The Voynich Ninja

In order to take a fresh count on the recipes I went through Quire 20 and numbered the paragraphs according to my best guess. I then checked my paragraphs against the transcription, as Marco suggested. There were some differences and I accepted or rejected the guesses of earlier transcribers as I thought best.

As an interim I numbered all the Recipes (which is the best name until we know what they are) with roman numerals. Thus we can refer to f113r.V without confusing it for line f113r.5. I hope that eventually we can have a full ordinal count, and that each recipe will have a simple and unique number.

I have attached a text file to this post showing my current best guess. It is ordered by folio, with the line numbers given for each recipe. Where I believe that a recipe is too long and should be two or more individual recipes I had left a comment. I intend to go back and try to split these using whatever distinguishing marks I can find.

I believe that the folios f104r, f104v, f105r, f105v, f106r, f106v, f107r, f107v, f112r, f112v, f113v, f114r, f114v, f115r, and You are not allowed to view links. Register or Login to view. are fully divided into recipes and there remain no problems on these folios. I believe that the last recipe of You are not allowed to view links. Register or Login to view. may be split into two, even though the number of recipes and stars already match on that page.

The folios 103r, f103v, f108r, f108v, f111r, f111v, and You are not allowed to view links. Register or Login to view. need further investigation. It is noteworthy that these folios form two bifolios, the outermost and the innermost (currently) of Quire 20. The lower half of You are not allowed to view links. Register or Login to view. was not numbered for recipes, and You are not allowed to view links. Register or Login to view. was not considered.

I will happily answer any questions with my guesses, and welcome anybody who can help with the troublesome folios.

So what I've decided to do is tabulate each recipe and the occurrences of the fifty most common words. I will then consider each recipe as a group of common words, and compare it with others. Some will be similar, others dissimilar. I hope that some recipes will match each other strongly, showing natural divisions between the 300 or so recipes.

I also hope that some of the most strongly matching recipes will show similarities in where and how the shared words occur within the text. For example, if two recipes share five words and they occur in the same order, it might reveal some underlying structure.

I think I could also link recipes using rare words or phrases (as far as they occur). There's the strong potential for building up a web of relationships which might lead to some useful observations. I can see so many ways of looking at and comparing these recipes. I'm surprised nobody has thought of this before.

Emma: some sections are just understudied Smile

Your idea looks very promising. Looking forward to reading the results.

...some things are well studied, but results were not published

Hi Emma,
I went through your analysis, comparing Takahashi's transcription and with the scans of the ms.
Here are my observations. I don't think I could add much of potential use. Your analysis seems excellent to me. There is a very high correlation between stars and recipe boundaries, but there are a few exceptions and dubious cases.

f103r - 19 stars, 18 recipes
III - 8-12 (5 lines): Possibly two recipes to account for extra star, but no obvious break.

The red star corresponds rather precisely with line 9, bu the first word (oteeos) does not seem like a candidate as the first of a recipe. I agree that considering this as a paragraph with an extra star is the best option.

f103v - 14 stars, 13 recipes
VII - 20-26 (7 lines): Are there two paragraphs here?

Recipe VII could include only line 20. Line 21 beginning with tokain could be the first of a new recipe. It is dubious.
Takahashi's transcription has some anomalies in line numbering (lines 28a, 36a): I can't see why, since all lines seem pretty regular to me.

f105r - 10 stars, 10 recipes - Complete
III - 10-12 (3 lines): Including short line written above.
X - 34-36 (3 lines):

Takahashi's lists 37 lines. He also counts separately the very last short line (which is centered, not aligned to the left). I think the 10 recipes are OK. Maybe we could list III has having 4 lines (counting the short line as well?).

f108r - 16 stars, 14 recipes
III - 8-13 (6 lines): Almost certainly two paragraphs, 8-10 and 11-13?
XII - 39-44 (6 lines): Likely two paragraphs, 39-40 and 41-44?

Takahashi splits III (he has a paragraph-end marker on line 10). I wonder why it did so, but that does fit with the position of the white star at the beginning of line 11.
Same for XII: Takahashi splits it in two. Possibly the way in which lines 40 and 41 diverge is what makes that point a likely paragraph boundary?
I think we could go on and split the two paragraphs as you suggested, so that we match the number of stars.

f103r - 19 stars, 18 recipes
III - 8-12 (5 lines): Possibly two recipes to account for extra star, but no obvious break.

The red star corresponds rather precisely with line 9, bu the first word (oteeos) does not seem like a candidate as the first of a recipe. I agree that considering this as a paragraph with an extra star is the best option.

f103v - 14 stars, 13 recipes
VII - 20-26 (7 lines): Are there two paragraphs here?

Recipe VII could include only line 20. Line 21 beginning with tokain could be the first of a new recipe. It is dubious.
Takahashi's transcription has some anomalies in line numbering (lines 28a, 36a): I can't see why, since all lines seem pretty regular to me.

f105r - 10 stars, 10 recipes - Complete
III - 10-12 (3 lines): Including short line written above.
X - 34-36 (3 lines):

Takahashi's lists 37 lines. He also counts separately the very last short line (which is centered, not aligned to the left). I think the 10 recipes are OK. Maybe we could list III has having 4 lines (counting the short line as well?).

f108r - 16 stars, 14 recipes
III - 8-13 (6 lines): Almost certainly two paragraphs, 8-10 and 11-13?
XII - 39-44 (6 lines): Likely two paragraphs, 39-40 and 41-44?

Takahashi splits III (he has a paragraph-end marker on line 10). I wonder why it did so, but that does fit with the position of the white star at the beginning of line 11.
Same for XII: Takahashi splits it in two. Possibly the way in which lines 40 and 41 diverge is what makes that point a likely paragraph boundary?
I think we could go on and split the two paragraphs as you suggested, so that we match the number of stars.

f108v - 16 stars, 14 recipes
VI - 20-29 (10 lines): May be three paragraphs, but no obvious breaks.

Takahashi splits it in two (20-27,28-29 – I ignore that he also lists line 27 as 26a).
Line 28 starts with a good candidate “pororaiin”, but it does not match any of the two extra stars. I can't see of any way of meaningfully solving this situation.

f111r - 17 stars, 5 recipes
II - 6-35 (30 lines): Multiple paragraphs
III - 36-43 (8 lines): May be two paragraphs
V - 48-54 (7 lines): May be two paragraphs

II: no idea.
III: the star could suggest a split at line 39 (starting with “saiin”). Very dubious. You are not allowed to view links. Register or Login to view. 26 has a white star near a line starting with “sair” and other similar (but more dubious) cases.
V: Takahashi splits it in two (48-50,51-54). Line 51 has a red star and starts with a very good candidate: “polkeeo”. I think this recipe can be split this way.

f111v - 19 stars, 7 recipes
I - 1-25 (25 lines): Multiple paragraphs
III - 29-33 (5 lines): May be more than one paragraph
V - 37-44 (8 lines): May be more than one paragraph

I: this recipe has 10 stars, but it seems impossible to split. It might as well be a single anomalous recipe.
III: this paragraph has three stars. The stars and the initials could allow a splitting into three recipes (29-30,31,32-33) starting with gallows: polkiin, pochey, tar.
V: two stars. Considering that line 38 is slightly shorter and line 39 is marked by the second star, it could maybe be split into (37-38, 39-44). The starting word of line 39 is not particularly appropriate (dsheey), but see also You are not allowed to view links. Register or Login to view. 49.

f112r - 12 stars, 12 recipes - Complete

Takahashi has fewer paragraphs. He ignores the second star and has lines 1-10 as a single paragraph. He also ignores the last two stars and has lines 37-45 as a single paragraph. Your proposal seems to me much preferable.

f112v - 13 stars, 13 recipes - Complete

Takahashi differs also in this case. He also counts 13 paragraphs, but they don't match the stars. He counts three paragraphs (instead of two) for lines 20-26. He counts the last 6 lines as a single paragraph (instead of two).

f113r - 16 stars, 16 recipes
IX - 25-26 (2 lines)
XVI - 45-51 (7 lines): May be two paragraphs.

Takahashi's count is similar. The last recipe has two stars and there is no star at line 25, but I have nothing different to suggest.

f114r - 13 stars, 13 recipes - Complete

Takahashi counts 4-10 as a single paragraph. They are clearly two distinct paragraphs, as stated in your analysis.

f114v - 12 stars, 12 recipes - Complete

Takahashi counts 6-10 as a single paragraph. Splitting it into two seems OK to me: it fits the position of the star on line 8 and the first word of line 8 is also OK (pchedchdy).

f116r - 10 stars, 6 recipes
IV - 10-14 (5 lines): Two paragraphs for two stars?
VI - 18-30 (13 lines): Multiple paragraphs

Takahashi also counts 10-14 as a single paragraph. There is an unaccounted star, but it's difficult to say if there is a recipe starting on line 12 or 13. I think the initial of 13 (sarain) is a better candidate than that of 12 (quoain) but both are dubious.
Takahashi counts 18-30 as a single paragraph. There are four stars corresponding to these lines, but they are very hard to split.
I also agree that the following star-less long paragraph starting at 31 doesn't seem to fit with the rest.

Hi Marco, thanks so much for taking the time to add your thoughts.

f103r.III: I agree that [oteeos] doesn't sound like the first word of a paragraph.

f103v.VII: A break after the first line is by far the best option, but I'm dubious.

f105r.III: I will make the changes you suggest.

f108r.III: I'll split this, as I think you and Takahashi are right on balance.
f108r.XII: I think Takahashi's suggestion is the best due to the divergence of the lines, but there's still a doubt in my mind that the break should be a line lower.

f111r.III: Splitting at 38/30 is the best suggestion, as 38 does also end a little short. But like you I don't find [saiin] a good candidate for the first word.
f111r.V: You're absolutely right, I will split this.

f111v.III: I'll split between 30/31, with a new recipe beginning [pochey], but I'm not convinced of the second split.

f112v: I can see where Takahashi is coming from with his different paragraphs. I admit there is some uncertainty here.

Hi Emma,
if you provide an updated list of recipes, I could try to "number-crunch" something. I was thinking of listing and counting the common words between all possible couples of recipes. Couples with a high rate of common words could then be manually searched for patterns (this step would be much harder to compute).

Hi Marco, I would be happy to post the list. Though I'm actually already do the work you suggest. Finding patterns is incredibly hard, however, and I think it will take weeks to do a decent job.

Here is a tentative matching of words using the transcription and recipe boundaries by Takahashi: I selected couples of recipes with a particularly high number of common words (normalized by the length of the paragraphs).
Emma, if you think this could be of use and you post an updated list of paragraph boundaries, I can adapt the scripts to your input.

You are not allowed to view links. Register or Login to view.

Sorry for the late reply, but I thank you, Marco, for your very generous offer.

I have attached the updated file to this post.

Emma May Smith

Emma May Smith

Koen G

Davidsch

MarcoP

Emma May Smith

MarcoP

Emma May Smith

MarcoP

Emma May Smith