Yes, a positively huge thank you to the London Fortean Society for having me along for one of their evenings. I had a lot of fun covering the Voynich Manuscript and I hope the 100-strong audience managed to walk away with a reasonable feel for what I find entertaining, intriguing and frustrating about it.

What I particularly appreciated was that when (during the Q-and-A section) anyone right at the back asked a question, the entire room went pin-drop quiet so that everyone could hear what was being said. That’s exactly the kind of audience every speaker would like to have. And the questions were really good too!

Oddly enough, I’d never given a talk on the Voynich Manuscript to non-specialists before, so the evening also offered me a nice opportunity to cover a lot of material that I’d thought about over the years but hadn’t really found a way of presenting in Cipher Mysteries.

At the same time, I did deliberately steer well away from Voynich theories (and indeed from almost all Voynich theorists): and noted that as time has gone by, my interest in (and desire to try to answer) any historical question that involves the word “why” has ended up so close to zero that you’d need Roger Bacon’s non-existent microscope to tell the two apart. (Frankly, I find trying to work out what genuinely happened more than difficult enough for me.)

As a salutory tale of what happens when Voynich theories go really bad, all I can really do is point to Dan Burisch and his wonderfully recursive timelines, catastrophes, J-Rods, DNA inventions, etc – here, here, here, and finally here.

Back in the present, the big problem I’m facing is that history tells us that a typical Voynich researcher will study the manuscript for twenty years before being stopped, either by choice (has this ever happened?) or by being forcibly raised to that nymphily balneological structure in the sky. No wonder I’m feeling an increasing sense of urgency, given that it would seem I now only have about four years left to crack it. (And as for Rene Zandbergen, he must be made of awesomely stern stuff, methinks.)

Oh well, all I can do is hope that I’ll be able to come back to the LFS for an update lecture in less than four years with some good news. Fingers crossed! 🙂

I’ve blogged before about the Voynich Manuscript talk I’ll be giving to the London Fortean Society this coming Thursday, but I need to make sure that anyone going realises that the venue has changed.

The new venue is The Pipeline, 94 Middlesex Street, London E1 7EZ, which is located in the City-style Bermuda Triangle of Liverpool Street, Aldgate, and Aldgate East; the last of which three always made one of my grandfathers laugh his head off: he’d once heard a Cockney bus conductor say “Aldgit East, all git aht“.

As to the contents of the talk… anyone expecting a regurgitatory recap of what you read in Wikipedia will (I sincerely hope!) be sadly disappointed, because I plan to cover a great deal of stuff to do with the Voynich Manuscript that you wouldn’t find there, or in fact hardly anywhere on the web. Such as the real deal with Voynich theories, for which I produced a special commemorative meme:

Anyhoo, I’m looking forward to the evening, and (barring any disasters) it should be fun. Very reasonably priced and timed (7.30pm for an 8pm start, £4/£2 concs), and I hope to see some of you there! 🙂

Statistical and cryptanalytical analyses tend to assume that ciphers will fit one of a small number of well-known and well-researched pigeonholes (e.g. Vigenère, autokey, etc). Unfortunately, this kind of “backwards attack” can often be stopped dead if the encipherer includes one or more additional steps sideways, unless the backwards attacker happens to be cunning or lucky enough to reconstruct those tiny steps.

But as they knew in Bletchley Park, it is sometimes possible to “forwards attack” cryptograms. There, a “crib” was the name BP codebreakers used to describe where you already had the plaintext, typically obtained by decrypting the same message enciphered using a different cipher system: having this would help the code-breakers reconstruct daily settings for the second cipher etc. Just so you know, this is precisely why you should never forward a received (and deciphered) message word-for-word using a different cipher, a lesson many WW2 code bureaux stubbornly failed to learn.

Similarly, the idea behind my “block paradigm” methodology is that if we can use secondary historical clues to determine the plaintext from which a given section of ciphertext was derived, we stand a reasonably good chance (I think) of reconstructing the cipher forwards from there. You can therefore think of it as a high-level “historical crib”, where the plaintext is reconstructed via in-depth research rather than by breaking a parallel cipher.

At the very least, this whole process could very possibly yield a completely different class of problem to solve, which in the case of the Voynich Manuscript shouldn’t be a bad thing, given that a century’s worth of backwards attacks has been largely unproductive. 🙁

But what might the plaintext for the Voynich zodiac look like? Would we even recognize it if we had it in front of us?

The Voynich zodiac section

In the same way that many people have long suspected that the Voynich Manuscript’s “Herbal” section(s) probably contains plant and/or remedy descriptions (albeit secret, valuable or unexpected ones), there has long been a strong – yet untested – historical hypothesis about what the Voynich zodiac section might well contain, which is simply this: per-degree astrology. This is because each sign seems to be divided into 30 elements (29 in the case of Pisces, though this may possibly have simply been a slip of the quill), and there are 30 degrees in each zodiac sign (i.e. 12 x 30 = 360).

The modern history of per-degree astrology is something I covered here before: it moved back from Marc Edmund Jones (1925) to the nineteenth century astrologers “Charubel” (who claimed he channelled his per-degree symbols) and “Sephariel” (who claimed that he copied his from “La Volasfera”, supposedly a Renaissance book by Antonio Borelli / Bonelli [did he mean Guido Bonatti?], but this has never turned up).

It’s often written that Western medieval per-degree astrology arrived from Arabic sources via Pietro d’Abano (while he was in Spain during the 13th century). Heidelberg has a 15th century German translation of his work in MS Cod. Pal. Germ. 832 (“Regensburg, nach 1491”), which Rene Zandbergen mentioned in a comment here back in 2009. (If you look at fol. 36r onwards, you can see a few lines of text for each of the thirty degrees in each of the signs in turn, along with some rather jaunty miniatures.)

Prior to the Arabs, you can doubtless trace all this back to the original Indian sources (Diane O’Donovan pointed to the encyclopaedia-sized “Brihat-Samhita” by Varahamihira), but taking things back that far falls way beyond my paygrade, so I’m not going to attempt it in this post. 🙂

However, if you take the time to read Chapter XII of volume III of Lynn Thorndike’s “History of Magic and Experimental Science”, you’ll see that another medieval writer famously wrote on per-degree judicial astrology: and this is where my search began.

Andalò di Negro

Andalò di Negro (fl. first half of the 14th century) was a noble from Genoa. Boccaccio, who he famously taught “in the movements of the stars”, noted that “since [Andalò] had traversed nearly the whole world, and had profited by experience under every clime and every horizon, he knew as an eye-witness what we learn from hearsay” (De genealogia deorum, XV, 6, quoted in Thorndike, p.195).

One of Andalò’s works (“introductorium ad iudicia astrologie”) that discussed per-degree judidicial astrology was of particular interest to me. So, back in 2009, I managed to get some working photocopies of it courtesy of the Warburg Institute’s Photographic Collection: these were of the two known documents listed further below.

Interestingly (and unlike the Pietro d’Abano-derived Heidelberg manuscript), the key feature that seems to oddly parallel what we see in the Voynich Manuscript’s zodiac section is that these two documents contain only a small amount of data per degree (though admittedly arranged in columns of a table rather than in the form of nymphs, stars and labels).

The first document is at the British Library: I managed to get a look at this in person, kindly thanks (if I recall correctly) to a letter of introduction from Dr David Juste, who was then a historian of astrology at the Warburg Institute. One unusual feature was that a few key parts of the tables were highlighted in different colours, something that wasn’t at all apparent from the black-and-white photographs taken for the Warburg in the (I guess) 1920s or 1930s. (Sadly, the colour notes I took at the time have long since disappeared).

* BL Add. MS 23770 (BL: “14th century”, Warburg: “circa 1350”)
http://searcharchives.bl.uk/IAMS_VU2:IAMS032-002097417 – “Letter of introduction required to view this manuscript”
1. “INTRODUCTORIUS ad iudicia astrologie co[m]positus ab A[n]dalo de Nigro de Janua;” with paintings of the signs of the Zodiac, the planetary Gods, etc., ff. 1-44.
Aries (8r), Taurus (9v), Gemini (11r), Cancer (12v)
Leo (13v), Virgo (15r), Libra (16v), Scorpio (17v)
Sagittarius (18v), Capricorn (19v), Aquarius (20v), Pisces (21v)
You can see monochrome thumbnails of these twelve images on the Warburg Institute’s Photographic Collection.

In the case of the second document, since 2009 Cod. Fonds 7272 has been placed online and made downloadable by BNF. As a result, I can include links directly into the Gallica pages for you (which is nice).

* BNF Cod. Fonds Latin 7272
http://gallica.bnf.fr/ark:/12148/btv1b8452771j
Aries (112r), Taurus (114r), Gemini (115v), Cancer (116v)
Leo (118v), Virgo (119v), Libra (121r), Scorpio (122v)
Sagittarius (124v), Capricorn (126r), Aquarius (127v), Pisces (129r)

Here is the Aries table as it appears in BNF Cod. Fonds Latin 7272, courtesy of Bibliothèque nationale de France via Gallica:

By way of a guide, fol. 22v of BL Add. MS 23770 explains (says Thorndike, p.192 n.5) Andalò di Negro’s “fivefold distinction of degrees within the signs: 1, masculine or feminine; 2, lucidus, tenebrosus, fumosi, or vacui; 3. putei; 4. azamena (like the putei, to be avoided); 5, augmentates fortunam.”

However, I have to mention at this point that according to Boncompagni’s (1875) “Un Catalogo dei Lavori di Andalò di Negro” (an offprint taken from “Bullettino di Bibliographia e di Storia delle Scienze, Matematiche E Fisiche”, Tomo VII – Luglio 1874, and for an original of which I paid a load of money several years ago but which is now available print-on-demand from Kessinger *sigh*), there might possibly be a third copy still floating around.

Boncompagni (pp.54-55) mentions that an 1834 alphabetical index of the Biblioteca Altieri di Roma published by Federico Blume lists: “de Nigro, Andali de Ianua, Introductorium ad iudicia astrologiae, Fogl. membr. V.E.5”. Moreover, Emilio Altieri’s index to the Biblioteca Altieri (car. 11a, recto, lin.2) reads: “Andalus de Nigro de Janua, de Astrologia, Pil. 13, Lett. A. Numo. 5”. Yet according to a 1690 index, “Il detto Altieri non possiede ora alcun esemplare manoscritto d’alcun lavoro di Andalo di Negro“, so it seems that it had already disappeared by then.

That sums up the known versions of this work tolerably well: but what might these tell us about the Voynich zodiac? Obviously, that’s a good question something I’ll have to leave for a follow-on post…

A “research tree” is the term I like to use to describe a whole group of evidence / artefacts / phenomena / ideas that are linked together in non-chain-like ways. The term is particularly relevant to unsolved cipher mysteries because you almost always start by not knowing where in particular research trees your cipher fits (if it even fits at all, which they typically don’t).

So, to both recap and expand slightly:

(1) I suspect that the Voynich Manuscript’s Quire 20 contains a sizeable collection of encrypted recipes.
– – Though an old suggestion, we now have a lot of secondary analysis to help reconstruct its original page order.

(2) I suspect we now know enough to be able to match Quire 20’s original structure with that of unencrypted recipe collections.
– – This matching trick is what I call the “block paradigm” approach to cracking historical ciphertexts.

(3) The most likely languages for the plaintext are Italian, Latin, and French.
– – The ‘michitonese’ handwriting appears to contain some Voynichese, and looks to have come from Savoy.

(4) I believe we can eliminate Latin as the plaintext language.
– – This is because Voynichese’s ‘8’ and ‘9’ characters appear to function as ‘contraction’ and ‘truncation’ shorthand tokens, making them essentially incompatible with Latin (where word endings hold a large amount of semantic content).

(5) My working hypothesis is that the plaintext is in Italian (Tuscan).
– – This is because there are a large number of Italian herbals, but very few French herbals.

(6) The various reliable dating evidences we have suggest that this was written between 1440 and 1470.
– – (…don’t get me started on this, or we’ll be here all night.)

(Feel free to disagree with any of the above! I’m not telling you what to think, but making clear the constraints I’m using to guide my own search.)

As a result, I’ve been looking for 13th / 14th / 15th century recipe lists written in Italian. My current hunch is that Quire 20’s plaintext might well be something close to BNF MS Latin 6741 – Jean le Bègue’s collection of paint, colouring, and gilding recipes.

Hence probably the best way to start is to build up a picture of the research tree in which that hunch is located, and then explore it a little…

The Italian colour recipes research tree

For building up an initial view of this research tree, I began with “Original Written Sources for the History of Mediaeval Painting Techniques and Materials: A List of Published Texts” by Salvador Muñoz Viñas, pp. 114-124 of Studies in Conservation, Vol. 43, No. 2 (1998) (many thanks to Juergen W. for this!). [Though because this concentrates on published sources, there may be several significant books of colour secrets out there that it misses, e.g. MS Sloane 416.]

Now that I have pruned the (initially somewhat overgrown) research tree down to more manageable proportions, this is the view I’m currently seeing through my research window:

This research tree has four main branches I now hope to explore in more depth:

* The recipes of Johannes Alcherius (as copied by Jehan Le Begue).
– – Jehan Le Begue’s translation is in volume 1 of Mary Merrifield’s book Original Treatises.
– – Alcherius was in fact a master builder working in Milan, had access to a large number of secrets, and was still alive in the early years of the 15th century: and so would seem to be an excellent candidate for the author of the Voynich Manuscript. 🙂

* Secreti per colori, Biblioteca Universitaria di Bologna MS 2861
– – Description here, transcription here, or in Volume 2 of Merrifield.

* “Ricepte d’affare Piu Colori”
– – There’s an article on this in Archeion, Vol. XV (1933), pp. 339-347 by Daniel V. Thompson Jr, which I hope to read soon. 🙂
– – There may possibly be more in “Trial index to some unpublished sources for the history of mediaeval craftsmanship” Daniel V. Thompson Jr – Speculum / Volume 10 / Issue 04 / October 1935, pp. 410-431.

* MS Sloane 416, “The Venetian Manuscript”.
– – This manuscript also contains a (brief) description of ciphers, which makes it doubly interesting to me.
– – Parts of this might be in Dutch, but I can’t tell properly from the description.
– – I can’t find any good description of this. I’ll probably have to spend a day at the British Library…

Have I missed anything important? Please say if I have! Oh, and I’ve ordered a copy of Mark Clarke’s (2001) “Art Of All Colours”, which looks to be extremely interesting as well….

In a recent post, I discussed a large number of features of the Voynich Manuscript’s Quire 20, to try to get under its vellum skin (so to speak). One thing I’d add is that if you compare the vellum colour of the front (recto) and back (verso) of pages f103 through to f108…

…you can see (I think) if you constrast enhance these…

…that f104, f105, f107 and f108 appear to match each other quite well.

Hence – given that I think f103 may well be separate anyway – I infer from this that f104, f105, f107 and f108 may well have been folded and cut from a single piece of vellum: and hence that they might very well have sat next to each other in Q20’s alpha state. Hence f106 and the missing f109-f110 bifolio may well have been from a different piece of vellum, and so (given that I suspect f105r was the first page of the book hidden in the quire) may well together have formed the two most central bifolios of the quire.

So: given all the above, and that there seems to be a good chance that f108v and f104r originally sat next to each other (as discussed before), I suspect we now know enough to reduce the large list of bifolio permutations for Q20’s original state down to just four good candidates (I’ve only listed the first folio of each bifolio pair for convenience):

a) f103 : f105 f108 f104 f107 f106 [f109]
b) f103 : f105 f108 f104 f107 [f109] f106
c) f103 : f105 f107 f108 f104 f106 [f109]
d) f103 : f105 f107 f108 f104 [f109] f106

Of course, I may be wrong… but I do now think there’s a high chance that this is basically correct.

But to use the block-paradigm trick (i.e. to decrypt a cipher by finding a separate copy of the text from which it came) with these possible candidates, though, we need to also find a structurally matching copy of the hidden book’s plaintext.

So the big question is surely this: how on earth do we find a copy of Quire 20’s plaintext?

Johannes Alcherius

Elsewhere in Cipher Mysteries, I mentioned BNF MS Latin 6741, which is a collection of 359 recipes collected together in Paris by Jean le Bègue / Jehan le Bègue in 1431, and which was discussed by Mary Merrifield in her 1849 book Original Treatises, Dating from the XIIth to the XVIIIth Centuries, on the Arts of Painting.

Yet having now read in rather more depth about Jean le Bègue’s recipe collection here, I suspect that the answer lies not in Paris (the city which BNF MS Latin 6741 seems never to have left) but instead in Milan, and specifically with Johannes Alcherius – arguably the key author whose recipes Jean le Bègue ripped off collected together.

However, the big problem is that while Alcherius wrote in Italian, Jean le Bègue wrote in Latin – so what we have in BNF MS Latin 6741 is actually a Latin translation of Alcherius’ Italian original.

So if I’m right about this, it would mean – unfortunately – that rather than finding the Italian plaintext from which I believe a large part of Quire 20 was derived, I strongly suspect that BNF MS Latin 6741 instead contains Jean le Bègue’s Latin translation of that same Italian plaintext.

So… even though I suspect that the block paradigm trick may have got me close to the finishing line in this instance, the Voynich Manuscript’s secrets still continue to elude us. Close, but no cigar. Or pizza. Oh well! 🙁

How to cross the Quire 20 line?

It seems to me that the single most important piece missing from this jigsaw is the Italian plaintext of Johannes Alcherius’ recipes for colours. I can see two possible routes to achieve this:

(1) Reverse translate Jean le Bègue’s Latin back into the Italian plaintext from which it was derived. Which would be exquisitely nuanced, and very hard to get right, but just about possible all the same. Or…

(2) Find fragments of Alcherius’ recipes floating in other documents, but in their original Italian form (rather than in Jean le Bègue’s Latin translation). It may be that, somewhere in the far recesses of Academe, someone has already searched for this, perhaps as part of a compeletely separate study. But if so, my own digging has been utterly unable to find it.

Perhaps you will have better luck, though!

Things you can help with

Does someone have (or can get access to) a PDF copy of “The recipe collection of Johannes Alcherius and the painting materials used in manuscript illumination in France and Northern Italy, c. 1380-1420” (1998) by Nancy Turner that they can send me, before I start throwing my money at Maney Online (now part of Taylor & Francis)?

Or a copy of “Painting Techniques : History, Materials and Studio Techniques, Proceeding of the IIC Dublin Congress, 7-11 December 1998”, where the same thing also seems to appear?

Or… does anyone have a study listing all pre-1500 Italian colour recipe manuscript fragments? (For what it’s worth, I was unable to find any mention of Alcherius or Jean le Bègue in Thorndike.)

Back in August 2010, I posted up some observations on the Voynich Manuscript’s Quire 20, which included:
* Tim Tattrie pointed out that ‘x’ appears on every folio of Q20 except the first (f103) and the last (f116)
* I noted that these ‘x’ characters often sat next to ‘ar’ and ‘or’ pairs, e.g. arxor / salxor / kedarxy / oxorshey / oxar / shoxar / lxorxoiin, etc.
* Tim Tattrie also pointed out that the paragraph stars on f103 and f116 are notable because they don’t seem to have tails
* The tail-less paragraph stars on f103r looked to me as though they had been added in at a later stage
* Elmar Vogt pointed out that most of the paragraph stars followed an empty-full-empty-full pattern, except for “f103r, f104r and f108r”
* “The notion that Q20 originally contained seven nested quires (as per the folio numbering) seems slightly over-the-top to me”
* f103r doesn’t “look” it should be the first page of Q20, but f105r (with a nice ornate gallows) does:-

From these, I tentatively concluded (way back then) that Q20 might well therefore have originally been written as two separate codicological parts, which I proposed calling “Q20A” and “Q20B”. It was certainly an interesting suggestion… and, as I’ll explain below, one that I suspect was quite close to the truth, though admittedly not the whole story.

The “ytem” / “ydem” star tails

I then posted again in September 2010, where I proposed that the tails of the paragraph stars had been written first: and that they had been then been accessorized with a star to hide them from view.

That is, the tail is the meaning (they all read ‘y’ if you look closely), and the star is the deception. But what does the ‘y’ mean? Well, lists in countless medieval documents are very often itemized using the word “item” or “ytem”: which is why we use the word “itemized”, of course. So it seemed (and still seems) overwhelmingly likely to me that the paragraph star tail ‘y’ was short for “ytem”, and that each Q20 star with a tail marked the start of an item.

If you count up all the paragraph stars with tails, you get the following numbers:

f103r 0 + 19 (i.e. no stars with tails, but 19 stars without tails)
f103v 0 + 14
f104r 13
f104v 13
f105r 10
f105v 10
f106r 15
f106v 14
f107r 15
f107v 15
f108r 16
f108v 16
f109r ?? \
f109v ?? -\ missing
f110r ?? -/ bifolio
f110v ?? /
f111r 17
f111v 5 + 14 (something a little odd would seem to be going on here)
f112r 12
f112v 12 + 1
f113r 16
f113v 15
f114r 13
f114v 12
f115r 12 + 1
f115v 13
f116r 0 + 10

Add up all the stars with tails and you get 264: if the four pages from the missing central bifolio (f109 / f110) each contained 15 stars, we would seem to get a total of closer to 264 + 4 x 15 = 324 items (as opposed to starred paragraphs). I don’t know exactly what that means, but it is what it is.

Quire 20 “Block Paradigm”

When I revisited Q20 in 2014 , it was to propose that we might also profitably look for a “block paradigm” match to Quire 20. By this, I meant that we should go a-hunting for an existing book of secrets from the right time frame from which Q20’s contents might well have been copied / encrypted. As a pretty good candidate, I suggested BnP MS. 6741, which contains a set of 359 numbered recipes (plus various rhymes) compiled from various sources by Jean le Bègue / Jehan le Bègue [1368-1457] in Paris in 1431: in Latin.

But then again, when in the past others have suggested that this section might just as well contain 360 elements (as in per-degree astrology), or even 365 elements (one for each day of the year), it has been pointed out by way of response that the number of starred paragraphs doesn’t seem to fit: we have too many stars. “My God, it’s (too) full of stars”, one might reasonably say (if you are a cinema buff, that is).

This is because if you take Q20 as a whole, you would expect its total number of paragraph stars to be around 323 + (15 x 4) = 383, which is about 20-25 stars too many for (what I, at least, consider) the most likely scenarios (359 / 360 / 365).

Yet if you restrict yourself to “ytem stars” (i.e. stars with tails), it seems that you end up with roughly 324 “ytems”, which would seem to be 35 or so stars too few.

So how do all these odd-shaped jigsaw pieces slot together? Quire 20 would seem to be quite the three pipe problem, as Sherlock Holmes would have (fictionally) said. 🙂

…or is it?

How to read this hidden book

Building on all of the preceding observations (and inspired by comments recently left here by Rene Zandbergen), here’s how I think Q20 was written, and how we should try to “read it” – that is, how our eyes should sequence its pages and comprehend its content.

It seems likely to me that the ornate gallows character on f105r marks not the start of a quire, but the start of a book – or, at the very least, the very first “ytem” in that book. Furthermore, if we group all the pages with “ytem” tailed paragraph stars together and put this at their front, f105r would have been either (a) the first part of a free-standing book (“Q20A”, as I proposed before), or (b) the first part of a book whose presence inside Q20 was concealed by fake paragraph stars. Either way, I now feel confident that we should be reading f105r as if it were the first part of the hidden book.

In which case, the fake-looking paragraph stars on f103r, f103v and f116r would indeed be fake, added to visually conceal the start and end of the book within Q20. f103r has nineteen of these fake stars crammed down its left margin: the more you look at them, the more fake they look, I think.

So: if we put the f103-f116 “fake paragraph star” bifolio to one side, and place the f105-f114 bifolio as the outermost ‘wrapper’ of the hidden book, the question then comes whether we can infer from any other statistical or visual properties what the nesting order (and orientation) of the other five bifolios inside it was (i.e. when the book was in its original, ‘alpha’ state).

Tim Tattrie insightfully noted in a comment here that:

* “lo” as a separate word is only found in f104r, f106r and f108v.
* “rl” as a separate word, or word beginning is only found in f104r, f108v and f113r.
* “llo” as a series of letters is only found in f104r, f108v, f111v, f113v, f116r

I think this suggests some kind of semantic, content-based link between f104r and f108v: and hence that f108v and f104r may well have originally sat facing each other in the original bifolio layout.

Unfortunately, I haven’t yet noticed any Q20 colour transfers dating back to its earliest phases of composition. The big reddish smudge on the top right of f103r (where some of the Voynichese letters inside it have been badly emended in a later hand) seems to have happened after the pages were in their final nesting order, and the same seems true of the (initially green, but then fading away quickly) stain at the centre top of the Q20 pages.

There are also some tiny green paint splodges on the very right hand edge of f104r: but these seem to me to have almost certainly been added by accident when the heavy painter was painting Quire 19 (f103 is a little bit smaller than f104, so the outermost edge of f104 stuck out a little beyond f103, catching the drips from the paint brush).

All in all, the problem with trying to reconstruct Q20’s original nesting order is that it was left in a very clean (i.e. unmessed-around-with) state, probably because it was the least visually intriguing quire. So unless we can find a magical Raman-style way of picking up close-to-invisible inter-folio paint transfers from the paragraph stars, we don’t seem to have much else to work from.

Hence at this point, I’m basically out of ideas: are there any other sources of information (whether visual, statistical etc) you can suggest that might help us reconstruct Quire 20’s original folio sequence?