The author’s rights are asserted.
Postscript – I should make this two separate posts, I suppose, but I’ll just wait twice as long before posting again. 🙂
earlier posts in the series
2b. The ‘4o’ (updated and revised)
Before being interrupted by revisiting folio 5v, we were considering how to approach the manuscript while still avoiding the constantly-recurring errors of the study’s last hundred years.
Among the most persistent recurring errors noted were
- (i) confusion of ‘gut feeling’ with evidence-derived opinion;
- (ii) guesses mis-called hypotheses – wrongly identifying an idea which has no foundation in fact with an opinion about solid preliminary data or precedent studies;
- (iii) over-estimation of abilities and skills;
- (iiii) expounding a fictional narrative as if laying out a logical or plausible argument, and
- (v) personal weakness which led various individuals to attempt – often successfully – to cover deficiencies in their knowledge, abilities or work by adopting a tone of authority and/or social superiority and
- (vi) unreasonable certainty.
Pronouncements made by Elizebeth Friedman offer an interesting example of (v) and (vi) combined.
N.B. In historical research what you think is less important what first led you to think so.
For those methodological errors, the first and most enduring model was Wilfrid Voynich’s narrative delivered in 1921. The vast majority who approached the manuscript later either adopted his assertions or his flawed methodology, or both.
From the available evidence, Mr. Voynich was a professional and competent evaluator of a manuscript-as-object (a question which had never been asked, so far as I can discover, until I looked into it a few years ago).
*The example I discovered, and discussed in detail, concerned a doubt that arose over whether an item which Voynich sold to Robert Garrett of Baltimore had been accurately dated and described. Before publishing a summary of my research into this question of Wilfrid’s competence, I looked for any precedent and for any previous mention of that item in Voynich writings. There was none.
However, Wilfrid was no historian – not by training or by inclination.
The crafting of history is called a discipline for good reason: it requires talent and training to be honed by keeping up with current studies and work from the historical documents. It also requires the researcher to constantly consider and to show that he/she has formed their opinions from a wide range of primary evidence and secondary sources. It is the range, nature and balance of evidence which marks good research. That’s why we add footnotes and bibliographies – so that an informed historian of the subject or period will have a clear picture of a work’s limits and omissions, as well as what we have taken into account.
Very little of what Wilfrid said on that day appears to be outcome of his own research, but is only material adopted and arbitrarily adapted from Newbold’s work – which Newbold presented in a lecture delivered on the same occasion in 1921. Wilfrid did not acknowledge the debt.
These days, among historians, to avoid honestly admitting where information or new insights came from is regarded even more seriously than it was in 1921.
Wilfrid says almost nothing about the manuscript’s history, apart from his having decided on seeing the manuscript that it was an original text written and enciphered by a thirteenth-century Englishman named Roger Bacon. .
Then, supposing his flash of imagination had been a flash of inspired insight, Wilfrid ran with that unfounded and still unsupported notion. It is not beyond all possibility that he might yet be proven right, at least about the current text’s copying a thirteenth-century one. At present, though, the balance of evidence is against an origin in England.
Provenience vs. History
The bulk of what Wilfrid presented in 1921 was not a ‘history’ for the manuscript but a quasi-historical story imagining its subsequent travels – the sort of information which, when accurate, was called in his day ‘provenience‘.
Let me clarify that.
Properly speaking, a manuscript’s history consists of the history of those things which make up the the manuscript: the materials, the history of techniques in writing-styles and image-making; and such precedent works as are directly reflected in the manuscript’s written text and images.
In this strict sense, then, manuscript’s history ends when the pages were inscribed though might continue to the time when those pages were bound.
The object’s provenience describes the finished (or more-or-less finished) item’s subsequent travels and vicissitudes.
*re: ‘provenience’. I’ve used the older term, seen as late d’Imperio’s Elegant Enigma, because it makes this distinction quite clear. The present-day habit is to speak of ‘provenance’ which simply means where something came from, and can refer both the manuscript’s provenience and to the history of e.g. an image used in it.
Describing an object’s provenience has always been part of the seller’s task, and of the librarian’s. It doesn’t aim to explain the lineage of text or imagery now in the manuscript, so much as to describe, owner by owner, how the (more or less) finished object came to be where it is now.
This is why a library’s catalogue description may describe a manuscript as a fifteenth-century French Psalter, but will not explain the history or origins of the membrane, nor technical details of inks and pigments, nor the Psalms’ history and origin. That Psalter’s provenience, but not its history, begins in fifteenth century France.
The history of the same manuscript, if investigated, may find that the membrane had been imported from England; that the scribe had been trained in Picardy; and that the written text originated in the near east, individual Psalms having been composed 1800 or 1500 etc. years BC.
Were a manuscript’s history identical with its provenience, we’d be asserting that the Psalms were invented in fifteenth-century France by a previously unknown French king David.
So a history of how the manuscript came to be is rather different from a provenience which says what happened to the object after it was made.
Most Voynich storylines focus on the time the vellum was inscribed or concentrate on that section of its provenience which occurred, or is alleged to have occurred, between the seventeenth to the twentieth centuries.
It should be fairly obvious that calling provenience-narratives the manuscript’s “history” is inappropriate and an unfortunate habit because it has often seen assertions about provenience, and especially seventeenth-century and later provenience, back-projected onto the manuscript’s materials and content as a means to explain them.
With the quasi-historical provenience narratives forming the single largest group of Voynich-related writings today, and serving as basis for the creation of still more tenuous guesses as hypotheses, I think the most pressing need at present in this study is a development of solid historical research in the strict sense of the word.
,That is to say, I think it would be to the study’s benefit if ‘where did the manuscript go?’ gave way to a focus on ‘where did the content come from?’
To illustrate why ‘match the picture’ isn’t that sort of investigation, I’ll revisit the Voynich archer (again, *sigh*) in a later post.
I do accept, absolutely, that formation of working hypotheses is essential to the research being done by those who are convinced (rightly or wrongly) that the written text is enciphered, I see no reason for research into the manuscript’s history to begin by inventing or adopting a quasi-historical narrative as theory.
Which brings me to another distortion resulting from supposing other aspects of Voynich research should use techniques employed by those who see the Voynich text as a cipher-text.
To sum up what has been said so far:
First, that there is a distinction to be made between the manuscript’s history on the one hand – defined as prelude to manufacture, and on the other hand narratives about the finished manuscript’s travels. The ‘travel-history’ is provenience.
Second, there is a distinction to be made between the methods of cryptography and those of translation.
Failure to observe the second distinction is why, for example, Mr. Ardiç’s attempting to explain the written text in terms of spoken language is meeting fairly aggressive opposition from individuals whose way of compiling statistical data about the written text has as its basic theory a one-to-one correlation between sound-values and subjectively-defined (and possibly erroneous) interpretations of the glyphs.
The potential for error is magnified by cryptologists’ habitual use of a Roman-alphabet analogy [EVA] for the Voynich script.
Late edit: (10th March 2022) I’ve just seen this same point made by Darius in a comment to Nick Pelling’s latest post, ‘Voynich Paper Suggestion #4: Unpicking the k/t Gallows‘, ciphermysteries, (March 9th 2022). I rather think it will prove to be an observation made earlier and more often. If I find a precedent in posts to Reeds’ mailing list, or if a kind ciphermysteries reader knows who should be credited as first, I’ll add a comment below this post.)
To appreciate the distinction, you might consider how a person trained in reading and writing English, French, Spanish, Italian or German perceives the relationship between the spoken and the written versions of their own language. They expect it to be fairly close, one sound to one glyph, though that has not always been the case.
In German, before library catalogues were computerised the sign for ‘sch’ was considered a single glyph and was listed last among the ‘s’ entries.
To impose that same expectation about the relationship between written and spoken Irish, however, would be foolish.
If you compiled statistical data from written Irish, minus (say) vowels and diacritics, there is little chance it could be read off by a novice to sound and sense as spoken Irish does.
On the other hand, a native speaker might be able to correct the errors and omissions of the transcription. The same is true of pointed and unpointed Hebrew texts where in addition to pointing you might have to consider cantillation.
What we’re seeing happen at present with Mr. Ardiç, is a not-exactly-bullying effort to force him to make his reading of the text conform to the observations and statistics of people working at one, or at two full removes – by first transcribing and then rendering as EVA a text whose language they do not know.
That they cannot get any consistent or intelligible results from the ‘transcribe, create Roman script analogy – match and translate’ method suggests the problem at present might be less Mr. Ardiç’s translation than that the model to which he is being all-but- ordered to conform is inadequate.
A speaker can elucidate the written representation of language, but the opposite isn’t necessarily true. I mean – can you, using EVA, write the sound “Ardiç” with no more than five symbols?
To the Conference organisers, I’d say “Do let the man speak his piece” and leave the critical comments to competent specialists attending. Is there any Voynichero claiming competence in medieval Turkic languages?
Since this series of ‘Notes’ is meant to suggest positive avenues for newer researchers, I’ll move on.
I think it’s entirely possible to engage in researching the manuscript’s history (sensu strictu) without having an preliminary new, or second-hand theory to follow.
One alternative is to look for a detail which intrigues you about the manuscript, its materials, writing or images and start looking more deeply into that.
It doesn’t have to be a grand, all-encompassing theme. It can be quite simple such as – how many scripts which existed before 1435 contained one or more ‘4’ shaped letters/glyphs and what sounds did they represent in those scripts?
Then, after you’ve done that, another question is likely to arise? How, if at all, are the scripts in my list related to one another? There should be a mass of historical evidence and discussion about those things. You don’t have to contribute something wholly without precedent to contribute valuable material to this study for the first time..
Another approach is to look for items where the usual story doesn’t seem to match the manuscript’s own testimony. For example, the usual story is that the Voynich calendar is a ‘zodiac’. If you actually look at it though, it is not. It contains only ten months, not twelve and while theorists will wave that way and just say the (imagined) missing months were cut away, that offers no reasonable evidence for why only ten-months of the year would be wanted. I can bear witness to the fact that this is no easy question to research.
Somewhat easier is the question of why the Voynich calendar has some doubled months. That they are not accidental duplications is clear enough; there are differences between the organisation of the tiered figures and also way the month-name is written differently on each of a pair.
There’s a lot of solid scholarship on the history of calendars in the world beyond Voynich studies and wiki articles, so attempting to discover where and when those particular months were, or might be, doubled in a pre-1435 calendar from somewhere in the world should prove interesting and if it’s hard work you may get a solid result – say a list of regional and religious calendars in which the same months are doubled. So there you have another contribution to the study. That list, properly researched and documented, is a contribution in itself. You don’t have to make it fit any Voynich theory. You don’t have to use it to create a Voynich theory. You’re doing the important work of putting what is actually in the manuscript into a still-fairly-general historical context.
What you can expect to find in doing original work of this sort, starting from ‘external’ (i.e. non-Voynich focused) historical studies is that each question researched will usually throw up additional questions about the manuscript and about the ‘provenience-narrative’ sort of assertions made about it.
You needn’t try to create, reject or endorse existing theories about, say, whether the calendar was supposed to serve astrology.
I think it helps, as you work, to keep in mind that the manuscript isn’t the problem. The problem is our ignorance; and that’s what you’re working to alleviate.
I suppose I should illustrate how a Question-First approach works in practice.
Here, in outline, is one question I’ve been working on intermittently for some time. Theorists tend to gravitate towards things they perceive as ‘similar’ to their theoretical norm, but in iconographic analysis I find that attention to differences is more interesting and is usually more informative. And this can apply to historical research as such.
Specialists and Wilfrid’s ’13thC England’.
I’ve always wondered why, when Wilfrid asserted the manuscript made in thirteenth century England, there was no outcry about that dating or provenance (origin) offered by the best-qualified and experienced commentators from whom opinions are recorded between1921 and c.1952.
Among the persons who thought that sounded a fair attribution of date and place (not authorship or content) were Goldschmidt, Panofsky, Thorndike and perhaps even Richard Garnett – then keeper of what is now the British Library’s collection of medieval manuscripts but which were held, in Wilfrid’s time, by the British Museum (which still owns them).
Garnett guided and advised Wilfrid, who came to London as a refugee and bookseller, specifically advising him to set up a shop in London rather than hawking volumes about to potential buyers. Garnett’s name is also on Wilfrid’s application for British citizenship.
*’Garnett’s name..citizenship’ see e.g. Colin MacKinnon, ‘The Naturalisation Papers of Wilfrid Michael Voynich'(2013). (online as a pdf). His sources are given.
Now we know the vellum dates to the early fifteenth century, so one has to ask what it was that prevented those specialists openly doubting or rejecting Wilfrid’s thirteenth-century date?
Was it the second-rate quality of the vellum? the manuscript’s dimensions? The absence of, say, logwood ‘pink’ in the palette? The unflattering depictions of female figures which most saw as ‘crude drawings’ because they are unlike anything we see in western Christian art during the thirteenth, fourteenth or fifteenth centuries?
We know that Panofsky insisted it pre-Renaissance, though in his time the ‘Renaissance’ was often treated as covering the mid- fifteenth and earlier sixteenth centuries.
This question – why did they not dispute Wifrid’s date and assignment to England – is why I looked into that question ‘ Was Wilfrid a competent evaluator of medieval manuscripts?’ I concluded that he was, but simply as someone who could give a fair attribution to date and place.
So should we conclude (as in fact was earlier done) that what we have is a fifteenth century copy of a thirteenth-century work?
But competent and experienced evaluators looked everything – the materials, the finish, the range of colours used, the disposition of material on the page and so on.
So what was it about the manuscript which led them all to see the Voynich manuscript as a thirteenth-century and probably English work?
And the question which intrigued me most was that none offered the objection by which Fr. Theodore Petersen rejected O’Neill’s ‘Columbus theory’ – palaeography.
It has become the norm that the Voynich script (‘hands’) are said to most resemble the humanist hand (so called) whose origin is ascribed to Poggio Bracciolini and other early students of Greek in medieval Italy. But if the flow of the Voynich script were obviously unlike any seen before the early 1400s, why didn’t those specialists, or Wilfrid himself, instantly opt for a date and attribution in early fifteenth century northern Italy?
The people I’ve named were pre-eminent in medieval manuscript (and art) studies. Goldschmidt’s two-volume study of Renaissance bindings is a classic. Panofsky was a specialist in medieval art and literature, who moved later into focus on Renaissance art and culture. We know from his correspondence, his published works and even his terse and information-neutral responses to Friedman’s embarrassing ‘questions’ that Panofsky studied palaeography too.
So why did not object, before 1952, to Wilfrid’s ‘thirteenth-century English’ attribution?
Robert Steele had reservations. He describes the velum as ‘coarse, even for the thirteenth century’. Though he, of course, also presumed a wholly Latin European origin for the manuscript and everything in it.
I made a list – vellum? palaeography? images? text-block’s dimensions? and when there was no more pressing issue to deal with, I researched one after another to discover what those specialists saw about the manuscript which suggest to them all that a thirteenth-century English origin looked ok to them.
I would have liked to add ‘colours’ but we must look to the Beinecke library for a full technical description of the Voynich palette. If we ever get it, the information could as easily be ‘ho-hum’ as rivetting and deeply informative. It all depends on the results.
To speak first of the dimensions – to see if there was something characteristically thirteenth century and/or English about the manuscript’s dimensions, I made a fairly rough survey of the British library holdings with results leading me to focus on a period from 1350AD to 1450 AD, and I took into account both folio dimensions but and text-box dimension, because a later copyist had no obligation to copy empty margins precisely.
In the usual way a survey of that type would be a waste of time; many manuscripts have text-blocks whose dimensions are no longer original, having been trimmed down in a later re-binding.
Had the Voynich manuscript shown signs of such uniform trimming, I’d not have troubled with that question unless I happened to find myself with weeks to spare and near some great library, However, on Rene Zandbergen’s asserting that the Voynich text-block had never been trimmed (the only time I had a question answered from that source), I went ahead.
Barbara Shailor, wrote the description for the Beinecke catalogue in 1967, giving the dimensions as 225mm x 160mm. I’ve seen other numbers on Voynich sites, though none with any explanation or evidence for their departing from these measurements.
Parchment. ff. 102 (foliation, s. xvi, Arabic numerals; not every leaf foliated) + i (paper), including 5 double-folio, 3 triple-folio, 1 quadruple-folio and 1 sextuple-folio folding leaves. 225 x 160 mm.
The results of my survey were more interesting than I’d expected and led on to research into the parchminers’ networks operating across Europe during the thirteenth and fourteenth century, especially that of the Jewish parchminers whose work is best documented for Avignon during the third quarter of the fourteenth century.
Two points in particular were of interest: one is that manuscripts having the longer measure of 225mm are* chiefly Jewish works and not only ones originating in the south-western Mediterranean, but in the Greek-speaking Aegean.
*within the British Library’s collection of mss dated 1250-1420 AD
My posts published through voynichimagery on this topic were: ‘Dimensions 160mm x 225 mm’ (Mar 6, 2015); ‘Dimensions and Proportions Brit.Lib. Mss 1340-1450 AD’ (20th June, 2013); a focus on Brit.Lib. MS Harley 623 – ‘Dimensions and Places’ (10th June, 2013). The posts cover both membrane and paper.
Points of interest were that (i) membrane sheets were subject to standardisation in some regions of Europe with dimensions for membrane and paper being brought into alignment and (ii) evidence exists for researching further the question of whether the Voynich dimensions’ agreeing so closely with those for certain non-Latin papers may be evidence that the parchminer rather than the person commissioning the copying is responsible for the present manuscript’s dimensions – that is, the dimensions of the text-block. Here, needless to say, I had to consult any recent studies of the pecia system.
*summaries of all the above research were published through voynichimagery.
Studying the parchminer’s networks and the text-block’s dimensions (or, more fruitfully relative proportions) had allowed me to resolve the question of why Panofsky said “Spain or somewhere southern with evidence of Jewish-Arabic influence” and to understand that it wasn’t irreconcilable with the opinion of other specialists who concluded the work made in England by one or more Latin scribes.
The work led me think, increasingly, that the flow of the Voynich script, its general if not compete absence of diacritics/cantillation and its evenness was due to interaction with Jewish scribes and more likely Sephardic than Ahkenazi script. It wasn’t a firm conclusion – more an sense of the research’s direction.
And there I left the problem. I’m not a trained codicologist or palaeograper and amateur efforts shouldn’t try to push too far, as I see it.
The best result from that research was the consistency with which three intertwined themes constantly recurred: the regions of Anglo-French culture; the Jews of the Aegean and south-western Mediterranean; the inter-actions of paper and of membrane suppliers in the south-western Mediterranean. This all linked easily with information gained by researching other sections of the manuscript. Perhaps these maps will help clarify why a provenance ‘English’ ‘French’ or ‘Spain or somewhere southern’ aren’t necessarily incompatible. England in France; Majorca in ‘France’.
A note included in ‘Dimensions 225mm x160mm’ reads:
“a leaf (vellum?) described by as “of Hugh of St.Victor, De sacramentis, II,6:V (Migne, Pat.Lat. 176, col.587D), 225mm. by 160mm., remains of double column, 35 lines in a fine early gothic bookhand, capitals touched in red, still mounted on a binding, France, mid twelfth century” – quoted from a Southeby’s catalogue of 2014
Hugh of St.Victor, who died in 1141, had developed a new system of mnemonics, fully described by Carruthers.
and into ..2022
As a result of looking further into the curious lack of ruled lines in the Voynich manuscript – the ‘difference’ in this case being between some of my earlier observation and description of points of comparison between details in the Voynich map and those in a couple of early cartes marine gridded ‘by the Rose’.
The current question was ‘Why are there no neat, Euclidian geometric forms in the Voynich manuscript?’ Compiling a preliminary reading list for Euclid’s geometry in medieval Europe, I came across this seminal paper by John Murdoch:
- John E. Murdoch, ‘Euclides Graeco-Latinus: a Hitherto Unknown Medieval Latin Translation of the Elements Made Directly from the Greek’, Harvard Studies in Classical Philology, Vol. 71 (1967), pp. 249-302. [JSTOR]
The article can be read online through JSTOR, but here I’ll first repeat that manuscript’s description… Paris, Bibliotheque Nationale, Fonds latin 7373, 178 fols., 18 x 27 cm., vellum, 13thC AD.
Then note that Murdoch attributes its making to ‘northern Spain/southern France’.
and finally show an example of what Murdoch describes as an ‘artificial humanist hand’ but which I prefer to consider a proto-humanist hand, one reflecting that fusion of Sephardic, Greek and (to a lesser extent) Latin scribal traditions:
So altogether, I think I have now a better idea of why none raised objection to a thirteenth-century English provenance. Leaves of vellum were being produced in the thirteenth century which came from the Anglo-French or ‘Norman’ regions, in which there were Jewish communities involved in a Greek-to-Latin translation movement and in the making and distributing of both membrane and paper for manuscript production. Only Panofsky had immediate second thoughts – because he knew you don’t get ‘shapely’ female forms in the west until somewhat later than the thirteenth century.
Even then, though, you don’t find in western art images of females having over-large heads, small high breasts, very thin arms, and heavy thighs but very thin, sometimes almost bone-thin shanks. It’s an odd combination but – to answer another research question – no, not altogether unknown. Here’s one example of images using similar forms as an established habit (and notice the ‘armoured’ fish. This isn’t the only region or period we find comparable forms, but there aren’t many.
12 thoughts on “O’Donovan notes #3 Selecting a question.”
“Was Wilfrid a competent evaluator of medieval manuscripts ?”
Well, I would say Wilfrid Voynich was a bookseller, no more and no less. He was certainly not an expert on the subject of “literature of the late Middle Ages”. In my opinion, the attribution of the VMS to Roger Bacon was only a marketing strategic measure. A “big name” simply sells better.
Personally, I am increasingly coming to the conclusion that the search for such big names is an aberration. Of course, it is tempting to consider well-known authors with an often well-documented biography. Objectively, however, it is at least as possible that the author in question is a complete unknown.
Matthias – The easiest way to explain how I came to conclude that Wilfrid was competent to date and place a manuscript – which is all I say – is to send you a copy of my essay. I’ve never for a moment believed the manuscript had a single author; the work is so clearly a compilation. I took quite a few years’ flak before my view on that point became more prevalent in Voynich studies. So in fact you and I agree on both these points.
P.S. It was part of the mindset of the nineteenth and early half of the twentieth century to think in terms of ‘important people’ as the movers of historical events. Pretty much everyone involved in Voynich studies from 1912 -c2009 believed the work must have a single author. Belief that the ms’ written text was a ciphertext played a large part of the idea too. Even Panofsky assumed an ‘author’ though he never accepted it was English, and very soon realised our present ms’ couldn’t be dated to the thirteenth century.
My point here was an opposite question – why didn’t Wilfrid’s ascribing the ms to the thirteenth century and to England meet immediate objection from the more eminent specialists? What was it about Wilfrid’s opinion AND about the manuscript-as-object which made that seem reasonable at the time? Wilfrid’s competence to date and place the works he sold was one factor to be considered. He was not ‘just’ a bookseller. He was a trader, specifically, in medieval and Renaissance works and I found only two instances where his description of a work was doubted; in one case he was positively vindicated; in another he withdrew the sale on his own initiative and repaid the purchase price because he, himself, had been misled. He was competent – and honourable. He was just not good at deeper historical research.
LikeLiked by 1 person
“… the observations and statistics of people working at one, or at two full removes – by first transcribing and then rendering as EVA a text whose language they do not know.”
Someone who isn’t using a transcription alphabet of some sort is making exactly the same type of judgements about which sets of ink strokes are “the same” as which other sets of ink strokes as someone doing an explicit transcription — the critical difference being that they’re doing it implicitly in their head rather than communicating those judgements explicitly in a way that allows replication of what they’re seeing/doing. This is a bug not a feature.
Brumbaugh provides a good example of the problem this creates. If you work through his supposed labels in f100r in detail, it turns out that Brumbaugh is seeing two different characters for Currier ‘R.’ In the first type of ‘R’ the backward c comes out of the middle of the diagonal stroke, in the second it attaches to the top. He lumps that 2nd type of Currier ‘R’ in with Currier ‘2’. Which is fine if that’s how he sees the script, but his readers shouldn’t be expected to have to excavate that by working backwards from his final result.
Having said that, it is absolutely a good idea for anyone working with the text to spend some time transcribing it for themselves to have a better feel for the raw script. If you don’t like EVA, use Currier or V101 or a scheme of your choosing. If you have some sort of hangup about using Latin alphanumerics, use whatever you want because for the purposes of transcription it doesn’t really make any difference whether you call a particular bunch of “similar” groups of adjacent ink strokes ‘H’ per Bennett and the First Study Group, ‘P’ per Currier or Darius, ‘qp’ per Frogguy, ‘t’ per EVA, or ‘k’ per v101.
“If you compiled statistical data from written Irish, minus (say) vowels and diacritics, there is little chance it could be read off by a novice to sound and sense as spoken Irish does.” This is true, but misses the point. The point is that if someone thought there was a mapping from Irish consonants to a Voynich transcription scheme of their choosing, then the statistical properties of the Voynich text should fall comfortably within the distributions of those properties derived by converting some body of Irish texts from the same period into “Voynichese” using the proposed mapping. That’s not a substitute for translating long enough continuous chunks of texts to convince specialists in Nth Century CE Irish that the result is plausible in terms of vocabulary, grammar, and syntax, but it’s a good bet that if the theory can’t pass the statistical comparison then it’s not going to pass a linguistic one.
LikeLiked by 1 person
Karl, I understand the point you’re making. My point is that the statistics are statistics about the doubly-removed transcription and may have as little relationship to the language informing Voynichese as would statistics compiled from a faulty transcription-and-romanisation of the way spoken Irish is rendered in script.
The usual computistical approaches in Voynich studies make the same dubious assumptions that the Friedmans did and which include e.g. an expectation of standardised grammar and orthography.
I sometimes feel that the theorists and statistical types have forgotten that just as spoken language is not written language, so statistics are only statistics of their interpretation of the glyphs, which itself involves quite a number of assumptions ranging from the dubious to the doubtful to the probable, but never reaches certainty.
In demanding that all attempted translations conform to one or more of the statistical studies they may be actually demanding that the translation conform to their erroneous model.
What I say is simply – let Mr. Ardic speak. Let people competent to speak about medieval Turkic languages judge his translation. The statistical types might even learn something useful.
“I sometimes feel that the theorists and statistical types have forgotten …statistics are only statistics of their interpretation of the glyphs,…” — That ceases to be an issue once someone like Darius drives a stake in the ground and puts forward a specific hypothesis because now we can talk about the statistics of *his* interpretation of the glyphs.
Darius claims that the underlying plaintext is in a 12th cent. CE dialect of Persian (https://theoricalist.blogspot.com/2021/08/voynich-manuscript-final-piece-of-puzzle.html) which has been converted from the normal script used for Persian to Voynich glyphs in accordance with the table at https://theoricalist.blogspot.com/2021/08/an-introduction-to-voynich-manuscript.html and then obfuscated with nulls that can be stripped out using the rules at https://theoricalist.blogspot.com/2021/09/iteration-theory-of-voynich-manuscript.html.
So…if you take some set of 12th cent. Persian texts and map them to Voynich glyphs using his theory, how do the entropy stats of those texts compare to the entropy stats of the Voynich text after you strip out the nulls? How similar are the glyph frequency distributions of the de-nulled Voynich sample and the converted-to-Voynich-glyphs Persian text samples? For that matter, where do the h1 and h2 values of the de-nulled Voynich text using Darius’ transcription scheme fall relative to known natural languages period? Those are questions that have nothing whatsoever to do with my (or Rene’s or Glen’s or anyone else’s) interpretation of the glyphs or any existing transcription of the mss, They are purely questions about what happens when you apply an empirical test _to his theory using his assumptions_.
(None of which is intended to pick on Darius, by the way, it’s just that his theory has been a recent topic of discussion, so it was convenient to use it as an example.)
“The usual computistical approaches in Voynich studies make the same dubious assumptions that the Friedmans did and which include e.g. an expectation of standardised grammar and orthography.” — I’m not sure that was an assumption as much as a reflection on the relative availability of machine-readable corpora of manuscript texts. When Bennett first wrote about the issue in ’76 he only had access to whatever he typed in himself. In principle, an earlier version of the Helsinki corpus referenced below was available on the web when Dennis Stallings wrote about the issue in ’98 (http://ixoloxi.com/voynich/mbpaper.htm), but in fairness to Dennis it probably wasn’t as easily discoverable if you didn’t know about it. The impact of variability in the grammar and spelling of different authors in older texts on high-level statistics like 1st and 2nd order conditional entropy (h1 and h2) is a purely empirical question, and can be addressed empirically. For example, see the discussion of entropy values from historical texts in Lindemann & Bowern (https://arxiv.org/abs/2010.14697).
Yesterday evening just for grins I pulled four Middle English texts from the corpus at https://quod.lib.umich.edu/c/cme/browse.html:
* The book of quinte essence or the fifth being,,,
* The Desputisoun Bitwen the Bodi and the Soule
* Iacob and Iosep; a Middle English poem of the thirteenth century
* The Life of St. Dunstan from The early South-English legendary ; or lives of saints. I. Ms. Laud, 108, in the Bodleian library
and four 17th century English texts (the period you were using for examples of spelling variation in a recent post) from the Helsinki Corpus (https://www.helsinkicorpus.arts.gla.ac.uk/display.py?what=index):
* an excerpt from Boyle, Robert: Electricity & Magnetism
* an excerpt from Evelyn, John: Diary of John Evelyn
* Strype, John: Letters (to his mother)
* Oxinden, Elizabeth: Letters (to her mother-in-law)
So 8 different authors, multiple genres, two different phases of English. Eliminating digits, punctuation, and spaces, and converting everything to upper case, the h1 values range from 4.12 to 4.21 and the h2 values range from 3.47 to 3.58, i.e. the values are still in a fairly narrow range despite the variability between authors (and even the substantial evolution of the language between the two sets of texts). A quick check of a sample from _Bullfinch’s Mythology_ yields h1 = 4.16 & h2 = 3.52, so basically even with 600 years of linguistic change between “Iacob & Ioseph” and Bullfinch the 1st and 2nd order entropies haven’t changed that much (which should be encouraging for Darius, as it suggests he could do at least an initial entropy test using newer samples of Persian if 12th cent. CE Persian texts aren’t readily available). Obviously, different languages/periods might give different results, but that’s a “might” that should be tested against data rather than assumed…
re precedent for Darius comment – and mine- see the very full survey in Karl’s comment to ciphermysteries
I was addressing transcription issues there rather than precedents for the idea that more than one plaintext script letter could map to a given Voynich glyph. With regard to that, in Brumbaugh’s theory any of the English plaintext letters (A, E, O, U, J) could map to any of the glyphs represented by Currier as (M, O, A); both plaintext Q & A mapped to Currier ‘8’. Given the limited number of published theories older than Brumbaugh’s, that’s probably the earliest one to suggest something like Darius’ point (2) in his comment. Again, that’s with the caveat that the same thing is true for functionally different reasons in a polyalphabetic cipher of the sort Strong suggested (i.e., different plaintext letters will map to different glyphs depending on where the plaintext letter’s position lines up with the repeating numeric shift sequence, as opposed to the mapping being position-independent as it is in Brumbaugh’s and Darius’ theories). Actually, it’s probably the case in Newbold’s cipher, but I’d have to dig into my notes on the details of how Newbold’s cipher worked to be sure. Not sure about Feely’s cipher.
Before receiving the comments from you that I’ve just approved, and which come up with date stamps April 15, 2022 2:24 pm and April 15, 2022 2:59 pm, I had already posted my notice about those precedents.
Your second comment adds to that, for which I thank you.
As for your first comment – I have no opinion about what language(s) if any may inform the written part of the text, and I see no sense in creating or maintaining theoretical narratives.
I have not devoted my energies to the written part of the text, and as a result can have no opinion about what, if any, language(s) might underlie Voynichese, not about the objective worth of earlier observations and/or theories about it, including e.g. theories that it contains ‘nulls’. For that, it would be better to consult a few
persons well qualified in non-Voynich-focused cryptology, linguistics and techniques of statistical analysis.
About your criticisms of a more recently-come researcher – is obvious that, in common with many Voynich theorists, you have the potential to exercise analytical-critical skills.
Perhaps, instead of directing those skills outward as weapons to discourage others, you might begin looking analytically and with an equally critical eye at the premises and evidence informing your preferred theories.
It is advice which should have been given Wilfrid Voynich in 1921, to the Friedmans in the 50s and to promoters of the theories which today dominate public discussion of this manuscript, so please don’t take the advice too personally.
Just go over, from the most basic of your premises, each link in your chain of argument and ask seriously of each – ‘why do I believe this is true?’
[comment edited to clarify and shorten]
Karl, I feel that my reply was a bit abrupt, so let me give an example of why you should test your premises. The examples you give test the evolution in formal, written English over a period. Fair enough. But all the examples *are* from formally-educated and fairly highly educated individuals using ‘written English’ which even today is distinguished from ‘spoken English’ directly rendered.
In doing that, you have assumed orthography itself to be standardised in earlier English works and to alter over time but not, for example, by region (as it did. Individual scribes can sometimes be identified by region because rendering the sound of a word in their own local dialect).
What you haven’t considered – perhaps – is that the Voynich text may not be in cipher, may not have been inscribed by someone highly educated in the underlying language, or might be a foreigner’s effort to reproduce a text, or the sound of a language, with which they had no previous familiarity.
If you were to collect all known examples of Latin efforts to reproduce Arabic words (that’s an easy example) then it would be interesting to see whether statistical analysis of that vocabulary returned results comparable to literary Arabic texts.
I do not think there is enough solid evidence to allow an opinion about whether, and how much, of the Voynich text is composed of prose, of poetry, or simply of vocabularies, technical instructions in brief, or (as another possibility) of litanies. Try doing statistical analysis of Hisperica famina and see how you go. That defeated ‘decryption’ for centuries, until Jane Stevenson published her translation.
I’ve also seen examples of what was called a ‘magical text’, and whose Phoenician origin was ridiculed by the author who transcribed it. What that author hadn’t the means to realise was that it was a multilingual litany to what I suppose you’d call the ‘great Mother’ or ‘Queen of Heaven’ and the litany consisted of her name in language as disparate as Hindu, Latin and classical Egyptian.
More things in heaven and earth, Karl… The history of human thoughts and deeds is not only stranger than we imagine, but stranger than we can imagine. So I try not to use imagination (theory-making) too much in undertaking historical research. 🙂
In case your interest is piqued – I first read Stevenson’s translation of Altus Prosator in an online journal devoted to Syriac studies. Can’t find it online today, but the journal of Celtic studies appears to have published a paper by her with the same title in 1999:
Jane Stevenson, “Altus Prosator”, Celtica 23 (1999): 326–368.
I am concerned that I’ve expressed myself badly in connection with Panofsky’s opinion.
For a short time, after seeing some photostat pages, he accepted that it might be very early – by which I suppose he meant as early as the thirteenth century. He never supported an English origin for the work and after spending two hours with the manuscript attributed it to ‘Spain or somewhere southern’, noted Jewish and ‘Arab’ influence in the drawings.
The reasons he gave for a later date were not – interestingly – due to consideration of the manuscript’s codicology OR palaeography.
As he says, he noted that ‘shapely ladies’ do not occur in western Christian (‘Latin’) art before the fifteenth century.
Panofsky’s error, if any, was to go along with Wilfrid’s assertion that the work had a single author; which in turn led to assumptions initially (i.e in 1932) undisputed by Panofsky, viz. that the text originated at the time of manufacture; and perhaps too that the content (not just manufacture) had its first origin at that time and within the the political territory of western Christian Europe.
But the opinion he gave in 1932 was the first in which a person of his eminence opposed Wilfrid’s theory of English origin and also – with reasons given – that thirteenth century date.
Few knew of that opinion and so the ‘Roger Bacon ciphertext’ story continued to dominate public comments for decades afterwards – for which, see Reeds’ Bibliography in the ‘Cumulative…’ page in the top bar.
For Karl and other readers, as example of the ‘who’d a thunk it?’ moments in research, here’s a little from another of Jane Stevenson’s papers. This information surely broadens our horizons in thinking about languages and scripts that might have been available to some thirteenth-century English scholar.
Stevenson speaks of..
the evidence which we have for the presence of a variety of Ephraimic and pseudo- Ephraimic writing in the library of Canterbury in the seventh century. This is the obvious place to start, since the presence of a number of Syriac and Greek works in this library, the contents of which formed part of a syllabus taught to an unknown number of Anglo-Saxon students, offers a clear and plausible avenue of transmission for such material into the far West.
The fact that the Persian saint Milus of Susa, whose life was not translated into Greek, is included in the Old English Martyrology, with the place names in corrupt Syriac forms suggests that this text, for one, might actually have come to England in its original language: if it did, Theodore is the most likely person to have brought it. It would hardly be surprising if Theodore knew Syriac. In his later life, he was clearly a man with a gift for languages: besides his native Greek, and the Latin he acquired in his middle years, he learned Old English in his old age, and a complicated pun in a work which I have suggested was written by him, suggests that he may even have picked up a little Irish….
Theodore had come from Tarsus and to keep him company in Anglo-Saxon England and see that his teaching conformed to the Latins’ theology and didn’t stray into the Syrian or Greek opinions, he was accompanied by the abbot who had recommended Theodore for the position as head of the Anglo-Saxon church. He was a dark-skinned and very learned North African Byzantine abbot, named Hadrian and he had been appointed abbot in Naples.
Jane Stevenson, ‘Ephraim the Syrian in Anglo-Saxon England’, Hugoye: Journal for Syriac Studies, Vol. 1.2 (1998 [and 2010]): 119–163.
I think I mis-remembered which of Stevenson’s articles I’d read in Hugoye and which in Celtica. I read both in 2000, as best I can recall.
On the library of Canterbury
“The Cathedral and priory of Christ Church, Canterbury was a leading centre of learning and book production from the early Middle Ages, and the library of the community was one of the largest known collections of manuscripts in medieval Britain. A surviving booklist dated before 1331 lists nearly 2,000 volumes held by the library … and it is estimated the library held several thousand books by the dissolution of the monastic foundation in 1539.
From the 1550s onwards, books were taken in large numbers into private collections, destroyed completely or recycled in parts as binding waste for later records.”
from ‘The lost library of Canterbury Cathedral: Digital resources to reunite manuscripts and fragments’, Teaching the Codex (blog) April 15th., 2020.