Starting from scratch #5 A gathering of gatherings

I have the following cluster diagram from Julian Bunn, whose analyses of the written text are published in his blog, Computational Attacks on the Voynich Manuscript. The original is included in ‘Folio Similarities’ (February 26th., 2010) online. The diagram shows degrees of similarity between vocabulary in each folio, and Julian says of it that he noticed that ‘incidentally’ “the red circles generally match Currier Hand 1 and the blue match Currier Hand 2”.  The purple rings around some folio-numbers in the version below mark the folios chosen by Rene Zandbergen and ORF for radiocarbon dating. As you can see, the choice does not constitute a representative sample in the scientific sense but has the advantage of providing a reasonable date-range for folios written in the “Currier A” hand and its related vocabulary – which some researchers refer to as a Currier language.

currier Hand FolioClusters and samples for radiocarbon test final

The four folios tested are, as best I can tell, those listed below, but the information varies between different sources, some referring to Folio 1,  (Bunn’s red group ~ near the dividing line) on which Jakub Hořčický’s name is inscribed, in Czech – then known as the Bohemian language – ‘Jakub z Tepence’.[1]:

*Fol.8;  *Fol.26; *Fol.47 (on which see comments here); *Fol.68 …. and * ? Fol 1.

In terms of the quires (gatherings), these equate to:

Folio 8 – Quire 1; Folio 26 – Quire 4; Folio 47 – Quire 6; Folio 68 – Quire 9. So in fact – apart from f.68 – all samples came from the top six quires – that is, the top third of the manuscript since it presently contains 18 quires, another two posited [16] and [18].

If our present manuscript has been compiled by extraction, then the dates returned for those –  in the ‘Currier A’ hand – may relate to a single prior manuscript or may indicate a consistent date for transcription. It does not appear likely that any original composition would be presented in this way, on the more expensive vellum but without the usual indications of formally-produced works.  There are exceptions, but the rule in the medieval world is that an original composition was presented formally, usually composed by eminent persons whose words were first transcribed on ephemeral media and then produced in formal style.  The type of the viliores is presumed by default to contain works copied from earlier sources. Formal works of Jewish provenance also follow the style of local Latin works. The North French Miscellany is one example. (details given in an earlier post.)

manuscript Hebrew Northern France_______________________

In a post entitled ‘Voynich radiocarbon dating, part 1‘, Nick Pelling posted graphs to represent the radiocarbon results.

The same information is also presented, with argument and expansion, at Rene Zandbergen’s site – from which I have taken only the date specified for each individual folio.

Pelling’s graphs treat the folios in their manuscript order. I have sliced his image to list them by the apparent (raw) result, which makes a more obvious connection to Julian’s cluster diagram – a correspondence which might be pure co-incidence, or which might indicate evolution over time. Nick stands by every word in his post linked above – he just said so when giving permission for me to re-use the graphs. 🙂

In fairness I have to add that I do not yet share assumptions informing much of the discussion of this data, chiefly: (i) that the work is properly considered homogeneous, so that (ii) one may take the mean of the four dates and apply it as dating for the manuscript as a whole. My own view is rather that there is so marked a variety of styles and content expressed by the imagery  between some sections and others, and the division between ‘Currier A’ and ‘Currier B’ so marked, that the question of whether our manuscript derives from a single source or constitutes a collection from disparate sources gained, perhaps, over years (whether by copying and/or by extraction) is at present a question that should remain open, given the bias [2] in sampling.

Whether the difference between “Currier A” and “Currier B” signals no more than a rota of scribes at work, or a substantial disparity between content, sources and dates of inscription is thus still unknown.

As for such descriptions as  “Herbal A” and “Herbal B” and so on:  they are convenient but should not be supposed substantive.


None of my available sources specifies whether the foliation used for reporting the tests, or for any particular writer’s discussion, is that of the Beinecke Library or a system adopted by some members of the Voynich mailing list. For want of better knowledge, I show the folios according to the Beinecke library.


folio 68-iii ~ recto (Bunn’s red group); verso Bunn’s blue group.

raw data returned: c.1400 AD

folio 68 r and vNick's calibrated-data folio 68______________________________________________________

folio 26 (Bunn’s blue group)

c. 1436 AD

folio 26 r and vNick's calibrated-data fol 26


folio 47 (Bunn’s red group ~ distant from the dividing line)

c.1444 AD

folio 47 r and vNick's calibrated-data fol 47


folio 8 (Bunn’s red group)

c.1461 AD

folio 8 r and vNick's calibrated-data folio 1~ for a full discussion of Pelling’s diagram and his opinion of the data see his recent post, ‘Radiocarbon dating and the Voynich manuscript’, ciphermysteries, May 14th., 2015 ~


The radiocarbon results suggest that while the material of folio 68 was made at the time of either the “Currier A” or the “Currier B” style, some of what is written on that folio comes from each period and/or hand – the date for the “B” remaining effectively unknown, folio 26 being the only folio sampled with text all of the ‘blue’ (Currier B) type.

It is interesting that the folios containing “B” script/language tend towards the higher (earlier) range, and those with the red towards the lower, so that “B” might be earlier than “A”, yet the test was insufficient for any firm conclusions.

Some researchers have expressed concern over the condition of folio 68, and especially that part of it from which the sample was taken, but for our purposes it is enough to note that the results’ apparent disparity (1400 as against the 1430s-60s) is matched by a marked difference of style in this folio’s drawings, as against the remainder of the manuscript, and even between the verso and recto.  Such things lend credibility to the idea of collation and stages of addition, against that of the work as an original, homogeneous, composition. There is a possibility raised that some folios – and here a fold-out – had originally a blank side later filled by a different hand.

If folio 68 is included in the set, then the results allow for incription of ‘Currier A’ folios to have taken place over an extended period –  more than two generations – (1400-1461) whereas if it is not included, the other three and, let us assume, their quires, can be seen fitting within a mere thirty years (1436-61). In the latter case, since one of these three (f.26) contains text exclusively from the ‘blue’ group, so the overall range for inscription would be narrowed. All this, of course, if the data were accepted as representative.

Quire 1 (fol.8) c.1461 – Currier ‘A’

Quire 4 (fol. 26) c. 1436 – Currier ‘B’

Quire 6 (fol.47)  c. 1444 – Currier ‘A’.

Like the disturbed binding and variation in drawing style – the insertion of a ‘B’ among the ‘A’ type could add further weight to the view that the material was collated from loose quires or from manuscripts in quaternis, the latter gaining further support from the various marks of former stitching and extraneous needle-holes.[2]  It is weakened by the fact that we have no other examples to date of script like that in the manuscript, and that these variations may not signify a change of hand or abbreviation method but a change of language or of subject, with consequent change in the vocabulary.

All this might hold –  if the data were solid rather than, as it is, rather fragile.

From a scientific point of view, the lack of randomisation, and of duplication leaves us still with a question-mark over whether the manuscript was transcribed within a short space of time, by several scribes, or over a far longer period.

On the other hand, manuscripts are routinely dated and analysed by other parameters, so this question-mark while disappointing is not a tragedy. 🙂

 Postscript – I have been asked why I refer so often to ciphermysteries and so little to other sites such as Zandbergen’s.  The reason is that Pelling is always perfectly clear about the difference between fact and personal opinion, offering comparison of his preferred view to a range of alternative opinion, and never leaves one in doubt as to whether a thing stated as ‘being so’ reflects the result of his own research, objective data, a personal opinion or the adoption of another’s work and its conclusions. I do not find these distinctions so clear on any other site, my own too often guilty of the same.  One cannot always recall from where a small piece of information came that was read several years ago.



1. I take the reading of that inscription from Jan Hurych’s letter to the old Voynich mailing list (14 Mar 2004). Jan says that it is written in Czech, rather than the more formal Latin.

2. ‘bias’ in the purely technical sense; I am not suggesting an ideological bias on the part of that television company or those conducting the tests.

3. On this general question, a useful set of studies in A. Giorgi, S. Moscadelli, D. Quaglioni, Il notariato nell’arco alpino. Produzione e conservazione delle carte notarili tra medioevo ed età moderna. Atti del convegno, Trento 24-26 febbraio 2011, Milan, 2014. [Title trans: The notarial profession in the Alps. Production and storage of notarial documents from the Middle Ages and the modern age. Proceedings of the conference, Trent, 24 to 26 February 2011].

  • For use of the term ‘gathering’ in bookbinding, definition from a good introductory site, the ‘Medieval Manuscript Manual’.


  1. Diane, I’m sympathetically reading your dialogues and the research you’ve done with every presentation. I’m hoping you will follow up on two references I’d like to contribute to your blog:
    The Medicinalibus Indorum Herbis written by Martin de la Cruz (an Aztec physician) and his collaborator and translator Juan Badiano. Both gentlemen were teachers at the Colegio of Santa Cruz in Tlaltelolco. The document was written for Fray Sahagun. It had been sent to Spain in 1552. It eventually ended up in Cardinal Barberini’s library in the 17th century. Subsequently, in 1902, Barberini’s Library became part of the Vatican Library.
    In May of 1990 Pope John Paul II returned the manuscript to Mexico.
    I can’t recap the entire 10-page explanatory discussion of how THE manuscript re-appeared some 400 (+ or -) years later — except to refer you Fermin Herrera’s fabulous little bi’lingual “Hippocrene Concise Dictionary: Nahuatl-English/English-Nahuatl . Also of consuming interest to me is another beautiful book: “An Aztec Herbal – The Classic Codex of 1552” written by William Gates (Dover Publications)
    Finally (ahem) one last referral to Fray Sahagun’s Twelve-Volume “Florentine Codex-General History of The Things of New Spain” — Books One and Eleven
    PS: I haven’t been able to get wordpress or other social sites work for me. Thanx for ‘listening’ !


  2. Bobette – if MS Beinecke 408 were a copy of any Nahuatl text, I would expect to see some evidence in the manuscript that the (posited) transcribers tried to reproduce the style or motifs characteristic of Aztec art. Lack of any such indications, especially with regard to representing the sun, is particularly telling, I think. But perhaps some of the linguists will unite in confirming your translations, which would justify revival of that old proposal made first, I think, by Jim Comegys, and which relied partly on an acceptance of O’Neill’s now discredited identification of a “sunflower” in the botanical section. Comegy’s book was/is entitled,
    Keys for the voynich scholar : necessary clues for the decipherment and reading of the world’s most mysterious manuscript which is a medical text in Nahuatl attributable to Francisco Hernández and his Aztec Ticiti collaborators

    Published in 2001 in Madera, California.
    Have you read it?


  3. Bobette – another generally negative indicator is absence of any recognised new world pigments in the Vms. There is a useful list of those found in Nahuatl codices at

    I’m afraid that MS Beinecke 408, as an artefact, offers nothing in support of the Nahuatl thesis, at least as far as its analyses and description have been provided so far. Some information may be being kept from the public, but one can’t argue from negative evidence in this case, I think, when the positive evidence runs completely counter to the thesis – to date.


  4. Footnote about frequency diagrams. I’ve just fallen over a conversation that took place about this manuscript in 2009. [t-b] had decided that the “gallows” letters might be no more than a form of rubrication, unrubricated…

    [A.C.] Very nice! By the way, have you had any luck with using a word frequency chart instead of a letter frequency chart, according to your new theory of how the letters are written? Perhaps that would give a much better hint at what language it might have originally been.

    Not got that far yet – did find 4 Peruvian, 4 Indonesian & a couple of African languages that matched those frequencies!!!



  5. Diane: the set of bifolios to be sampled was selected with a specific set of criteria in mind (mainly trying to select as wide a range of different types of pages within the four for which they had budget).

    However, these criteria imperfectly overlap numerous other criteria that people blithely tend to assume were used, such as reliability, dating precision, consistency, uniformity, etc: and they were also all assumed to have been used at once in order for the maths formulae to work.

    It’s a difficult old thing to comprehend: and once The Scientific Hand Hath Writ a date range with 95% statistical confidence, people just assume that it’s game over for anything that doesn’t fit this perfectly. Of course, historical research never plays out that easily… but what’s a person to do, eh? 😦


    • Nick – the program-makers were of course perfectly entitled to choose whichever folios they wanted. But in the purely objective sense, the choice was not remotely representative, nor even randomised. So what you get is a set of individual tests, but not ones which can be applied overall.

      As far as I can work it out, the selection was based on purely subjective ideas about the different sorts of subject matter in the pictures on a given folio – not the way to date a manuscript, if that’s what you want to do.

      I can’t help it that the program-maker’s choices cut across the normal method for scientific testing; but that’s what happened. The lab was only asked to test those folios the program-makers chose, whereas in the normal way you’d hand the thing over and say “we have a budget of x amount; we have permission to remove slivers from four folios. Can you get a dating for the whole manuscript from that?” Personally, I’d have said it wasn’t possible. With more than a hundred folios, and only four to be tested, you would only have a hope of getting a fair dating for the whole if you were absolutely certain already that it was a ‘homogeneous’ production – e.g. a printed manuscript whose binding was original. If obliged, you’d do the randomisation, and adhere to the standard procedure from the first. In this case, it wasn’t possible. The scientists were asked only to act as lab technicans, and that basic slog I’m sure they did meticulously.


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s