In Part I of this series I discussed how a Xerox WorkCentre emerged as the likely source for the PDF file scan of the President’s long form birth certificate that was released to the public via a press conference on April 27, 2011, and also posted the same day on the White House web site.
Blogger NBC began to focus on the Xerox as the “forger” with a series of fine articles in June of this year. NBC found a number of documents posted online that were created on a Xerox WorkCentre 7655. These documents exhibited many of the the same features of MRC compression seen in the WH LFBC PDF. He also found an odd comment string embedded in the JPG portion of the PDF’s, “YCbCr”. As I have mentioned before this string refers to the color space used for encoding a JPG in such a way that the file can be greatly reduced in size without changing the way it is perceived by the human eye. The human eye is more sensitive to shades of grey than variations in color. The comment string was found in every color scan known to be produced on a Xerox WorkCentre. NBC found it was also in the WH LFNC PDF.
Now it was time to test the theory. Since no one has access to the original document the simplest place to start was by printing a color copy of the WH PFBC PDF and then scanning it on a Xerox WorkCentre 7655. NBC did exactly that and the results were nothing short of astounding. Nearly all the anomalies were there including
- The green background 8 bit JPG layer including the form
- Multiple monochrome 1 bit layers
- Separation of the seal and the date stamp into separate layers that can be moved around in Adobe Illustrator
- A Xerox WorkCentre, through edge erase, creates a masking layer, just like the one found on President Obama’s PDF A WorkCentre creates using MRC and JBIG2 compression an optimized document which recreates at least the following artifacts:
- -Multiple monochrome bitmap layers
- -A JPEG layer at half the resolution of the text layer
- -Color under the text layer due to JPEG artifacts
- -Rotated images
- -A JPEG with embedded comment ‘YCbCr’
- -JBIG2 compression causing characters to be identical
- When saved as a PDF in preview, after a 180 degree rotation, the relevant metadata identifying the WorkCentre disappears but various tell-tale signs remain
- I followed up by performing a similar tests on a Xerox WorkCentre 7535 using a color printout of the WH PFBC. My results confirmed what NBC had found and the only significant difference was the number of layers and the details exactly what items were separated into the background and primary text layers. Something I did, which the CCP had never done, was to publish the resultant files themselves in this article: “
- The WorkCentre 7535 appears to use the same MRC compression algorithm as several other machines in the WorkCentre line so one would expect similar results. As a matter of fact when Dave Kriesel, a graduate student in Germany posted an
- about a problem he had discovered with the MRC/JBIG2 algorithm changing numbers on scans in some rare cases both the 76XX and the 75XX series were listed as affected models.
- We had now demonstrated that the laundry list of anomalies cited by a coterie of Birther self-proclaimed experts could be explained by the Xerox WorkCentre –> Mac Preview workflow process. One feature of the LFBC PDF that could not be tested using printout of the LFBC PDF was the white halos found around some of the letters. The explanation is simple. The halos are were already there so we would have to demonstrate that the Xerox created them with another test specimen.
Some commenters on anti-Birther blogs had located examples of Xerox WorkCentre sourced documents that demonstrated halos. Fortunately, Xerox did us a favor by making the default file name for PDF scans to email “Scanned from a Xerox multifunction device001”. A Google search for that phrase will result in hits for PDF’s posted online with that file name.
Here are examples posted by commenter JPotter at OCT of two files demonstrating the creation of halos:
http://piledrivers2404.ca/wp-content/uploads/2012/04/Scanned-from-a-Xerox-multifunction-device001.pdf and http://atmuseum.org/1983_exhibit_pdfs/Scanned%20from%20a%20Xerox%20multifunction%20device001.pdf
The former is a scan of a Pile Drivers Union news letter. It displays clear examples of halo effects in the header:
The second document is a scan of some press clippings posted at the Appalachian Trail Museum web site. The halos are more subtle on this background but are still present.
JPotter explained the halo effect very well:
All it takes is sharp lines/text on a non-white background*, and a certain level of contrast between the lines/text and the background. When scanned on a Workcentre on which MRC is enabled, *poof*, “halos”.
* Of course, any time a foreground layer is created, the “halos” are there regardless of the background color or pattern … it’s just difficult for mere mortals to distinguish white on white.
JPotter makes an important point that the halos are really there on other documents but you just cannot see them on a white background. The ideal conditions for detecting halos is in a document is one with black lettering on a light colored background. Like a document with black letters on a green basket weave pattern for example. They are the natural result of the MRC compression process.
Preparing the Test Specimen
The next step was to create as best we could a replica of the LFBC. We had to use what was available. I suppose I could have called the White House and asked to borrow one of the two certified copies. I decided that would not really be a good idea.
We decided the next best thing was to obtain some sheets of green security paper and print the best copy we could find of the LFBC on one of these. Kevin Davidson (Doctor Conspiracy) stepped up to the plate and obtained a couple of different types of green basket weave security paper. The first type was not a good match but the second one was a very good match. Here is what it looked like:
Now we had to pick the best image to print on a sheet of this security paper. The image with the best resolution of the text on the LFBC was a photo taken by AP photographer Scott Applewhite at the press conference where the LFBC was first released. Applewhite took photos of all four pages of the handout given to members of the media that morning. The Applewhite photo appears to be of the copy of the LFBC in this handout. We can infer that the image was from that handout because we can see a faint copy of the President’s short form bleeding through the image. It was apparently behind the LFBC when Applewhite shot the photo.
One problem with the AP photo is that the background had a slight bluish tint. Kevin Davidson was able to produce a white background version of the AP photo:
This image is the one I used to prepare the test specimen. One additional step was required however. We know that the image of the LFBC was reduced when Hawaii copied it on to the green security paper. I believe this was done to make sure that the copy showed the entire document to the edge and still left a green border around the image.
I was able to determine through some measurements and trial and error testing that the right scale was about a 74% reduction. I printed a scaled version using a high quality color laser printer (coincidentally not a Xerox WorkCentre) on a sheet of the green security paper. I believe the result was a good approximation of that the original LFBC certified copy must have looked like.
Here is a link to a high resolution version: LFBC Test Specimen
Ready to Test!
As you can see the test specimen had no white borders and no halos. Now I was ready for some testing. What did I expect to demonstrate?
- That I could take an “original” document that was similar in appearance to the WH LFBC and produce a PDF on a Xerox WorkCentre with features similar to the LFBC PDF.
- The number of layers would be similar.
- The separation of the document into layers would be similar.
- That some characters in the monochrome mask layers would be identical pixel for pixel.
- Layers like the seal and date stamp could be moved in Illustrator.
- A white border would be produced when using default settings on the Xerox WorkCentre.
- The additional step of resaving the file in Preview would produce a top level clipping mask and even larger white border.
- That halos would be produced where they did not exist in the original.
- That the PDF would contain a JPG layer with the “YCbCr” comment tag.
The CCP claimed in their Media Supplemental Report written by Garrett Papit that they had conducted similar tests using a copy of the LFBC printed on green security paper. Of course they claimed they were not able to duplicate any of the anomalies in the LFBC and concluded it was a forgery. Their testing was fundamentally flawed. In my next article I will explain why and look at the results of the tests I ran on a Xerox WorkCentre.
Buckle up and stay tuned.