(NOTE: I’ve updated this post quite a few times to make corrections and clarifications. Sorry about that. I just want to be clear about what I’m saying and what I’m not. Also, please read the updates at the bottom, as they add further, important information than what is in the main body. Last update: 2:12pm 9/13)

LittleGreenFootballs has presented some quite convincing cases (September archives) that the documents pertaining to Bush’s National Guard Service are forgeries. I have some thoughts on this as well. Here are my creds:

Besides my lifelong interest in the medium, I studied typography and letterpress printing at Univ. of Southern California for two years under a master printer who has since moved on to teach digital typography at Art Center College of Design, in Pasadena, CA, and I have used digital typesetting tools since their inception, in addition to working with lead type, Linotype, and photo typesetting machines. So, I feel I can at least make some well-informed, though not expert, comments on the material that has been discussed in various blogs so far. I believe there are typographic issues with these documents that to date (9/13) still have not been even remotely addressed by any of the so-called experts employed by CBS, or, an explanation about “why” some of these issues are important in the blog posts I’ve seen so far with the general exceptions of Shape of Days and Flounder.com. Therefore, this post is not as much about actual specific comparisions of the two sets of documents, as it is about the theory behind the typographic comparisons that many other posts have made hitherto. I am merely attempting to fill in certain gaps left in the discussion.

The man who would really be the expert, dealing with issues that have not hitherto been addressed, however, would be my former professor, since he is probably one of the most well-informed people on digital typefaces and their relation to their historical predecessors. He has a letterpress shop in Los Angeles under his own imprint, and besides his current teaching post, is frequently asked to review digital typefaces for various publications, and type designers. He does not have a blog, but we have discussed this material on the phone and the following is a summary of that discussion. I won’t publish his information in the blogosphere, for a variety of reasons. I will, however, forward his name and phone number to any organizations that wish to speak with him.

So then, the Bush docs made available on the CBS website were digitized at too low a resolution to accurately compare serfis, thicks and thins against a similarly .pdf encoded version of LGF’s Microsoft word document. They needed to be at digitized at 2400 dpi and then translated to .pdf at full res for an accurate comparison to be made, which would have made this process very easy. Nevertheless, there are some things about typefaces in general that are important to note:

Just because a typeface is said to be at 12pt, doesn’t mean it is the same size or that it fills the same space at another typeface that also is labeled 12pt. Nor does it mean the same typeface would produce the exact same results via a different typographic medium. Point size isn’t a typographic absolute, with regard to the size of typefaces. 12pt. Times New Roman isn’t necessarily the exact same physical size as 12pt. Caslon or 12pt. Garamond, nor is it the same size as 12pt TNR from photo typesetting, Linotype or Monotype matrices. They are close, maybe very close in size, but not the same size. This has always been true of moveable type since it’s inception, and every other typographic medium since. The reason for this is that the design of each typeface was hand crafted individually, not by machines. And most of the time, when the typeface was translated into a new medium, the actual size was not exactly matched. Between typefaces that look extremely similar (or “font”, and computer users understand it), the width of the body of the letter (x-height) can be different, the ascenders (the tall part on the “h” or “d” or “b” (descenders for “p”, “q” and “y”, etc.) can be longer or shorter, and the natural letterspacing is usually smaller or larger. These are, in fact, some of the factors which visually distinguish one typeface from another. So, 12pt isn’t 12pt. isn’t 12pt across typefaces, or across media.

Furthermore, metal type, the basis upon which all other subsequent forms of original typeface designs were copied, is not measured exactly 72 points to the inch. It’s almost 72, but not exactly. When Apple created “fonts”, they fixed that problem, and so all typefaces since have been measured by a fixed and absolute 72 ppi. If the resolution of the documents in question were better, you could measure the difference between the so-called “original” document and LGF’s new MS Word doc with a typefinder (a magnifier with absolute point measurements on it used for comparing typefaces) because in all likelihood, 12pt. TNR from 1973 would be slightly larger or smaller than the MS Word doc, because of the units of measurement (there are other reasons as well, which I’ll get to next). While on one line of type this would be difficult to notice, over many lines of type the differences would be made more apparent. Even so much as a 1/8th of a point difference would result in noticeable drift over the course of a page. The 18 August 1973 document in question has 14 lines of type, enough to notice any progressive distinctions.

Given the coarse nature of the documents currently on the web, the Microsoft Word versions that LGF made appear to match all but exactly. For this to be possible, the physical processes involved for creating lines of type must account for a wide variety of typographic factors, the collusion of which would be nearly impossible to replicate, given the inherent limitations of physical media such as an IBM Selectric or phototypesetting, i.e., these machines are very inflexible, relative to the computer. Type size, letter-spacing (kerning), word-spacing, and leading (the space between lines of type) would all have to be exactly identical, and even the slightest deviation would be very easy to spot over many lines of type. Basically, for the LGF Microsoft word doc and the Bush docs to match obviously do on LGFs blog, with regard to spacing, it pretty much requires one to stretch of the imagination to allow for this: that two nearly exact typefaces were used crossing a completely different medium (which involve different physical, mechanical processes that directly affect how the type is printed) resulting in the serendipitous confluence of not one or two, but all three axes of spacing — letter, word and line spacing — such that the older, more coarse document when overlaid by the new one looks virtually identical to a document created with the default parameters of Microsoft Word. The likelihood of this is next to zero.

Overall, the IBM Selectric theory really isn’t a viable option, for reasons that follow:

Even if the use of proportionally spaced Times New Roman (which, BTW was originally invented in 1931 and there was suspicion that it was copied from another earlier typeface by Starling Burgess) on that machine could recreate the MS Word kerning (space in between letters), it would not be able to do so in terms of word spacing. Word spacing on a typewriter is mono-spaced, that is, there is one unit of measurement for a word space, which is defined by the internal grid of the machine, not the ball that is used. That is why it was common practice amongst the lay public that correspondence which was typed on a typewriter used a double-space after the period, to demarcate the difference between a typical wordspace, and the beginning of a new sentence. Professional typesetters never did that (then or now), because they could control the proportionality of wordspacing in a hand-set document, but this practice wasn’t widely known outside printing circles until Robin William’s book The Mac is Not a Typewriter came out in 1990. Just so it’s said, both the LGF MS Word doc and the so-called “original” reveal a double space after a period. Whoever created these documents, at least they got that part right.

So, since the actual physical mechanical mechanism of a typewriter wouldn’t have anything else but one space measurement for hitting the space bar, and the documents in question are proportionally wordspaced (and not center justified - see Update #3), that further makes it incredibly unlikely these documents are originals from 1973. Not only that, but a carriage return on a typewriter would also cause monospaced leading (line spacing), also according to the internal grid of the machine. Some machines would have multiple settings like single, 1 1/2, and double spacing, but those were all of fixed widths by virtue of the physical mechanics of the machine. Microsoft Word, however, usually has proportional leading as the default option that is calculated/determined according to the size of the type chosen by the user. Usually this is “single” spacing as defined by MS Word. If the leading was not automatically set to a proportional spacing, lines of type would either be too far apart for typesizes under 10pt, or grossly overlap from top to bottom for type sizes over 16pt. Since that would be a variable that the user would have to manually change every time s/he decided to change the type size, the auto function creates a generally pleasing look to the line spacing, and saves the user the trouble, but it is based upon a percentage of the size of the type, not necessarily upon traditional typographic leading widths. It is very unlikely the two sets of documents are from a different era, as it would require the highly improbably occurrence that the auto single spaced leading in MS Word would have exactly matched the monospaced leading of the typewriter. Any difference at all, however slight in the leading of either document would be quite visibly exaggerated over the 14 lines of type when the two documents are laid on top of one another (see the link at the bottom for just such a comparison).

Also, the balls that the Selectric used would have come largely in two sizes in the United States: pica and elite (12pt. type (approx) and 10pt type (elite designated a typewriter typeface that was 12 chars to the inch, or about 10pt.)). As I stated before, 12pt Times New Roman on a typewriter isn’t necessarily 12pt Times New Roman on computer, and even the slightest variation would produce drastically different results (see link at bottom again).

Photographic typesetting, on the other hand, which is really the only possibility for these documents not to be forgeries, had it’s own spacing paramaters that would have been fixed given the nature of the project. I cannot tell you if the mechanism in a photo typesetting machine wouldn’t make the two documents look very similar, but they are difficult machines to use and generally output film or transparencies. Firstly, the transfer of the photomatrices to the negative would cause slight distortions, but arguably imperceptible if the job was done right, which almost no job ever was, since they are not exactly user friendly machines. Secondly, these negatives would have then been sent to a printer who would have made plates and printed from this film. These were just not used to print directly to a sheet of paper. Besides this, there is a lot of translation in this phototypesetting process before you get to the final document, each stage of which adds variables which make a near perfect overlay as seen on LGF, highly improbable.

Now, there was technology out there that could do some of these things (some of which is more sophisticated than what is being used today), but the problem with that theory is that those machines were incredibly expensive, and, generally operated by tradesmen. There would have been no point for the military to have paid such an expensive price for a machine used by a National Guard Fighter Wing so that a Lt. Col. could bang out a memo. It would have been used to typeset a book, but not a memo. Besides that, the Lt. Col would never have waited to use that machine to type a memo, because in the time it took to set up the machine, the memo would have already been typed on a typewriter. It’s like sending this blog to an offset printer, scanning it, and then having a text reader decypher the scan. It’s just an incredible amount of overkill for something so minor.

So, here’s a summary: using computers, it is very possible (though painstaking) to exactly recreate a historical document, but the reverse is almost statistically impossible for the reasons described above. As my former prof put it, it would be the equivalent of winning the California lottery, and you actually have a greater likelihood of being eaten by pigs.

Here’s the proof in the pudding:

• View the original documents under a typefinder, which is basically a magnifying glass with absolute type sizes imprinted on the lens.

• Compare the two sets of documents around the serifs, thicks and thins of the letterform.

• Measure the proportional wordspacing.

• Measure the leading.

• Find an old IBM Selectric with a TNR ball and type something out.

You’ll know in a half a second whether or not it’s a legit document. Personally, based on the evidence I think the docs are a forgery.

UPDATE #1 - Changes in Times Roman, and Times New Roman from 1973 to the present.

My former professor pointed me to this reprinted letter by Charles Bigelow (who is a very distinguished typographic professor): Times Roman and Times New Roman were made by different foundries (Linotype and Monotype respectively), and the design changed over time, esp. since the 1980s. This implies that if the two documents were actually crafted in different time periods, they would most likely look different and be spaced differently, since they would be based on different historical precedents. Since they look nearly exactly the same, as LGF (linked above) has shown, the likelihood that they were created in two different time periods is next to zero. Here are some excerpt from the Bigelow letter:

“During the decades of transatlantic “sharing” of the Times designs, and the transfer of the faces from metal to photo to digital, various differences developed between the versions marketed by Linotype and Monotype. Especially these became evident when Adobe released the PostScript version, for various reasons having to do with how Adobe produced the original PostScript implementations of Times. The width metrics were different, as well as various proportions and details.”

Also,

“When Microsoft produced its version of Times New Roman, licensed from Monotype, in TrueType format, and when Apple produced its version of Times Roman, licensed from Linotype, in TrueType format, the subtle competition took on a new aspect, because both Microsoft and Apple expended a great deal of time and effort to make the TrueType versions as good as, or better than, the PostScript version. During the same period, Adobe released ATM along with upgraded versions of its core set of fonts, for improved rasterization on screen. Also, firms like Imagen, now part of QMS, and Sun developed rival font scaling technologies, and labored to make sure that their renderings of Times, licensed from Linotype in both cases, were equal to those of their competitors. Hence, the perceived quality of the Times design became a litmus for the quality of several font formats. Never before, and probably never again, would the precise placement of pixels in the serifs or ’s’ curves etc. of Times Roman occupy the attention of so many engineers and computer scientists. It was perhaps the supreme era of the Digital Fontologist.”

UPDATE #2:

Here are some great links ‚Äî Shape of Days, “The IBM Selectric Composer” talks about exactly what I’ve listed here, with visual examples from an IBM Selectric Composer.

From the Selectric Typewriter Museum, an opinion that the documents were made in MS Word. (hat tip: Wizbang)

RatherBiased brings up many of the same points here (Hat tip: Leroy Leontin, commenter below)

UPDATE #3:

A Freeper pointed out that proportional wordspacing was available on some IBM Selectric Composers, however, this seems to only have been available when typing justified lines (meaning, both the right margins and left margins of the copy formed a straight downward line). The memos in question are all left justified, right ragged lines, meaning, the left margin is justified, the right margin is variable. Yet, the words are proportionally spaced on the memos anyway (look for the link entitled “Justification sans digital electronics”). Good job pointing that out though!

Here are more Freeper comments on this post, with links: Free Republic.org

And, of course, the blog that started it all: Powerline Blog

UPDATE #4

I can’t believe I forgot to mention this, but the absolutely simplest way to tell if the documents in question were created using a typewriter is by looking at the back of the sheet of paper. The hammer of a typewriter key will leave a visible impression. For obvious reasons, a computer will not.

Someone has mentioned to me that CBS is claiming that they don’t have any originals, only first or 2nd generation copies. Obviously, that prevents us from checking, but my question is: why would they only have a copy? Where would the original be kept, if not in the proper file?

UPDATE #5

Some good stuff on Hugh Hewitt. Personally, I think Hugh needs to call a typographer or typographic historian, not a “documents expert”. People keep harping on the kerning issue, when the more important issues are proportional word spacing on a non-justified line of type, and proportional leading. These considerations are far more damning to the authenticity of the documents.

UPDATE #6

This site is an absolute must see: Flounder.com. He is a qualified expert in electronic typesetting, and does a far more thorough job of explaining details I did not, especially regarding kerning. What I described in generalities, he describes in painstaking specifics and should, even given the narrow range of his discussion, put to rest any lingering doubts about the authenticity of the documents.

Also, Powerline linked this post here.

bLogicus has a pretty thorough list of articles and blogs that reference this issue.