The Powerline blog now has this story, picked up from here, with valuable insight from a former military clerk/typist.
I was a clerk/typist for the US Navy at the Naval Underwater Systems Center (NUSC) in Newport RI for my summer job in 1971 when I was in college. I note the following with regard to the Killian memos:
1) Tom Mortensen is absolutely correct. Variable type was used only for special printing jobs, like official pamphlets. These documents are forgeries, and not even good ones. Someone could have at least found an old pre-Selectric IBM (introduced around 1962). Actually, I believe we were using IBM Model C's at the time, which was the precursor to the Selectric.
2) I also used a Variype machine in 1971. I fooled around with it in my spare time. It was incredibly difficult to set up and use. It was also extremely hard to correct mistakes on the machine. Most small letters used two spaces. Capital letters generally used three spaces. I think letters like "i" may have used one space. Anyway, you can see that this type of machine was piloted by an expert, and it would NEVER be used for a routine memo. A Lt. Colonel would not be able to identify a Varitype machine, let alone use it.
3) US Navy paper at the time was not 8 1/2 x 11. It was 8 x 10 1/2. I believe this was the same throughout the military, but someone will have to check on that. This should show up in the Xeroxing, which should have lines running along the sides of the Xerox copy.
4) I am amused by the way "147 th Ftr.Intrcp Gp." appears in the August 1, 1972 document. It may have been written that way in non-forged documents, but as somone who worked for ComCruDesLant, I know the military liked to bunch things together. I find "147 th" suspicious looking. 147th looks better to me, but the problem with Microsoft Word is that it keeps turning the "th" tiny if it is connected to a number like 147. And finally......
5) MORE DEFINITIVE PROOF OF FORGERY: I had neglected even to look at the August 18, 1973 memo to file. This forger was a fool. This fake document actually does have the tiny "th" in "187th" and there is simply no way this could have occurred in 1973. There are no keys on any typewriter in common use in 1973 which could produce a tiny "th." The forger got careless after creating the August 1, 1972 document and slipped up big-time.
In summary, the variable type reveals the Killian memos to be crude forgeries, the tiny "th" confirms it in the 8/18/73 memo, and I offer my other points as icing on the cake.
Item #4 on this link is the clincher for me, that drives me crazy with Microsoft Word.