The Voynich Ninja

Quote:." In Old French, the diphthongs oi/ai/ei were interchangeable and varied according to region and dialect.

We'll, it's those regions an dialects we're trying to identify after all.

Great work, eggy! I don't even want to imagine how much effort this took. But we are learning, gradually.

When playing around with ordering by date, I searched through all of the "16th,15th,14th,13th century" entries to check if the dates couldn't be narrowed down any further. I discovered that there are often other references that sometimes helpfully narrow the date range. However, to my absolute dismay, I also discovered that Gallica sometimes gets the entire century wrong. I don't know how, I don't know why, but it's Gallica's fault!   I assumed it must be the classic 15th century = 1500s mistake, but it was the other way around.. 

For example, Épinal. Ms. 243 was correctly entered as 16thC (Gallica's displayed date), but actually dates to between 1470-1480. To be fair, I guess it's technically only 20 years off, but "16th century" implies a potential date up to 1600AD so it's quite the difference. 

Anyway, I went through the uncertain dated items and updated them where possible. If I updated the date or the location, I added the source(s) that made those claims in a new column. Many are in a far narrower date range now, but it is important to note that there are multiple that have conflicting claims. Biblissima, JONAS, ARCA, and other sites often include expanded information too: handwriting style, locations, languages, provenance, and descriptions of specific passages, which may be helpful in the future.

The lesson to be learned from this: Trust no one!
Especially not in academia. I've witnessed similar dating errors or conflicting citations by different authors. Sometimes such dates are simply matters of opinions. I know it's tedoius but it pays off to verify whatever you can yourself. Thanks for your hard work!

Yeah, this is an issue. When a date is given as "16th century", it may mean that they didn't have all that much to go on.

The problem, I've found, is that these digital catalogue entries are often taken from stuffy old paper catalogues made at some point in the past. These were not always as thorough as we'd expect today, and either way, much of that information may now be obsolete thanks to new insights in the research.

Case in point, the Beinecke description for the VM, which was based on a pre-radiocarbon dating description made by someone under the spell of the Sunflower. Even though the people at the Beinecke I communicated with at the time agreed that it should be updated (including the then-curator Ray Clemens), this was apparently not the most straightforward thing to do. Nor the most pressing matter on the library's hands, for that matter :)


Thanks for all the work you keep putting in, eggy! I also see you found another interesting German entry in the bottom there.

(28-05-2026, 09:02 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Yeah, this is an issue. When a date is given as "16th century", it may mean that they didn't have all that much to go on.

The problem, I've found, is that these digital catalogue entries are often taken from stuffy old paper catalogues made at some point in the past. These were not always as thorough as we'd expect today, and either way, much of that information may now be obsolete thanks to new insights in the research.

Case in point, the Beinecke description for the VM, which was based on a pre-radiocarbon dating description made by someone under the spell of the Sunflower. Even though the people at the Beinecke I communicated with at the time agreed that it should be updated (including the then-curator Ray Clemens), this was apparently not the most straightforward thing to do. Nor the most pressing matter on the library's hands, for that matter 


Thanks for all the work you keep putting in, eggy! I also see you found another interesting German entry in the bottom there. 

You're very welcome. Don't worry either way, I'm sure i will run out of new sources soon! But I hope that this data will be useful to someone out there, at some point, for something. If anyone finds anything wrong with any additions, please do not hesitate to PM me, email me, or post in this thread. I have done my best to ensure that no repeat entries or mistakes occur, but after seeing 100s of examples - many of which copy eachother in style and spelling - it's genuinely difficult to make sure. Once I finish adding the lists I have found, I will post my methods and sources in full. 

But now that there is a decently sized dataset, I would like to discuss octēbre. It is one of the only VMS month names that represents a semi-common variant that also has dating consequences AND is found in a statistically significant amount of examples. For a while now, the assumption has been that the month names were written (far) later than the VMS itself. If this is the case, a confirmatory piece of evidence would be frequent usage of octembre in the late 1400s or the 1500s (within a lifetime or 2 after the carbon dating of the VMS parchment). 

Using the currently available data, I listed all of the spreadsheet entries and plotted the counts of octobre vs octembre. 

To do this, I took every entry and did the following: 

1) For each entry, set the date to the middle of each given dating range (1450-1500AD = 1475AD)
2) Organise these dates into decades (1410-1419AD, 1420-1429AD)
3) Display the counts of octembre vs octobre over time

Here are the results for 20yr periods: 

[attachment=15848]

Let's imagine for a moment that we live in the world before VMS carbon dating. These results point STRONGLY to ~1360-1420AD date statistically, which is really interesting. To be clear, this DOES NOT imply that the month names were written at the same time (or before) the VMS was written, and I am not claiming that to be the case. It does imply a likely dating to the 15th century latest, but it is of course possible that some areas continued to write this way the the 16th century.

All of my current work is being done in a seperate area (a copy of the spreadsheet) before committing to the community spreadsheet, which can be viewed here for anyone interested: You are not allowed to view links. Register or Login to view. 
(forgive the absolute mess it is)

It's interesting data, even though it somewhat clashes with expectations. The impressions I've been getting from people studying the handwriting (which includes myself) is that the month names are still 15th century but later than the marginalia. The standard narrative is that someone added them afterwards, potentially to order the sheets.

But is that absolutely necessary? I think not, though it's still the most likely in my opinion.

Do the month names fit in the spaces left by the images, or do the images try to take the month names into account and work around them? The former seems most likely, given the month names' irregular placement in the circles. Which does give them the appearance of "later notes" rather than instructions for the artist or proper labels.

And if they are later notes, we might expect a bit of distance between the section's completion and their addition.

I wonder if the many octembres that weren't added to the sheet (because their other months aren't interesting) follow the same trend?

Edit to add: I would always compare octembres in absolute numbers, since the number of Octobres is basically infinite.

Very interesting, I hadn't expected that!
But I think with the broad dating ranges, such graphs can be misleading. How do you handle entries like 'early 14th century' or '15th century'?

I made an alternative graph. Again Octobre in red, Octembre in green, according to your spreadsheet. I used the absolute dating ranges. 'Early' was handled as first quarter of a century, 'Late' as last. Highest scores are at the bottom, worst on top. I omitted different spelling variants like 'October'.

[attachment=15855]

We see that Octembre in prominent 1250-1400, but also has a significant number of entries 1400-1500. But those just drown in the vast amount of Octobre entries for this time period. While Octobre is scarce pre-1400. So while Octembre loses significance at later dates percentage-wise, the absolute number isn't that much lower until 1500. The problem lies within a large number of samples of both month name variants that are just dated '15th century'.

But I think even with the abysmal date ranges we have to deal with, it's fair to say that Octembre clusters around an earlier time period that Octobre, mainly the early 15th century, and disappears post-1500. How much this is sampling bias as Koen indicates is another question. We'd have to collect all instances of those variants regardless of their score of other month names. Maybe with enough samples we could find a regional trend?

(31-05-2026, 12:31 AM)Koen G Wrote: You are not allowed to view links. Register or Login to view.I wonder if the many octembres that weren't added to the sheet (because their other months aren't interesting) follow the same trend?

(31-05-2026, 02:34 AM)Bernd Wrote: You are not allowed to view links. Register or Login to view.How much this is sampling bias as Koen indicates is another question. We'd have to collect all instances of those variants regardless of their score of other month names. Maybe with enough samples we could find a regional trend?

My criteria for adding an entry has effectively been: "are the month names french?". Once I find something and see "janvier" (or anything non Januarius), I start filling in the manuscript's details before even looking at the rest of the month names. If I see latin names, I flick through to make sure they are all latin before disregarding. There are many times that I start filling in the names and it's just another boring set. In that case it goes something like: "mars, auril, may, juing, juillet, aoust, septembre, ooo octembre, nouembre, decembre". 

If a set has octembre or octobre, it will be in the data regardless of the other months. 

(31-05-2026, 12:31 AM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Edit to add: I would always compare octembres in absolute numbers, since the number of Octobres is basically infinite.

Honestly, the amount of Octobres actually isn't that high near the end of the 14th century. Octembre is - as of the current data - far more common than Octobre in that range. There is certainly an infinite amount of the latin "October", though. Here's a graph with the absolute numbers (which are still too low honestly, so more data will be helpful)
[attachment=15857]

(31-05-2026, 02:34 AM)Bernd Wrote: You are not allowed to view links. Register or Login to view.But I think with the broad dating ranges, such graphs can be misleading. How do you handle entries like 'early 14th century' or '15th century'?

"Early 14th century" is set as 1301-1325 normally, but I think for the purposes of the octembre graph I put "Early 14thC" to 1310 for ease. This is part of the reason that there are graphs for 30yr groupings and 20yr groupings, as most of these entries will find themselves in roughly the right place. 

There are actually 2 sheets specifically to deal with the imprecise "Xth Century" dates. One sheet omits them, and the other sets them to the middle of the century (1350, 1450). The stacked data really didn't change that much with them included, but the absolute numbers obviously see a peak at 1450. At some point I also used the "latest possible date" instead, and came up with similar results, I think.

Eventually I will plot this data more thoroughly; this was mainly to see if the trend I thought I saw was actually there.

My post crossed with yours, but I removed it after seeing what you wrote: low octobre for the late 14th century is really unexpected to me. I would have thought that octobre was always the baseline in the vernacular. If it's actually octembre or a toss-up, that would be very interesting. 

I still suspect some sampling bias, since our aim was to collectively hunt for octembres.

It's possible that it was like that at first, but at this point I would confidently say that the vast majority of the entries that have been added have not been selected based on specific name variants. I would even go as far to say that their selection has been independent of specific month names in general; most of them were found via searching for "livre d'heures" or "calendrier" and then added regardless of how interesting they are.

If there was a negative selection bias against octobre, we would probably not expect such high numbers of them. In total, over 2x more Octobres have been added than Octembres, so if Octobre was common in the late 14thC, I would have expected to have found them more often. 

It's very possible that there is another bias somewhere. Perhaps calendars that contain Octembre are typically dated earlier because of the vernacular october? There's no evidence for that so far though, afaik. Most of the dating information leaves out why they are dated where they are, but many i've seen point out other reasons such as specific saints, dates, illustrations, or handwriting that point to a dating.

Edit: Another possible reason may be that many of the "15thC" entries are actually from the first quarter of the 15thC, but aren't contributing to that section. That doesn't address the lack of Octobres at the end of the 14thC though.