The Voynich Ninja

(27-08-2026, 09:47 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.

• For cooperative projects, we already work with a cutoff point at 1450, which I find generously late. Making this any later would prohibit the accumulation of knowledge.
  

I don't understand why that is generously late, considering 1450 is within the dating range for 1 of the folios tested. It's also well within one human lifetime of the dating of the parchment. Why couldn't someone who wrote the VMS in 1430 -in their 20s- also have written/drawn something similar elsewhere in 1470, in their 60s? 

I would also argue that the opposite is true, and that the cutoff explicitly prohibits the accumulation of knowledge. (there has to be a line somewhere though, obviously)

(27-08-2026, 10:16 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.I think the data strongly suggests roughly simultaneous creation of the parchment. I've overlaid all 4 images of distinct pages. I've colored with yellow the overlapping range of the actual measurement made (on the Y axis), these actual measurements look very consistent with measuring the same sample material, the overlap is huge.

Sure, but to combine those ranges into a smaller range by assuming they are the same object is not correct. If these were 4 samples from 1 folio, I would totally agree. But it's not. The individual results have FAR larger ranges, and those ranges are falsely shortened based on assumptions that aren't true.

Even if the parchment genuinely was all made around the same time - which is entirely possible, or perhaps even probable - the statistical shortening of the range has a large effect on the research conducted on the manuscript. Is it not entirely possible that one folio was made in 1390, and another was made in 1430? That would have a real effect on how the production of the VMS was viewed.

(27-08-2026, 10:43 PM)eggyk Wrote: You are not allowed to view links. Register or Login to view.Sure, but to combine those ranges into a smaller range by assuming they are the same object is not correct. If these were 4 samples from 1 folio, I would totally agree. But it's not. The individual results have FAR larger ranges, and those ranges are falsely shortened based on assumptions that aren't true.

The individual results do not have far larger ranges in the actual metric that was measured (y axis). All these samples are grouped together quite well, and I think the conclusions in the paper are the right ones - this strongly suggests a single object with a single narrow range of years. Suppose, the calibration curve was monotonous throughout this period, and we get the following results for the time period:

[attachment=17445]

Would you rather think that the object was created somewhere in the yellow range, which would be consistent with all 4 measurements, or would you say that there likely were two different time ranges hidden there?

But this above is exactly what the actual values measured in the lab show, before it's artificially split in a bimodal distribution by the double-humped calibration curve. These curves above are what we should use to assess whether it's a single object or not, and not the computed time range.

Also, consider the radical shift in the curve of f68 depending on which chemical processing was used, I've made the other 3 curves transparent in the image below, the two remaining curves are two versions of f68, the results for the very same sample, just differently processed. Basically, f68 lies on the opposite sides of the spectrum of all the samples depending on how much of the possible contamination with modern carbon was removed. Which means the error potentially produced by inconsistencies of the source material is nearly as high as the observed range of results.

[attachment=17446]

I don't think it makes sense to pay a lot of attention to this bimodal distribution of dates, it's not very likely actual 50 years passed between one folio and the other.

Eggyk, all I know is that people who know much more of radiocarbon dating than both of us combined, have determined the often cited date range. Michelle Lewis, hoping for a later date, consulted with an external expert, who confirmed that the conclusion is sound. Fashion matches very well with manuscripts produced in this date range, including the tail end in the 1430's and 1440's. Paleographical evidence of the last folio aligns well with the date range. 

Also, to the best of my knowledge, there is nothing in the manuscript that can only have been made after 1450. But of course, people will keep pushing for a later date. Everything that's earlier is already fair game, so an earlier date is not interesting.

(27-08-2026, 11:16 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.The individual results do not have far larger ranges in the actual metric that was measured (y axis). All these samples are grouped together quite well, and I think the conclusions in the paper are the right ones - this strongly suggests a single object with a single narrow range of years. 

Yes they do? The individual folios had ranges of ±35 years. The final combined range has a range of ±15 years. That's a huge difference.

(27-08-2026, 11:16 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.But this above is exactly what the actual values measured in the lab show, before it's artificially split in a bimodal distribution by the double-humped calibration curve. These curves above are what we should use to assess whether it's a single object or not, and not the computed time range.

These folios are literally not the same object. Just because they all potentially come from roughly the same set of decades does not make them so. 

Let's say hypothetically that we have another composite manuscript, with some pages from 1290AD, and some other seperate pages from 1330AD. If you were to test a few of the pages, and get these results: 

1) 1275-1335
2) 1285-1340
3) 1310-1352
4) 1320-1358

If you treat these as samples from the same object, you will get a tight range where they overlap of 1320-1335, and I think a similar range statistically. It would produce the wrong result. 
 

(27-08-2026, 11:16 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.Would you rather think that the object was created somewhere in the yellow range, which would be consistent with all 4 measurements, or would you say that there likely were two different time ranges hidden there?

If this was one folio measured 4 times, I would of course assume the yellow range. But as this is 4 seperate objects, it's entirely plausible that they are two seperate time ranges.

(27-08-2026, 07:02 PM)eggyk Wrote: You are not allowed to view links. Register or Login to view.1) The individual leaves of parchment are part of a singular "whole" object, and thus can all be assumed to be samples from that same "object"

...

- The dating range of 1404-1435AD is based on what I believe are potentially incorrect assumptions. 
- Based on the results, we should probably have a likely working date-range of ~1385-1458AD
- More carbon testing could potentially find more pages similar to f68, with a large peak in the mid 14th century
- To say that the VMS was written in a short period of time, or saying that the VMS is dated fom 1404-1435, even if true, puts unneccesary restrictions on various aspects of research.
- We should be treating works from the 9th-15th century as possible direct inspirations for the content in the VMS, instead of broadly dismissing most as probably not relevant.
- We should not presume that the content cannot have been post-1435AD

It has been a while since I last went through the original C14 report, but I don't recall it having the second assumption you described. And if it did mention that, it would certainly not be relevant, or used in getting their results. When any part of the manuscript was written, and over what period, is entirely independent of any C14 measurements and calculations.

The only original report I have ever seen is in the form of letter from Greg Hodgins to Christine McCarthy. Are you working from that, or do you have access to another version of the report? If another report exists, I would genuinely like to obtain it. I know Rene redid the analysis too but didn't he use the same assumptions and get essentially the same results. Or am I mistaken?

I had some issues of my own with how the analysis was described and reported, and I have discussed them with others here before. But those issues never altered the overall result enough to concern me.

One problem with the report, IMO, is that it does not accurately express the real assumption underlying the combination. I forget their exact words (and don't have the report to look at right now), but they said the assumption was basically that all the parchment was from the same source. That is actually not the effective assumption. The real assumption is more like this:

• The differences in the results from the four samples are mostly due to measurement error, residual contamination or other sources of uncertainty and, if that could all. be removed, then the dates for the four samples would very likely be the same... because the samples are very likely all from the same stock (or source) of parchment.[
  

In other words, the observed numerical differences are being treated as products of those various uncertainties, rather than as evidence that the parchment sheets were actually produced decades apart. The assumption of being from common stock is not the whole assumption, nor is it even the primary part of the assumption.

The primary part was glossed over—ignored, actually. (And it almost always does in the carbon dating reports.) But it is important to recognize that, even though the different samples are measured independently, the formal probability range quoted for each doesn't include every uncertainty relevant to the real historical date. It incorporates some of the uncertainties included in the calibration model, but there are also broader, largely subsumed uncertainties: possible residual contamination of the samples, systematic measurement effects, differences between the individual animals from which the different samples came, possible local atmospheric or dietary effects, and the assumptions underlying the calibration record itself, etc. etc.

The calibration curve typically used (and I believe was used in the Voynich analysis) is a curve called IntCal -- (IntCal04 for the VMS report and there are even more recent versions of it now). It is derived largely from tree rings and it has had plenty of review and scrutiny -- a LOT of carbon dating results depend on it being reliable after all. So it is actually pretty well-established and is generally reliable, but it is still just a constructed estimate of historical atmospheric carbon levels -- not a perfect global record. And the carbon record isn't uniform for all regions of the planet. Different calibration curves are used for different regions (and the appropriate region to use is yet another assumption and source of uncertainty.)

Some of the calibration-curve uncertainty was included in the VMS analysis but not all, and uncertainties around possible sample contamination and measurement errors, are not included at all. This is why using prior knowledge and beliefs about a set of related samples matters. Given that the manuscript folios share a common script, seem to have similar or related content, are all the same size, and were all bound together, it is pretty reasonable to assume they were all part of one production, put together around the same time, and using a common source of parchment. That may be completely wrong! But it is certainly more likely to be correct than not. So our prior belief is somewhat > 50%, at least. And since the four results overlapped substantially, it is reasonable to use that prior belief to resolve the (relatively minor) differences in the dating results across the samples. If they had varied wildly, then the assumption that their differences were all due to subsumed measurement errors, contamination uncertainties and such, would be difficult to defend.

While we quote carbon dating results as if the range's probability is accurate to the single percent level and the date range itself is quoted with the precision of a year, practically speaking the uncertainties probably rise to the level of one or even several decades. But those subsumed uncertainties are inherent in the whole technique and, consequently, they are most often simply ignored.

Another reasonable practice, by the way, is to discard one or two of the most extreme of the sample results as a way to artificially narrow the range in compensation for the subsumed measurement errors. But of course, that isn’t feasible when one starts with only four samples.

The important thing though is that none of this materially changes the broad historical conclusion. Whether the four results are combined or considered individually, they are in rough agreement and point predominantly toward parchment produced around the first half of the 15th century. In other words, the animals who put their skins in the game most probably died within that general period. The exact endpoints and the strength of the conclusion vary depending on the assumptions, but the overall historical placement does not change significantly. Perhaps if we find more reliable evidence about the manuscript's early provenance, the precision of the C14 data may become more relevant, but unless and until that happens, it's largely a moot point.  And it tells us nothing  about when the parchment got used to make the manuscript anyway. (I've said it before and I'll say it again -- the Vinland Map forgery was created in the 20th century, yet it was carbon dated, coincidentally, to an almost identical date range as the Voynich.)

(28-08-2026, 08:15 AM)asteckley Wrote: You are not allowed to view links. Register or Login to view.One problem with the report, IMO, is that it does not accurately express the real assumption underlying the combination. I forget their exact words (and don't have the report to look at right now), but they said the assumption was basically that all the parchment was from the same source. That is actually not the effective assumption. The real assumption is more like this:

• The differences in the results from the four samples are mostly due to measurement error, residual contamination or other sources of uncertainty and, if that could all. be removed, then the dates for the four samples would very likely be the same... because the samples are very likely all from the same stock (or source) of parchment.[
  

I'm taking from the document linked on voynich.nu: You are not allowed to view links. Register or Login to view.

Here is the quote in that document that justifies averaging the results: 

Quote:The precision of any measurement can be increased if measurements are repeated. Given the clustering of the dates for the four Soxhlet treated samples, and the nature of the object (a book), one can argue that in fact, a single object was measured 4 times.

This is pretty explicitly saying that they are treating the 4 seperate folio samples as 4 samples of 1 object, specifically for the purpose of increasing precision.

(28-08-2026, 08:15 AM)asteckley Wrote: You are not allowed to view links. Register or Login to view.Given that the manuscript folios share a common script, seem to have similar or related content, are all the same size, and were all bound together, it is pretty reasonable to assume they were all part of one production, put together around the same time, and using a common source of parchment. That may be completely wrong! But it is certainly more likely to be correct than not. So our prior belief is somewhat > 50%, at least. And since the four results overlapped substantially, it is reasonable to use that prior belief to resolve the (relatively minor) differences in the dating results across the samples. If they had varied wildly, then the assumption that their differences were all due to subsumed measurement errors, contamination uncertainties and such, would be difficult to defend.

As I stated before, the logic of book -> single object is not necessarily true. Many composite manuscripts containing similar content with different dates exist. Other manuscripts contain various works written at different times on the same parchment.

I was also under the impression that the VMS is assumed to have been bound later, meaning the folios were at one point seperate and unbound. It seems interesting to me that the justification behind this calculation was that because they were bound together in a book, the seperate folios can be considered one object. While at the same time, other reseach seperately suspects that the sheets of parchment were originally unbound and seperate. 

And you are effectively confirming what I said in the OP, which is that this assumption fundamentally comes down to a common script shared across the pages. This presumes that we know the context of voynichese - how many people could write it, understand it, and for how long a period of time. Why is it more likely that it was one production, instead of a student continuing the work of his teacher? 

I agree with much of the rest of what you are saying, too. As I mentioned, I am okay with much of the uncertainty and inherent variability that comes from carbon dating. My issue is basically with the final averaged range that was calculated. And yes indeed, it doesn't mean that the VMS date is wildly wrong either; it is almost certainly still a medieval object from broadly the same time period. But a range of, say, 1380-1450 has quite different implications for research than 1404-1435. I am of the opinion that we should seriously consider the wider range for purposes of research, to prevent missing crucial leads.

It's much less separate objects than you'd think. You can get several bifolios out of the same skin. And as far as I'm aware, nothing in the VM shows that some skins are of a different supply. They are all of similar size and quality.

(27-08-2026, 07:02 PM)eggyk Wrote: You are not allowed to view links. Register or Login to view.- More carbon testing could potentially find more pages similar to f68, with a large peak in the mid 14th century

This is pure speculation. Since we have hard data, I think sticking to them is the best possible approach. More carbon testing could potentially find that there are pages that are more than 60,000 years old, therefore time-travel or alien technology is involved.

I'm under the impression that the samples measured were not assumed to be from the same calf skin. The justification used was that they constituted a single object because they were bound within the same book. 

One sample was specifcally chosen due to a different thickness, 2 under the assumption that currier A->B may consitute a jump in time (along with one being a on the same bifolium as folio 1), and one because it was from an interesting astronomical page. This leaves out half of the entire VMS.

Note that two folios of similar size and quality were specifically chosen due to perceived difference in the content on them. That implies that a difference in content can represent a different time of writing, and that noticeable differences in the parchment are not necessary for dating differences to be investigated. I'm pretty sure the sample selection leaves out quire 13 entirely, right? If what is now known about the entropy of that section was known at the time of these tests, it would likely have been chosen as a sample specifically to test whether it was made at a different time. Why should we now assume that the results would have been in keeping with the rest of the samples?

There is zero direct evidence that those other sections date to the exact same time as the tested samples, other than the parchment looks roughly the same and the script looks roughly the same. We are effectively saying "this half of the book definitely dates to the same time as the other half of the book, because they are bound in the same binding and don't look all that different". 

It is entirely possible that the recipes section dates completely differently. There is no evidence for that being the case, but there is also no evidence against this being the case, in my opinion. The only evidence is that we are assuming it was probably written and added at the same time as the measured samples. That's literally it.

(28-08-2026, 11:06 AM)MarcoP Wrote: You are not allowed to view links. Register or Login to view.This is pure speculation. Since we have hard data, I think sticking to them is the best possible approach. More carbon testing could potentially find that there are pages that are more than 60,000 years old, therefore time-travel or alien technology is involved.

Yes, that point is pure speculation.

But why are you characterising what i've said in that way? I haven't claimed once that the results for the samples are incorrect. How is saying "we might find more samples like one of the samples we used" in any way similar to potential time travel and alien influence? 

Technically, the combining of the samples makes more assumptions about the rest of the manuscript folios than what I am saying. I am explicitly saying we should stick to the hard data instead of interpreting it in the way it has been interpreted.