From Indc Journal:

Using something called the Haas Atlas, the definitive collection of various typefaces, Mr. Bouffard (and other forensic document examiners) examined the veracity of various documents for over 30 years. Beginning in 1988, Mr. Bouffard hired a programmer to write a computer database program that catalogues the nearly 4,000 typefaces that appear in the Haas Atlas. This computer program is now a forensic standard that is sold as a companion to the Haas Atlas by American Society of Questioned Document Examiners (ASQDE).

What did Dr. Bouffard think of the documents?

First, the necessary caveats:

• The pdf document is of poor quality. It seems to have been copied and recopied several times, blurring letter characteristics.

• Also, certain types of analysis can only be done on the original documents, which don’t seem to be available, even to CBS.

So Dr. Bouffard is very clear that his analysis is not 100% positive. That being said …

“It’s just possible that this might be a Times Roman font, which means that it would have been created on a computer. It’s very possible that someone decided to create this document on a computer... I’ve run across this situation before … my gut is this could just well be a fabrication.”

The reasons why?

• Right off the bat, Dr. Bouffard noted what others in the blogosphere have been talking about – something called “proportional spacing,” which means that each letter does not take up the same amount of width on the page. On old typewriters that do not have proportional spacing, the letter “i” would be as wide as the letter “m.” Except for professional typesetting, proportional spacing was only available on a very few models (an IBM model, "Composer" and perhaps one or two other models) that were not widely available in 1972-73; the vast majority of typewriters did not have proportional spacing. Because of this, Dr. Bouffard’s computer program immediately eliminated “over 90%” of the possible fonts from typewriters that could create such a document, narrowing it down to perhaps 15 fonts used by a very few models.

• Next, Dr. Bouffard began entering individual characters in an attempt to match them to the remaining fonts that were available on proportional spacing typewriters of that era, focusing on numbers. Thus far, one character stood out, the number “4.” In the document provided by CBS News, the number 4 does not "have a foot" and has a “closed top,” which is indicative of Times New Roman, a font exclusive to more modern computer word processing programs. other characters matched the old proportional spacing fonts (available on only a small few typewriters of the era), but this number did not (please note that this is only an initial analysis with numerical characters).

Dr. Bouffard ran this number and could not find a match in his entire database of over 4,000 typewriter fonts that have been maintained and collected into his computer database since 1988. Otherwise, the font is very indicative of Times New Roman, the font that is only available on computer word processing programs.

The final word?

Once again, let’s not forget the qualifications: it's a bad copy of a copy and we have no original document for review, but, based on the initial analysis of the documents by an industry expert with over 30 years of experience in typesetting and forensic document examination, the documents “could just well be a fabrication.”

In light of this information, I think that it would be highly appropriate for CBS News and the Boston Globe to attempt to obtain a copy of the original document for more thorough vetting, and run a correction/addendum to the story.

I still have two other forensic document examiners that are examining the pdf file, and I will update if/when they get back to me. I also plan to ask Dr. Bouffard more detail about the nature of the "th" on the end of dates, though in our first conversation he indicated that some typewriters had the capability to do something in that format.

UPDATE: Dr. Bouffard called me again, and after further analysis, he says that he's pretty certain that it's a fake.

Here's why

* He looked through old papers he's written, and noted that he's come up against the inconsistency of the "4" several previous times with forgeries that attempt to duplicate old proportional spaced documents with a computer word processing program.

* Regarding the small "th" after the date, Dr. Bouffard told me that it was possible to order specialty keys that would duplicate the automatic miniaturization completed by word processors after a numerical date, but it was certainly not standard, and wouldn't make a lot of sense in a military setting. "That by itself, while suspicious, is not impossible, but in conjunction with the (font irregularity of the) number four, it is really significant," he said.

* Dr. Bouffard said that signature analysis isn't that relevant because the signature could have easily been copied and pasted onto one of the photocopied forgeries from another document.

* He said that he didn't know who CBS contacted to verify the document's authenticity, but that there is really only one other man that may be more qualified to determine authentic typefaces than himself. I think that the burden of proof may be on CBS to reveal this information.

I asked him to put a percentage on the chances that this was a fake, and he said that was "hard to put a number on it." I then suggested "90%?" Again he said it's "hard to put an exact number, but I'd say it's at least that high, sure. I pretty much agree that that font is Times New Roman."

But of course, he must be lying, right?