Search for Hidden Numbers in the VM

SEARCH FOR HIDDEN NUMBERS IN THE VM
Jan. B. Hurych

(originally published 2 August 2004)

Following are the efforts made by me to recover the digits hidden below the blue paint on f102v2 ( Beinecke 1006252) using the area Jorge Stolfi suggested there may be some numbers there.
Comparing with the other methods, the deconvolution (kindly provided by VM-list conference, I do not recall the programmer) seems to remove also the areas where blue is over the brown, where it should leave the brown instead. Now we know that the picture consist of pixels and the filter apparently leaves only true brown pixels. What happens to pixels with the colors that are not true? So far it looks like the method reduces the brown text into a series of unrecognizable spots which does not help us too much. The reality, on the other hand, is that the overpaint by brush does not go by pixels but layers and some pixels are so well masked by blue that there is no trace of brown color in them. So I did not concentrate on one method only, but did comparisons with others.

I based my decision on the possibility that first six brown characters under the blue overpaint are all recognizable as certain numbers. In the case of one, two or even three, I would admit the coincidence, but if there is more, the case is worth studying. Unfortunately, the "masker" did quite a good job. On the picture, we see the difference between coloring (marked "Y") , which is clearly avoiding the decorative circles and masking ( marked "X") where we can clearly see the vertical overpaint slashes covering almost exactly the symbols and moreover only the height of symbols, leaving quite a large area uncovered. Now this was not coloring, it was definitely the masking.

As for the size of numbers: they are almost of the same height as some VM "letters" but for proper investigation, the size of the sample of course has to be large enough (Beinecke scans allow magnification up to the size I used, but not larger, due to additional distortion). The bumps in the vellum, lumps in the ink, and artefacts of the image compression of course cannot be discounted, but it is obvious they will not effect the result that much. They would be random and certainly would not make up for the shapes so close to certain numbers.

True, there are strings or circles or dashes somewhere else on the folio, but considering the spots in question as a decoration only is just jumping to conclusion before proper investigation is done. As for the claim "there is no reason for numbers to be there" it is putting the wagon before the horse - that's exactly what steganography counts on.

I realize that if we consider those are really the numbers, we would have to make some unwelcome conclusions like the one that it was surely "hidden" there intentionally (by locating it in the place nobody would look for it and besides, their original size it much smaller then shown here). And like it was not enough, they were additionally masked by overpaint The author's knowledge of Arabic numbers was of course expected long time ago (there are several numbers in the VM already) and he apparently knew about steganography as well. And what is more important: we would have to admit that the VM is something more than just what meets the eye, but that we have known all along :-).

EVALUATION

1) FIRST BY INTENSITY CHANGE:

Obviously, it was not enough and same poor results stemmed from contrast variations.

2) NEXT BY "COLOR SEPARATOR", KINDLY PROVIDED IN THE VM-LIST

The next step would be color filtering or separation. I guess with this filtering, I removed too much since I did not know how to optimize the composite colors of brown and blue. However, the results provider by other people seemed to have the same drawback.

3) THE TRIAL TO REMOVE THE BLUE COLOR BY SOME OTHER MEANS:

It didn't work too well either, apparently the "blue" is not real, pure blue of computer basic color.

4) THE WORK ON NEGATIVES, WITH VARIOUS DEGREES OF SUCCESS:

5) FINALLY THE TRIAL No.3. WITH THE REMOVAL OF THE BACKGROUND COLOR:

6) FOR COMPARISON, I ALSO THE USED OF PROGRAM GEOMATICA, WITH HIGH DISCRIMINATION FILTERS.

They all confirmed that there is something underneath the blue mask, looking suspiciously similar to Arab numerals.

This is as far as we can go to convince somebody that there surely are some numbers there. The other problem is of course to decide what those numbers are and to provide an educated guess what do they represent. After all, this is still the beginning . . .