The Voynich Ninja

Full Version: Help? Idea for how to test a hypothesis?
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2
Okay, so, my own favorite idea for the Voynich at this point is that some/most of the writing is nulls or other junk writing just put in to confuse things. I'm inclining to think that there was some kind of grille or lattice that one put over the page to obscure the unimportant writing, and that this device was at some time lost or destroyed, leaving us with the mystery we now have.

Reason for this is, I notice a particular diagonal line formed by the text on the left side of almost every page, like letters have been forced to fit into the space in a certain way, and I also see evidence of more such lines on nearly every page. I see also a strong tendency for certain shapes to repeat within some of these diagonal gatherings. Sometimes the diagonal lines might be partially left blank but the page regions that have writing still make the diagonal shapes. 

The only idea I have for how to test this, would be to line up printouts of the pages over a light table, and compare the text placements on top of each other -- I would expect there to always be diagonal lines in all the same places if this were correct. The trouble is I cannot actually test it like that, because the MS was not flat-scanned -- the pictures we have are just photos of the open book, so the pages don't lay flat, and as a result they cannot be accurately compared due to arching, bending, camera angles and other distortions. (I mean, I can't even get the front and back edges of the same folio to line up accurately most of the time because of the way it was photographed -- and I know from having tried it, that if you can't get it to line up accurately it's hopeless to guess where the text truly does or doesn't overlap.) 

Transcripts are useless for this because the actual number of letters wouldn't stay the same, just the page-placement (and it might even be that only parts and pieces of the individual letters are important.)

I've been trying various simple methods like looking at pages through cheese graters and laying sticks or strips of paper over them, but there are hundreds or maybe even thousands of possible combinations there (and of course I don't even know what language I'm looking for underneath.) I don't think the idea's been disproven yet, it's just I don't have a proper way to check it. 

So, I am asking -- does anyone have any idea how I could perhaps check this, with available resources? Some way to see what parts of the pages are favored for writing? 

If you can't understand what I'm talking about, feel free to ask more questions. I've been ridiculed elsewhere for asking for help with this, by people who can't seem to understand what I'm getting at.
In order to test a grille, you will have to make a device (cut us piece of carton) and put it over the text yourself.
Myself i ordered a printed version of the manuscript, in order to check such a thing.
You can also use it on your screen if you do not have a paper version of course.

There is no automated possibility to do this by my knowledge, because you would need an OCR possibility and need to connect to the OCR software to test your algorithm. Such software is bit valuable and not easily available.

What you could do also, is place pages on top of each other in a graphical environment in (transparent) layers, and see if there is something lined up.  I believe there was someone on youtube doing the same with the manuscript, i believe, and presented it as a "solution", although there wasn't any.

Problem with any method is this:  basis of the test are the photographs of the manuscript. These are NOT 1:1 scans, but photo's. This means one page's text is in another angle than the other and in another scale. You can not be sure if the size of the page is actual size.
(05-09-2016, 10:05 AM)Botis Wrote: You are not allowed to view links. Register or Login to view.[intro deleted for brevity]

So, I am asking -- does anyone have any idea how I could perhaps check this, with available resources? Some way to see what parts of the pages are favored for writing? 

If you can't understand what I'm talking about, feel free to ask more questions. I've been ridiculed elsewhere for asking for help with this, by people who can't seem to understand what I'm getting at.

As you've pointed out, the pages probably aren't scanned at the same flatness (and may not be scanned at the same scale) which means laying them on top of each other is a challenge, but one way to do it would be this:

•  As David has pointed out, you can load each page into a graphics program on a different layer.

•  Take the uppermost layer and set it to about 60% transparency. Do this successively in pairs or, set each to diminishing levels as the layers go up (40%, 60%, 80%, etc.)


By the way... there are diagonal spaces throughout the VMS. On some pages they are very obvious and where they are, it looks like someone may have added in letters to fill in some of the spaces. Whether this is due to the text being created in more than one pass or perhaps has a steganographic purpose, I don't know, but the weirdness is definitely there. You can see it on some of the mid and later plant pages especially.
There is special software aimed at "normalizing" page scans (quite useful for scans of paperbacks). Don't remember its name offhand, but should be found on Google.

Could be tried on the VMS.
(05-09-2016, 03:48 PM)Anton Wrote: You are not allowed to view links. Register or Login to view.There is special software aimed at "normalizing" page scans (quite useful for scans of paperbacks). Don't remember its name offhand, but should be found on Google.

Could be tried on the VMS.

Far as I can tell those normalizers just straighten the text and maybe make the pages a consistent size, but don't do anything about the big issue of the images being derived from not-flat paper. Thing is, if the image of the Voynich page is distorted, I could only get an accurate comparison against another page that was distorted in the exact same way. 

That said I did manage to pick out two pages that looked similarly distorted and compare them. The results are very interesting in how words line up.
Quote:Far as I can tell those normalizers just straighten the text and maybe make the pages a consistent size, but don't do anything about the big issue of the images being derived from not-flat paper.

Those two issues are the very issues resulting from the images being derived from non-flat paper, aren't they?

Anyway, I have no experience with such tools, so can give no further comments. Sad
(08-09-2016, 02:29 PM)Anton Wrote: You are not allowed to view links. Register or Login to view.
Quote:Far as I can tell those normalizers just straighten the text and maybe make the pages a consistent size, but don't do anything about the big issue of the images being derived from not-flat paper.

Those two issues are the very issues resulting from the images being derived from non-flat paper, aren't they?

Anyway, I have no experience with such tools, so can give no further comments. Sad

No. The big issue is with the paper being arched in various places so that it's not flat -- think three dimensions and forced perspective. The text, on a single page, curves and shrinks/grows because it's nearer or closer to the camera depending on the arch. The programs fix only tilted text or page-to-page size problems, not curved/warped text and size problems on a single page.

For example, You are not allowed to view links. Register or Login to view., making any writing in that area useless for comparison to a page You are not allowed to view links. Register or Login to view. where the right hand side is more flat -- through an overlay method, anyway, they're useless.
One of the features of the 2014 scans, compared to the 2004 scans, is that more effort was made to make them flat.
This was in view of the fact that they were already planning to make the facsimile / book of essays that will appear in 2 months, and the news scans are the basis for this book.

The 2004 scans are at Jason Davies' site, and the 2014 scans at the Beinecke digital library.
Quote:The programs fix only tilted text or page-to-page size problems, not curved/warped text and size problems on a single page.

Actually I had an impression that the first thing they do is they fix the curved text. But, as I said above, I've no experience and may be mistaken. Nevermind. Smile
How thoughtless and short-sighted of the VMs author to create a separate grille that would be so easily lost. Instead, why not include a template from which a grille can be made and disguise that template in one of the illustrations?
Pages: 1 2