QandO

September 09, 2004

Document Forgeries? My Take.
Posted by Dale Franks

Rather than cover the ground that McQ and Jon have covered, I've decided to offer some observations about the forgery allegations over the CBS Documents. I served for nearly ten years on active duty in the USAF, two of them in a staff position, so I'm extremely familiar with the the USAF's Tongue and Quill (WARNING: Very large PDF file) style guide, also known as AFH 33-337. Additionally, I have spent the last several years working as an author of computer training manuals, technical training instructor, and software and web developer.

I point out my background to note that I have some relevant expertise, and that, based on that expertise, these memos appear odd.

It's not particularly relevant, but I was also in the Civil Air Patrol unit at Ellington AFB. I was born in Houston, and grew up in that part of town. When I was in the CAP there in the late 70s, the 111th was still flying Voodoos.

First, the form of the memos are odd. The signature at the bottom of the memos is improper. The memos are signed as follows:

This is not a proper Air Force signature block. Indeed, via Powerline, we have an example of Lt. Col. Killian's signature in the proper form:

I find it difficult to believe that Lt Col Killian would sign an official document without using the prescribed USAF signature block format. It just isn't done.

Next, some of the documents have manually typed letterheads. this isn't done either. Since 1948, the Air Force has had an official letterhead which is required for use by every USAF unit in the world:

It consists of the DOD Seal at left, then the words DEPARTMENT OF THE AIR FORCE, under which is the unit name, and, optionally, the unit's full address. If you run out of unit letterhead, you are allowed to use the blank letterhead, which merely has the DOD seal and the top line DoAF text. What you do not do, however, is just type in your own letterhead at the top of a page.

Sure, you can run out of letterhead at any time. I can remember several times when we were out of unit letterhead, and USAF letterhead both. And, in every case, I had to tritty-trot down to the base printing office and pick some up, so that we could write official documents.

What is really odd is that the USAF style for Memorandums for Record is that they do not use letterhead at all. Instead, you use plain white paper, and the top line simply says MEMORANDUM FOR RECORD.

One of the documents wildly violates the prescribed style:

1. The letterhead is typed.
2. The date is not aligned to the left margin, as required.
3. The MEMORANDUM TO... is incorrect. The proper form is shown below:

MEMORANDUM FOR RECORD

TO: 1Lt Bush, George W.

Additionally, the rank abbreviations are not in the appropriate USAF style. The rank of First Lieutenant is abbreviated as "1Lt" in the USAF; Lieutenant Colonel is abbreviated "Lt Col", with no periods in the abbreviation.

Finally, the unit abbreviation for the 111th FIS incorrectly uses periods. USAF acronyms for units use capital letters only with no periods, as I have written it above.

By the way, the address that has been redacted is 5000 Longmont #8. In case you care.

Now, I freely admit that none of this proves the documents are forgeries. Maybe the guys in the 111th FIS were a bunch of free-wheeling cowboys, who didn't give a fig about the USAFs prescribed styles, because they were TEXAN national guardsmen, who were above all the pretty rules of those lace-panty boys in Washington. Indeed, there's probably some truth in that, because, even as a teenager in the CAP, I remember that the bathroom in the control tower at Ellington AFB was liberally stocked with porn.

Still, the absolute lack of adherence to long-standing USAF standards is odd, at the very least.

Enough of the tedious USAF style stuff. On to the geekery.

In typesetting, there is a process called "kerning". This is the process of grouping letter in a word so that their spacing overlaps for readability. Instead of each letter taking up a discrete amunt of horizontal space, the letters in a word are squished together so that they overlap horizontally. Take a look at this from the 18 Aug 73 memo:

See how the horizontal space of the F and E overlap? This appears to be a case of kerning. Here is another example from the same document:

In this case, it's the F and the R that appear to be kerned, although the RO appear slightly kerned as well, but it's simply hard to tell with the poor resolution of the document.

Computer programs like Microsoft Word do this automatically. They know what the letters are in each word, so they can adjust the letter spacing horizontally to kern them.

No typewriter in the world can do this, because typewriters do not know what the letters of each word are. Not even proportional font typewriters can kern. Typrewiters have know way of knowing whether you've typed an FO, which can be kerned, or an OO which cannot. Only word processors can do this.

Moreover, Charles at Little Green Footballs, did an interesting experiment that reproduced one of the memos exactly using Microsoft Word with the default font and margin settings. In fact, the match was so close that when he superimposed his recreation in Word over the CBS original, he got the result below:

I think all of this raises very serious questions about the provenance of these documents. At the very least, CBS should publicly announce who did the document authentication, along with a precise explanation of how the authenticator arrived at his conclusion.

Oh, and finally, Hugh Hewitt has posted the transcript of his interview with Farrell Shiver, the owner of Shiver & Nelson, a document investigation lab in Woodstock, Georgia. Mr Shriver is also of the opinion that these documents are forgeries.

CBS has some questions to answer.

UPDATE:

Powerline has another blue-suiter who confirms my statements about the Air Force's style requirements.

UPDATE II:

The kerning argument appears to be an artifact of the Times New Roman font itself, which extends some letters beyond their alloted horizontal space. The document itself appears to be completely unkerned. See here for a fuller explanation.

TrackBack