The Voynich Ninja

Full Version: Assumptions baked into the VMS carbon dating
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3 4 5 6 7
(28-08-2026, 05:18 PM)eggyk Wrote: You are not allowed to view links. Register or Login to view.You are also saying I seem to have some prior belief about the nature of the parchment, when I literally don't. 

I'm sorry, but this is a good example of your not understanding what has been explained throughout several comments in this thread. You see, I neither said nor implied any like this. 

You misunderstood my references to "prior beliefs" and my descriptions of their role. I usually assume that such confusion is my fault -- due to me not expressing things clearly enough -- but I don't see that here. Which is why I just don't think further explanation is efficient for either of us.
(28-08-2026, 05:35 PM)asteckley Wrote: You are not allowed to view links. Register or Login to view.
(28-08-2026, 05:18 PM)eggyk Wrote: You are not allowed to view links. Register or Login to view.You are also saying I seem to have some prior belief about the nature of the parchment, when I literally don't. 

I'm sorry, but this is a good example of your not understanding what has been explained throughout several comments in this thread. You see, I neither said nor implied any like this. 

You misunderstood my references to "prior beliefs" and my descriptions of their role. I usually assume that such confusion is my fault -- due to me not expressing things clearly enough -- but I don't see that here. Which is why I just don't think further explanation is efficient for either of us.

Well, okay, this is kinda embarrassing for me, because I genuinely, honestly misread what you typed and it looks pretty damning given what we've just been discussing. Sorry. I'm not aiming to be obtuse and dismissive.  Sad

I would still appreciate if you are willing to explain some things that I have not grasped, because I can't come to the right conclusion if I don't know what exactly i'm not understanding. I would like to know why the fact that other books can have different sources of parchment does not hold the weight I think it does, for example.
In order to have a book with sources that can be differentiated by C-14 dating, it would have to contain material from different historical time periods bound together. Examples do exist, but most texts were produced in a span of time too brief for C-14 testing to discern.

The primary assumption at issue is the "single source" hypothesis. Given the test data and the range of uncertainty, the dating overlaps. The four samples come from a single text. It is statistically possible to calculate a combined average date-range. [[ 1404 - 1438]   > Not 1435!! 

That is with 95% probability. What about 96%? 90% is pretty good.

Nothing contradicts the "single source" hypothesis - other than the fact that it is not the only hypothesis. The first choice here is the "two source" hypothesis. Less likely, perhaps. but still viable. A separation of a few decades is all the same to C-14, but not in human terms. And that is just the parchment date. Text creation was 'somewhat' later.
(28-08-2026, 05:58 PM)eggyk Wrote: You are not allowed to view links. Register or Login to view.I would still appreciate if you are willing to explain some things that I have not grasped, because I can't come to the right conclusion if I don't know what exactly i'm not understanding. I would like to know why the fact that other books can have different sources of parchment does not hold the weight I think it does, for example.

Well, I'm not trying t put you off, and if I find some time to try to explain my points better than I did,  I will. But I've just don't have the bandwidth right now.  

But on your last point: there are indeed other books with different sources of parchment-- sure --but that is not the most common case one will find. I had mentioned, I think, about four features of this book that would suggest it was probably created in a single production effort (i.e. using the same stock of parchment). That suggestion follows from those observed features because the majority of books (I am sure, by far) that have those particular features also turn out to be the result of a single book production effort. So the fact that you can find some instances of books with disparate paper sources doesn't mean that it isn't still far more likely than not that when you find those features, you are looking at a book whose pages all had a common source. (In other words, your observation of selected exceptions holds little weight.) And -- given the amount of overlap of C14 dating results from four individual samples, which itself suggests there is a physical relationship between them and that the differences are probably due to subsumed measurement errors alone-- it becomes reasonable to treat the samples as a "single source"; that choice of treatment dictates a particular calculation process to reduce measurement errors.

(By the way, you pointed out that being bound together was not a good indicator of a common source, especially since it is believed that the manuscript might have been bound together only after it had been long completed as separate projects. That might be true, but it would also be a demonstration that someone in the past once noticed the rest of the features and also drew the same conclusion -- that the pages were most likely all part of a single book and so they ought to be bound together. And so they bound them.)
(28-08-2026, 09:01 AM)eggyk Wrote: You are not allowed to view links. Register or Login to view.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.
I agree.  

Averaging the results (that is, treating all samples as samples of the same object) is appropriate if one assumes that all the bifolios were produced within a short period of a couple of years. 

But the C14 data does not tell us that.  We know that the vellum comes from at least 10 different animals.  The Author possibly, if not probably, procured vellum from merchants or other sources (like the rejects box of monasteries and other vellum users) that could have stocked and mixed skins of arbitrary ages.  Therefore the assumption above cannot be made.  The proper way to present the results is by listing the individual probability distributions, not by averaging the results. 

That is, the C14 measurements only say that 
  • the vellum of bifolio f1+f8 (two panels) was probably made in 1394-1458 (92.2%) or maybe 1327-1342 (3%) 
  • the vellum of bifolio  f67+f68 (5.5 panels) was probability made in 1385-1436 (52.9%) or maybe 1308-1362 (42.5%)
and similarly for the other two bifolios that were tested.  From those measurements we cannot say anything about the bifolios that were not tested.

If we want a result for the book as a whole, the only thing we can say is that at least some parts of the Scottish cow VMS (such as f1+f8) were written after ~1420, with maybe ~70% probability.  Or something like that.  

But there is a non-negligible probability that some of the tested bifolios could have been written before 1400.  And, even if we assume that the whole VMS was produced with the lifetime of a single author,  some of the bifolios that were not tested could have been written in 1340 or earlier.

As it has been pointed out many times, the C15 report results do not give any upper bound to the date of writing.  Not even if we assume that the whole book was produced by the same Author and there was no forgery.  The Author could have found a pile of vellum in the attic, left by his great-great-grandfather, and could have given mostly 15th-century books to the Scribe to copy the figures from.

Moreover, the washing with solvent may have removed some original material together with later contamination.  Like the resin which was normally rubbed on the vellum to make the surface slightly hydrophobic.  Thus the measurement on the unwashed f67+f68 sample could be more correct than the one after washing.

All the best, --stolfi
(28-08-2026, 09:32 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.Averaging the results (that is, treating all samples as samples of the same object) is appropriate if one assumes that all the bifolios were produced within a short period of a couple of years. 

But the C14 data does not tell us that.  We know that the vellum comes from at least 10 different animals.  The Author possibly, if not probably, procured vellum from merchants or other sources (like the rejects box of monasteries and other vellum users) that could have stocked and mixed skins of arbitrary ages.  Therefore the assumption above cannot be made.  The proper way to present the results is by listing the individual probability distributions, not by averaging the results.

I still see no problem with the conclusions in the radiocarbon dating report. The most likely range for the creation of the set of folios that have been tested is ~1404-1435. Other possibilities are considerably less likely. If the parchment was produced over the course of 50+ years, it appears a very strange coincidence that all tested folios cluster in the range that maps both to the peak of ~1340 and to the upslope of ~1420 and none of them appear in the trove of ~1375 where they would immediately stand out and it would be a certainty that the parchment was not all from the same period. So, while it's possible that untested folios were created earlier or later, the four tested folios are most likely from the range specified in the report.

Overall, I'm not sure why this is so important. As you said yourself, any later date is certainly possible, especially given the poor quality of the vellum I can imagine someone sourcing old scraps unused for 50+ years, on the other hand if the manuscript is a copy, the original could be from any date preceding 1300s. Maybe the very things that the visual analysis is based on, like dress styles, were modified when copying.
(28-08-2026, 10:07 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.The most likely range for the creation of the set of folios that have been tested is ~1404-1435. Other possibilities are considerably less likely. If the parchment was produced over the course of 50+ years, it appears a very strange coincidence that all tested folios cluster in the range that maps both to the peak of ~1340 and to the upslope of ~1420 and none of them appear in the trove of ~1375 where they would immediately stand out and it would be a certainty that the parchment was not all from the same period.

Let's assume that each section of the VMS was written on vellum made from animals that were slaughtered over a span of a few years.  (This already is not a trivial assumption, given the many ways that the vellum could have been procured.  But let's adopt it as working hypothesis.)

With that assumption and the C14  results, we then could say that 
  • the Herbal bifolios (3 sampled) were almost certainly (more than 80%) made sometime in 1390-1460. 
  • the Astro/Cosmo bifolios (1 sampled) could have been made in 1308-1362 (~40%) or 1385-1436 (~50%).
And we could not say anything about any of the other sections. 

On the other hand, suppose instead that a random subset of 5 of the 50(?) bifolios were made in ~1500, while the other 45 were all made in ~1425, which is compatible with the measurements.  The probability that a sample of four bifolios would miss all those 5 anomalous ones is (1 - 5/50)^4 = ~65%.   But even a single bifolio from ~1500 would push the probable date of the writing to post-1500...

All the best, --stolfi
(27-08-2026, 09:47 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.The fashion of the zodiac figures knows its peak well before 1430. Later examples exist, but those are in the tail end. Even if the VM is in that tail end, it's still within the RC frame.

I agree that early 1400s is still the most likely date and can be assumed as a working hypothesis. 

However, if the Zodiac figures were copied from some other manuscript (like f85r2 quite probably was, for example), the dress styles would only tell the date of the source manuscript.

Quote:Both Marco and I, after analysing hundreds of handwriting samples to the point of complete burn-out, felt like the You are not allowed to view links. Register or Login to view. handwriting is purely medieval.

You know my explanation for f116v.  Even if that accident and "restoration" took place in 1580 or 1880, the person who tried to restore the lost writing surely would have tried to use letter shapes from the presumed date of the manuscript, not contemporaneous letter shapes.  (But if he was in Germany he would probably have tried to use German letter shapes.  So that analysis is still important to establish the nationality of the restorer, if not the date.)

And while you have found samples from the early 1400s that match many of the letters, you have not found a single sample whose handwriting matches the whole You are not allowed to view links. Register or Login to view. text -- have you?

All the best, --stolfi
(29-08-2026, 03:16 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.With that assumption and the C14  results, we then could say that 
  • the Herbal bifolios (3 sampled) were almost certainly (more than 80%) made sometime in 1390-1460. 
  • the Astro/Cosmo bifolios (1 sampled) could have been made in 1308-1362 (~40%) or 1385-1436 (~50%).

These are all probabilities for individual folios. Given that the folios nowadays belong and for a long time have belonged to the same object and share the script, the image styles, and are of compatible sizes, one cannot ignore the probability of 4 random samples from this object producing a coherent largely overlapping result in the carbon isotope measurement. Your logic would totally apply if we were identifying when 4 totally distinct objects were produced. The questions one needs to answer are of the kind "given these ranges of three of four similar looking folios, how should we adjust the probability distribution of the creation date of the fourth?", etc. For me, all these probabilities give the overall very high likelihood of the range provided in the report.

Unless there is some real evidence for a multidecade (40+ years) production, I see no real reason to go beyond 1404-1438 range provided in the report, it looks conservative enough for me.
(28-08-2026, 08:09 PM)asteckley Wrote: You are not allowed to view links. Register or Login to view.Well, I'm not trying t put you off, and if I find some time to try to explain my points better than I did,  I will. But I've just don't have the bandwidth right now.  

No I understand, and I appreciate it. 

(28-08-2026, 08:09 PM)asteckley Wrote: You are not allowed to view links. Register or Login to view.But on your last point: there are indeed other books with different sources of parchment-- sure --but that is not the most common case one will find. I had mentioned, I think, about four features of this book that would suggest it was probably created in a single production effort (i.e. using the same stock of parchment). That suggestion follows from those observed features because the majority of books (I am sure, by far) that have those particular features also turn out to be the result of a single book production effort. So the fact that you can find some instances of books with disparate paper sources doesn't mean that it isn't still far more likely than not that when you find those features, you are looking at a book whose pages all had a common source. (In other words, your observation of selected exceptions holds little weight.) And -- given the amount of overlap of C14 dating results from four individual samples, which itself suggests there is a physical relationship between them and that the differences are probably due to subsumed measurement errors alone-- it becomes reasonable to treat the samples as a "single source"; that choice of treatment dictates a particular calculation process to reduce measurement errors.

(By the way, you pointed out that being bound together was not a good indicator of a common source, especially since it is believed that the manuscript might have been bound together only after it had been long completed as separate projects. That might be true, but it would also be a demonstration that someone in the past once noticed the rest of the features and also drew the same conclusion -- that the pages were most likely all part of a single book and so they ought to be bound together. And so they bound them.)

So, I have had a different feeling about how common this is, I think predominantly through adding sources to the spreadsheets. It is more common than not that everything has one date, yes, and it is more common that these manuscripts appear to be a single production process. 

However, I have lost count of just how many times i've had to dive into the descriptions and official reports to work out which sections date to when. That situation is less common, yes, but it happens quite frequently and can be rather frustrating. Now, I don't have a number to give you as to how many of those manuscripts sourced different parchment for each differently dated section, but it is definitely more than a few (obviously, the ones i'm thinking of were plain and obvious - the non-obvious ones will likely not have been carbon dated). And these are sometimes dating differences of up to 100 years or more, not always 5 or 10 years or something. I assume they got bound together due to their shared topic in most cases. 

So, is it more likely that it was a single production? Yes perhaps, I would say so. But the fact that finding the alternative is common enough that it is not even really notable anymore would suggest that the odds are more even than is being assumed. Of course, "More likely than not" encompasses many sets of varying odds, everywhere from 99%-1% all the way to 51%-49%. So how reasonable an assumption this is is effectively a value judgment based on that % split. 99%-1% is probably a very good bet, while 70%-30% would arguably be too risky to rely on completely. 

(It is worth saying that in my experience, such examples of various different dated sections have been far more common in southern germany than elsewhere, but I don't have hard data to back that up right now.)

Now, it is also worth mentioning that part of this is not helped by the choices of samples. IF the VMS was such a case with differently dated topical sections - however likely or unlikely that is - the most logical differences in date would be between the sections. If older and newer parchment were seperately added, it would be less likely to find that in the middle of a section. So if the 4 samples had included 1 sample from 4 sections and still overlapped, I would personally hold that with higher weight than the current results. 

Unfortunately, multiple large different sections were not tested. If they were to be, its quite possible they would overlap much the same, but it is also entirely possible that it would differ.  (reasonably possible, not with the same chance as being made 60,000 years ago or made by aliens as put to me earlier in this thread).
Pages: 1 2 3 4 5 6 7