Weaving the voynich manuscript ~ seriously

[2700 words – with some of Julian’s data added]

I am convinced that it is not as simple as it appears (i.e. that the words are not words at all)

– Julian Bunn

to which the natural response is – well if not words, what are they?


Julian’s post entitled ‘Page positional Gallows’  is filled with data, and in my original post I had planned to set some of his illustrations side by side with others showing pattern-correspondences to standard series in woven fabrics.


I have now received Julian’s permission to include his pictures, so one image in this post has been changed.  Otherwise, I have to ask you to flip back and forth. I have not included many examples; this is about possibilities, not any insistence that you accept the idea. Here is   Julian’s blogpost.


As he says, the distributions of gallows on those pictures aren’t exact matches with positions on the folios, but I don’t think it matters very much in this case. The artificial regularity may be helpful in demonstrating a pattern, if it is a weaving design. In medieval times the series might just be read out, line by line. But one thing I’d expect to be constant and that’s a separate representation for the threading pattern, and (if a more sophisticated loom) the treading pattern.


In Julian’ s picture for  f.58v  for example,  the left-hand margin shows a vertical series separated from the rest –  by grey in Julian’s picture. In fact the interval is filled with non-gallows glyphs.


One section in the margin resembles one of the very simplest  arrangements for thread in a woven fabric. The same folio includes, to the right of the margin a staggered, vertical line of blocks which is typical of some woven patterns derived from the first.

fabric gold blog

techne weaving basic design vertical

Julian 58v marked blog

A weaving pattern will now use the margin (left or right) to record the treading pattern.


The basic threading pattern is placed across the top of the pattern, as shown in the illustration below.

weaving pattern thread and treddleThe treading pattern (vertical series) tells the weaver how to create the wanted variations on that basic pattern.  It’s more-or-less a profile view of the design, and appears in the selvage of many types of fabric.


(The small box to the upper right in the pattern design shown below refers to a second process in the threading-up).


I have never seen any weavers’ instructions from the fifteenth century; I don’t know whether they were gridded. But in any case, it is often easier to record a complex pattern – say for a fine carpet – simply by having someone call out in order the sequence of colours used for the knots. This could be what has happened in some (?) folios of the Vms.


Today a weaver will find the full pattern in any reference book, but  an experienced weaver wouldn’t trouble to draw up the whole design unless it were extremely complex.


Some weavers use numbers for the threading and treading order, but if the warp is multicoloured, colour might be used instead.


Most  threading patterns can be varied  to produce many others.  Here for example are twelve from a single threading-up.


4th.Feb.  The link to that diagram is now broken, though in case you want to chase it through ‘Wayback’ etc., just undo the caps below.



The artificial regularity of Julian’s illustrations, if any were weaving patterns, might come closer to a pattern’s intended form; but in any case the margins should be fairly literal; they are essential in forming any design, where the rest could be  dictated directly, and for a rug, even knot-by-knot.


(not that tribal weavers, in their own regions, would have needed instructions. Customary patterns were known well,  and could be treated fairly casually unless kings and factories formed part of the chain of production and exchange..)


The two most basic forms of weave are linen weave (‘tabby’) and twill. Tabby is the usual 1×1 over-and-under pattern used for linen,

Twill twill fabricis shown  left. Variations on twill include herringbone and ‘birdseye’ which became a specialty of the Scandinavians,  who developed it into an enormous number of variations. The following pair shows (a) variations of birdseye twill contrasted with (b) a random dot-matrix imageweaving birdseye vars and Lorem ipsum blogGridded patterns present theoretically even thickness and distances for both the warp and the weft.


In practice, and by comparison with the pattern-grid, a fabric may seem distorted. By way of example, the illustration at right is  a copy of Julian’s data for  f.103v, stretched, and that below left an 103v stretched detail bloginner Asian [Turkmen] rug. I’ve chosen it because a number of Julian’s diagrams show the equivalent: facing diagonals forming only two sides of a potential lozenge (diamond).

Turcoman rugs inner asia

textile Turkmen carpet gul detail diagonals

To colour differently the opposite sides in a medallion (gul/gol) is characteristic of the Turkmen style, sometimes called the Kurdish.


Glyphs, Number and colour.

This brings up the question of the glyphs, their number and their relation to colour. When I wrote the original post, being ignorant of cipher-techniques, I hadn’t any notion that a system I suggested for arrangement of glyphs and colours was, in a way, a re-statement of the cipher-person’s ‘bilateral cipher’. [three weeks later – it isn’t. The weaving idea is, sort-of. D.]


Here is an old post (ciphermysteries of course) referring and crediting Tony Gaffney with making the suggestion.  Nick himself raised the theme of bilateral ciphers again very recently and  shortly after my original ‘weaving’ post went up a few days ago. This is co-incidental, and about another topic entirely, but perhaps cipher-people may see a technical link in it. Thanks, Nick.


What I had suggested was that if there were seventeen distinct gallows glyphs (an idea I’d picked up at some time), they would offer a natural system for someone used to describing directions the way many eastern maritime peoples did: by reference to seventeen points for a 32-point compass.


Their points were named by stars, rather than by places or winds (as was the habit in the older Mediterranean), so the system gave a unique name to the points of north and south, but used each of the others twice: once on the eastern or bright curve for the point where a star nominally rose, and again on the western or dark side for its formal place of setting.


It was effectively a formalised system, a way of indicating a direction. and its points need not have precise astronomical reference in each case.  But in that way 2 + (15 x 2) made 17 names apply to 32 points, and thirty-two is a nice number, well-suited to recording languages.


A quick check of Omniglot gives 32 components for:

Islandic; Estonian; modern Persian; Coptic; Tigrinya; Russian; Lithuanian; Perso-Arabic; Belarus; Kurdish (Kurmanji-) in former USSR but now using Cyrillic alphabet; Bulgarian.

(Most of them conveniently near the northern road, too).


To use the ’17’ system, you’d need a  couple of  glyphs that were unique, and then fifteen with closely similar forms – for the ‘eastern’ and the ‘western’ as it were.


I’m not suggesting that the Voynich text refers only to weaving and colours; I’d assume colour would be one of many sets of associations built on such a series, the foundation probably alpha-numeric and similar to the way that the western world built upon the alphabet and the content of the Psalter. Ibn Arabi gave sets of associations for each of  27 lunar mansions.


So relationship between a ‘letter’ and sound might be direct, but that between a letter, its position, and/or a colour need not be so simple. As example, the word for green might start with  ‘g’ but by custom the colour imagined the primary element in a people’s cosmology and for that reason be associated with the position/value of the letter ‘A’.


However, two more ideas followed from this:-


First that the notebook is evidently meant to be portable, and so any patterns in it might be meant for use elsewhere – and since it is proverbially difficult to describe colour precisely, it occurred to me that the set of colours relevant to the set of [gallows-?] glyphs might themselves be present, in the manuscript’s palette of  inks and pigments.


Dyes were mostly vegetable dyes, [but see additional post on this point]; in the form of a dye, they won’t usually work ‘as-is’ on parchment.


Often used as a powder in dyeing, vegetable matter has to be mixed with some form of mordant or used as semi-liquid  mixed with a fixative gum. I haven’t properly researched the cross-over between materials used for dyes, and for pigments, yet.


According to this pdf, early printing inks used a mixture of “coloured earth, soot and plant matter for pigments,  mixed with gums for a binder”; in effect, a cross between a dye and an ink.  Calligraphy inks in China used animal-based gums as binders; gum arabic (a vegetable gum) was suited to water-colours; the oldest oil-paint we know was found in a mural in the Tarim basin. – again, see subsequent post/s.


But there seems no reason why it wouldn’t be possible to use some of the same dye-materials to form a manuscript’s pigment, wash or ink since already some served both purposes. Armenian manuscripts tended generally to use vegetable pigments, where Byzantine manuscripts used minerals far more often.


McCrone’s report on the manuscript mentions azurite, a mineral which isn’t used in dyeing (so far as I know – I’m checking this). Otherwise the report is a little vague on the subject, mentioning only that many of the pigments are mixed heavily with a vegetable? gum, one whose nature isn’t specified but which was not south American mopa-mopa.


Mary d’Imperio includes a full account of the description given by a Dr. Carter. I’ve re-arranged it slightly but otherwise here is the list:


Some of the colors appear to be colored ink or water-color, some a kind of crayon, and some an opaque kind of paint like poster paint.


A good, strong brown

… an amber-like ink, like British tan leather goods;

A red ink just like ordinary red ink today,

[ink/watercolour ‘wash’]

a bright, not quite brilliant blue ink or water-color,


an opaque aquamarine,

a good strong red, carmine rather than scarlet or vermillion;

a red that looks like face rouge in color and texture (sic!);

a thick red that makes dots of color that you could scrape with your finger nail.

a red that looks like a bloodstain about a week old.

a blue that sparkles with tiny fragments (not apparently by design)

a dirty yellow (the yellow and browns of the sunflower illustration are like those, only a little faded, of the Van Gogh sunflower picture; the greens are less brilliant):

an opaque green;

a dirty green,

a kind of green crayon,

and several other greens of various hues, intensity, value. and texture.


No black, nothing in the plain pink-to- purple range.  So unless it was made under difficult conditions, this is not a painter’s palette.


What it does match is the medieval textile palette, right across the northern road.


From the western end, and the fifteenth century, this example from the V&A: see article. (On cloth dyes used in medieval Europe, another colour wheel and article here).


15thC colour range V&A


Compare with a  vegetable-dyed silk-swatches from India)


In many cases, especially along the silk- and spice-routes,  the same palette which seems a little  dingy in the west was being used to much better effect, especially by setting similar shades into formal gradations, so as to gain maximum contrast and effect. (The Asian colour palette long ago recognised that colour is perceived by a process of subtraction).


Chinese silk pheasant embroidery

As you see from the example above,  it would be quite possible to get a string which read Blue#1 (i … iiii), Blue#2 (i … iiii), Blue3 (i … iiii) where the (i…iiii) equals the number of stitches, or of threads, over which the same element is continued in a given line.


In this case, it might be agreed that (as Nick Pelling argues, following a line of thought begun by John Tiltman) that the written text includes a great number of what appear to be letter-numbers combinations. Nick believed they referred to page numbers (which might still be so), but they could also apply to description of  a weaving or an embroidery pattern, because no more than four threads are usually covered by an unfixed float.


If there were a correspondence here between the set of colours and the set of  glyphs ( or simply of gallows glyphs), and they were disposed as pairs (and or as darker and lighter tones), that correspondence, if it could be established,  might be a way into identifying the language natural to the maker of the manuscript. A slender hope but a possibility.


vegetable dye colour wheel blog

OMNIGLOT has a table for colours in many languages. I think it was extremely brave of them to post it!


A good bibliography on the subject of medieval handweaving is offered online by Caroline Priest-Dorman.


An excellent article on methods and results in RAMAN  analysis of some western medieval manuscripts is:


Lucia Burgio, Robin J.H.Clark, Richard R. Hark, ‘Raman microscopy and x-ray fluorescence analysis of pigments on medieval and Renaissance Italian manuscript cuttings’, Proceedings of the National Academy of Sciences.. America, Vol.107 (20 ) 5276-5731.


You can see the figures from it online here:


There’s no need to enlarge on the subject of silk- and other weaving, but I would like to make one point. So called ‘Byzantine’ silks appear to have been largely Syrian in origin. Agnes Geijer discusses the point, and it is given emphasis by Charlemagne’s biographers who describe his humiliation at the gates of Byzantium when clothes he had been given there were taken back by the guards. Work done on the issue of magical practices in Byzantium (Dumbarton Oaks) also speaks about the magical or quasi-magical ideas attached to imagery, weaving and specific fabric and dress.I don’t see much point in enlarging on the theme here because the fabric idea is no more than a possibility, but it might be relevant to note at least that one tabu prohibited ‘automatic repetition’ in garments from the iconoclastic period onwards, and the same attitude to automatic repetition of forms  is thought to be a reason why printing technology was first begun, but almost immediately abandoned, in early medieval Islam.


So if the designs were repeat-patterns, I should be interested to know. It would also indicate a possible origin for the type, since the repeat medallion is a typical form for ‘Persian’ carpets of the inner Asian type.


By the way, on the subject of Charlemagne, I must say I was surprised to read in the wiki article about Charlemagne’s silk shroud supposedly being made in Constantinople.

fabric silk Charlemagne medallion

It’s fairly typical of the ‘star-pictures’ found in works from northern Syria, from whose weavers most of the fine silks sold in Constantinople came. WE are told that in response to the insult he received when the city guards of Constantinople demanded back the silk garments he had there, Charlemagne had imported a community of Syrian weavers to his own domains, settling them somewhere on the eastern or south-eastern coast  of France.


Silk fabric which allegorises or embodies constellations and their characters is found well into inner Asia, and in Coptic  woollen tapestries. The custom informs some manuscript illustrations, as I showed earlier, and even made it to the far west for a time.


I won’t elaborate on that either, but as you see Charlemagne’s shroud  is akin to forms used in the  month-roundels, and that celestial ‘car’ defining the “helios” (so called) which occupies the centre of some Roman zodiacs.


Here,  the disposition of the animals and so forth is literally the description for the northern constellation Ursa Major, which classical Greece and Rome considered the north-marker.  Following them, so did medieval Europe as a rule. It had  been known in western Latin regions as the rustic’s cart, in Old English Carles wægn, but thereafter it came to be taken Charlemagne’s car in popular belief. (In renaissance Italy, there is a clever pun on these ideas embodied in a rather bold church fresco).


after – not that the difference was always recognised in Byzantine or medieval works, but the true sun-chariot rightly has a pair of front-facing horses and may but need not always have a second pair harnessed behind the first. Manilius’ text describing Ursa Major is worth reading in the original here.. (D)


For comparison of design elements more generally, this is a  Chinese silk made in the century after Charlemagne lived. (source)

fabric silk Chinese imperial 9thC

The flowers are highly stylised. The yellow colour probably indicates an imperial court robe. The emperors regulated which olours and clothes might be worn by a given class.


So altogether:


Though  Julian was not talking about weaving, he was one of the first to conclude from results of his own research, so as far as I know (and it is often difficult to know who has done or said what in Voynich studies),


I am convinced that it is not as simple as it appears (i.e. that the words are not words at all.

(Julian Bunn, “Computational attacks on the Voynich”, August 23rd.,  2012)


Final note

This isn’t the only suggestion of textiles and textile arts in the manuscript.  In discussing the botanical section last year (not here) I illustrated fairly copiously the way the drawing style in many folios from that  section echoes those of inner Asia ( especially patterns of Suzani work in Uzbekistan) and also types of textiles created near ports of medieval India and south-east Asia.  I wont’ repeat it all, but here are a small number of the illustrations from traditional Suzani work. (The picture in the bottom row, a Javanese batik,  was allegedly made in the 10thC AD, the century after Charlemagne). I haven’t been able to check its provenance, but I do wonder if the caption hasn’t mistaken the Islamic and Christian calendars.









fabric. Coromandel coast 9thC Burma



  1. Hi Diane,

    I love “out of the box” conjectures like these! I’m glad the folio graphics I posted gave you the inspiration. What would be quite breathtaking is if the “text” on each VMs folio turned out to be weaving instructions for a pattern that somehow matched the illustration (if any) on the folio.



    • Thank you Julian. 😀
      I wish I were one of those people who is smitten by sudden inspiration, but I’m afraid I’m more of a slogger.
      The idea gradually dawned because of so many other elements in the manuscript, such as the use of weft-faced woven cloth (clavi) as mnemonic element in fol. 1v, and the similar stylistics in the botanical section etc. to those on fabrics widely traded through the Great Sea during the medieval period, and the arrangement of the figures around the month-roundels (similar to women in coptic tapestries.)

      So it was not an inspiration; more like a possible extension of what I’d found in other sections, and a possible between our very different sorts of work.

      I’m so glad you enjoy the post. Of course the next question raised by this possibility is whether a person well-accustomed to the textile trade mightn’t use the underlying system as a basis for a ciphering system.

      Now I think of it, a weaving loom can be imagined as a kind of enigma machine, wouldn’t you say? It certainly has enough variables.

      Or it could be seen as like a stringed instrument, with each warp-thread like a string/note in a harp or piano, the different treadings thenlike modulations of key.

      Someone else (sorry, can’t recall who at the second) has put a lot of work into translating some of the bathy- section as if it were musical notation – could be another input there.. who knows. I’d not supposed I’d get involved at all with the written text.

      .. New Year’s morning here. Forgive the ramble. And thanks again for the comment.


  2. Postscript (Jan 6th., 2014) I should have added this earlier.

    To my deep shame, the person who had tested possible links between Voynichese and music was … yes, Julian himself!

    It was an idea he tested nearly three years ago, now, and which is recorded in a post to his blog “Computational Attacks on the Voynich Manuscript” (February 26, 2010: ‘Mapping the Voynich Manuscript to Music’).

    Please don’t think that Julian is a bit off-centre. On the contrary, I think it would be fair to describe his approach to the manuscript as highly conservative, and quite in keeping with the work done from the 1950s to 1970s, beginning with Colonel William Friedman’s post-war ‘think-thank’ and concluding with Mary d’Imperio’s booklet entitled ‘An Elegant Enigma’ which was produced for the History of Cryptological Studies in the 1970s, and released by NASA as a downloadable pdf a couple of months after I began working on the manuscript. One of the nicer coincidences.


Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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