dialect' are those of ff. 103, 107, 108, 111, 112 and 116, which are three bifolia. The other three bifolia with ff. 104, 105, 106, 113, 114 and 115 use a more varied vocabulary". The earliest note that vocabulary in quire 20 clusters by sheet.

René Zandbergen, "Origin of the MS", voynich.nu, latest update 24 February 2026 (as printed), https://www.voynich.nu/origin.html
Mark: READ in full.
Claims: "There is every reason to believe that the Voynich MS was basically produced bifolio per bifolio (or foldout folio), as the properties discovered by Currier (different hands and different text statistics throughout the MS) are generally consistent for each bifolio." The sentence in which Currier's folio-level finding becomes a sheet-level claim. Its limit, on the same page: "It seems impossible to say whether all four drawings on a bifolio were done before all four text sections, or it was done page by page." He claims consistency of hand and statistics per sheet; he does not claim the order in which a sheet's four pages were written.

## 5. Bunn, with Pelling's comment

Julian Bunn, posting as "JB", "Folio Similarities", Computational Attacks on the Voynich Manuscript, 26 February 2010, https://voynichattacks.wordpress.com/2010/02/26/folio-similarities/
Mark: READ, with the comments. (The first name appears in the comments; the surname is the blog's indexed authorship.)
Claims: Jaccard similarity of each folio pair's set of unique words; the clusters reproduce Currier A and B. Control: "the Latin Herb Garden" split into 20 folios by herb entry, similarity about 0.02; a real text in its own division, not the manuscript's page layout. Nick Pelling's comment, printed on the post: "f104v is extremely similar to f115r, while f104r is extremely similar to f115v ... the closeness between f111v and f108r suggests that these originally formed the central bifolio (but reversed) ... However, f105 / f114 seem quite unconnected, as do f106 / f113 and f107 / f112." Qualitative, whole vocabulary, no null.

## 6. Pelling's two posts

Nick Pelling, "Evidence of bifolio reordering in the Voynich Manuscript...", Cipher Mysteries, 30 May 2013, https://ciphermysteries.com/2013/05/30/evidence-of-bifolio-reordering-in-the-voynich-manuscript
Mark: READ.
Claims (item 11 of fifteen): "any given bifolio has only 'A' or 'B' writing on it: this broadly supports the idea that these correspond to two broadly separate writing phases, rather than two separate writers writing in parallel." The sheet as the unit of Currier's dialect, stated as a codicological argument.

Nick Pelling, "Voynich Manuscript: Q20 (in-)homogeneity and BAAFU...", Cipher Mysteries, 30 December 2022, https://ciphermysteries.com/2022/12/30/voynich-manuscript-q20-in-homogeneity-and-baafu
Mark: READ.
Claims: glyph-pair densities ("ed", "iin", "ho") per page of quire 20, read off voynichese.com: "pages within a folio (and indeed pages within a bifolio, though to a slightly lesser extent) are actually reasonably consistent with each other", "true for most Q20 bifolios (e.g. f103 and the top half of f116r, f104-f115, f105-f114, f106-f113) but certainly not for others (e.g. f107-f112 and f108-f111)". Coins BAAFU, "Bifolio As A Functional Unit". Visual, glyph-level, no statistic, no control.
Its two preceding posts in the same series, both READ: "Quire 20, order from chaos... (Part 1: Evidence)", 20 December 2022, https://ciphermysteries.com/2022/12/20/quire-20-order-from-chaos-part-1-evidence, collects the physical evidence for the quire's order (Dulov's wormholes and pre-binding mark; vellum colour grouping 105|114, 104|115, 107|112, 108|111 against 103|116 and 106|113; paint transfers from f116r to f115v and from f113v to f115r; Davis's Scribe 3 throughout, with the first twelve lines of f115r by Scribe 2) and then reprints his 2010 comment to Bunn, with a leaf-by-leaf rating of how alike each Q20 leaf's two sides are on Bunn's metric ("103 good, 104 very bad, 105 very good, 106 bad, 107 excellent, 108 excellent, 111 excellent, 112 good, 113 excellent, 114 excellent, 115 very bad"), against which the two halves of sheet 104|115 are alike across the sheet although each leaf's own sides are "very bad". It closes: "I'm unaware if there are any studies that have specifically used page-to-page word list difference metrics (of the kind Julian Bunn did) to exhaustively evaluate all the different nesting permutations, i.e. to suggest what the original nesting order was" (December 2022; Layfield and Davis did this in 2026). "(Part 2: Suggestions)", 22 December 2022, https://ciphermysteries.com/2022/12/22/quire-20-order-from-chaos-part-2-suggestions, proposes vellum thickness, DNA, and a return to Bunn's metric, "excluding high-frequency words", with "f106 may well have been followed by f104"; suggestions only, nothing computed.

## 7. Davis

Lisa Fagin Davis, "How Many Glyphs and How Many Scribes? Digital Paleography and the Voynich Manuscript", Manuscript Studies 5.1 (2020), 164-180, https://repository.upenn.edu/mss_sims/vol5/iss1/6
Mark: READ (PDF).
Claims (p. 176): "This very unusual collaboration method bears emphasizing: the work of Scribes 1 and 2 (and 5) in the botanical section varies by bifolia—not by page, text, or quire." The premise behind holding hand fixed, and behind Layfield and Davis's "all four pages of each of the fifty-two extant bifolia were written by a single scribe", with the two exceptions 57|66 and 85|86.

## 8. Dulov

Vladimir Dulov, "Анализ распределения слов в Q13 и Q20" (Analysis of the distribution of words in Q13 and Q20), Mysteries of the Voynich manuscript, https://vladimirdulov.blogspot.com/, 17 January 2019, in Russian, nine diagrams.
Mark: READ (the text in Russian; the figures from the printed bar labels on the diagram images, which reproduce his per-sheet averages to the second decimal). Announced in English as Wladimir D, "Word Q13 and Q20 analysis", voynich.ninja thread 2633, 25 January 2019, https://www.voynich.ninja/thread-2633.html: READ in the archive copy, two pages, which carries no diagrams.
Claims: the share of each page's word tokens whose types occur in a fixed base text, the last paragraph of f116r, averaged per sheet of quire 20 (per cent): 103|116 40.71, 108|111 35.47, 107|112 26.3, 106|113 20.14, 104|115 18.01, 105|114 15.44. "The text on each bifolio was written before they were sewn into the quire Q20" (OURS: the translation). Affinity of each page to one reference text, not a likeness of a sheet's two halves; no null.

## 9. The line-wrapping control

Liudmila Rozanova and Alexander Temerev, "A Glyph Is Not a Letter, a Token Is Not a Word, a Space Is Not a Space: What the Units of Voynichese Are Not", arXiv 2608.17096v1, 17 August 2026, https://arxiv.org/abs/2608.17096
Mark: READ.
Claims: control texts are wrapped to the manuscript's own line-length sequence before glyph statistics are compared (sections 2.2 and 2.4). The nearest published relative of the poured-book control; no page-pair or sheet analysis.

## Not reached

- The academia.edu review (entry 2): 403 on every fetch; DESCRIBED from its index entry only.
- The NSA PDF of the 1976 seminar proceedings (entry 3): 403; Currier READ in the voynich.nu transcription instead.
- The ResearchGate copy of the DHQ paper (entry 1, companion): 403; READ at DHQ instead.
- Dulov's blog was unreachable on the first pass, because the forum's archive hides outbound links; it was found by search on the fourth pass and READ.

## Other works read for the report and cited there, not part of the page's argument

Torsten Timm, "How the Voynich Manuscript was created", arXiv 1407.6639 (2014); Torsten Timm and Andreas Schinner, Cryptologia 48(4) (2023), preprint lingbuzz 007480; Rachel Sterneck, Annie Polish and Claire Bowern (2020); Adam Gaskell and Claire Bowern (2022); Sravana Reddy and Kevin Knight, "What We Know About the Voynich Manuscript" (2011): all READ from PDFs on the first pass, per the report's own record of works read in full (sheet_finding_prior_work.md, "Searches run"); none analyses sheets. voynich.ninja thread 5940 (July 2026), a re-run of the Layfield and Davis quire 20 ordering on ZL3b: READ on the second pass. These marks are for the report, not for the page.