Many people have criticized my original post about Dr. Hailey as being too cryptic. I have over a dozen years of experience teaching and training graphics arts professionals. My eyes saw the problem instantaneously. Perhaps dealing with so many people who understand this stuff, I failed to appreciate how many people would not see the problem instantly. Further, this being such a serious charge, I saw great value in the “see for yourself” method. However, after multiple people asked me to explain it in detail, I am laying out the case in a more user friendly fashion. Feel free to ignore the post if you’ve seen enough already.
]]>< 
For those of you who are, by chance, not familiar with this graphic, it is an animation of one of the CBS memos overlaid with the exact same words typed into Microsoft Word WITH ALL DEFAULT SETTINGS. Including default tab stops and the default typeface of Times New Roman. This was the “Ah HA!” moment for many people.[See Footnote 1]
To get to the point where I had a problem with Dr. Hailey’s analysis, we’ll start with page 7 of the pdf where the professor is making the case that all the people who say it is Times New Roman are wrong:
…The only real questions are
So, when Hailey says that he created a virtually flawless replica “using Typewriter condensed as my font of choice,” what he really meant was:
“using Typewriter condensed as my font of choice except when the characters were obviously, dramatically different as with the numbers 7 and 2, and except when Typewriter had no such character (like the superscript th) in which case I dropped in a superscript from some other font I won’t identify for you.”
Boy, if that doesn’t rise to “intent to deceive” it certainly rises to “reckless disregard for the truth.”
Dr. Hailey claims that he achieved an exact match:. But it’s not even CLOSE. After all his hard work constructing this document, there are significant and obvious deviations between his version and the original. I mean, it is so incredibly obvious, that alone should have been sufficient to dismiss his argument. (Though I am glad there is ample additional evidence.)
I’ll bet an LGF-style animated overlay would make that all the more obvious.
Anyway, this is great work, and I appreciate the recap.
Great analysis job, Paul. So I guess Americans are expected to believe that Col. Killian — a man reportedly adverse to typewriters — created his Bush memos: a) using typographical gymnastics; b) knowing that someday Bush would rise to prominence in the political world and the memos provided proof that Bush sucks, a position coincidentally held by the majority of the MSM.
You nailed this one like a one-man blogosphere, and you nailed it fast. IMO, everybody owes you thanks and congratulations.
And you don’t have to apologize for laying on the technical voodoo (that you do so well). That’s part of the power of the blogs.
Let me see if I understand the professor’s points:
1. The memo could not have been created on a computer.
2. I found a typewriter font that matches the memos.
3. I recreated the memo on my computer using the typewriter font I mentioned.
Now, I am no expert on fonts or memos or even Photoshop, but it does appear to me that points 1 and 3 directly contradict each other. Even in his corrected version of the pdf, the logic is wrong.
Paul, you’re missing a red line that should be there — under the missing “2” beginning the second graf. Peculiar that the period is missing there, too. Could he not get it correctly spaced with the period in?
good eyes
I don’t know if this guy commited fraud or not. But I do know he’s not real bright. I saw this stuff about a week before wizbang posted about the flaws and I dismissed his argument in less than a minute. Its silly.
His argument is completely pointless. He says that the “font” used could have been from a typewriter and that Times Roman. Even forgetting the various historical problems with that , it doesn’t matter because there are now a number of computer experts who have stated that the algorithms used for text spacing on a computer versus the very primitive version of proportional spacing. The complex spacing caclulations of a WYSIWYG editor just aren’t going to be duplicated by a typewriuter.
So even if this guy proved that there were typewriters capable of printing leters that look like the letters in the memo, so what? If your arranging those letters by yourself you’ve proved nothing. And the “lack of critical indicators of digital production” is silly if they were run through a copier 20 times.
Is it fraud? I dunno but its clearly pretty damn bad scholarship. Maybe he’s so dumb he’s incapabvle understanding Dr. Newcomer and other computer science professors. I suspect its something along these line because his starting premise is ridiculous. Personally I think he embarked on a project that was flawed from the beginning and lacks an understanding of the problem.
It would seem that the Queen of fraud, Ms. Mary Mapes, referred the honorable Retired Major General Bobby Hodges to this same liar/fraud. You can find the whole story in this article from the Dallas Observer dated 30 Sept 2004:
http://www.dallasobserver.com/issues/2004-09-30/news/news.html
Professor Hailey is a man guilty of willful fraud, at the VERY LEAST. I’d like to see his academic credentials. Some GRE scores, something. But then, he might just be able to create those, too.
What an idiot.
Great work on this!
KCTrio
IN addition to the other problems, he’s missing a period after “instructions” on the final line.
His only defense is that this is so inept it’s not really fraud. It’s sort of like a counterfeit $8 bill.
Paul,
good eyes
Not really, since I didn’t catch the missing 4 at the end of the first line of the body of the memo 😉 I don’t really know what’s the deal with that, since the 4 in the Typewriter character set isn’t wildly different from the numeral in the memo (unlike the 2 and the 3 and the 5 and the 7, &c.).
Has anyone else noticed that the character set Paul provides above from Hailey’s site and the ITC American Typewriter Condensed in his link aren’t the same font? I mean, the latter really is “condensed,” meaning squished.
But neither capital R looks anything like the one in the memo. The right-hand leg of the R in the memo’s font goes pretty much straight down to baseline at a 60-degree angle to horizontal or so. The ones in “Typewriter” follow the upper part of the R, then go essentially vertical, with a serif at the end. Not remotely the same.
IOW, the R in “Report” didn’t come from that font either. I wonder where it did come from. (It also seems to be a little darker than the surrounding text, by the way, kind of like the “2” next to it. But faded in nicely — rather odd that he could blend that in so plausibly but make such a hash of the superscript on the same line.)
If anyone has pointed all this out already, I apologize for retreading old ground. I just haven’t had the chance to read all the commentary here yet.
The right-hand leg of the R in the memo’s font goes pretty much straight down to baseline at a 60-degree angle to horizontal or so.
No, it’s more like 45 degrees. Sorry. But, again, it’s totally different from the “Typewriter” cap-R.
Just wanted to point out that the R in the reproduction matches the R in the original document, but it does NOT match the R in American Typewriter.
The foot of the R in the font is curly. It starts towards the right of the R’s loop, curves out a bit, then back, then out again before hitting the baseline. It’s quite elegant really.
The foot of the Rs in the memo is straight. It starts at the left of the R’s loop, and heads straight out and down to the baseline. It’s spartan and indeed, the R is the easiest way to distinguish between Spartan and Helvetica, which used to be the two most common sans-serif faces.
The F and the S don’t seem to match either. I just think Dr. Hailey got tired of tenure.
Michelle, I noticed all sorts of inconsistencies between what Dr. Hailey claimed the font was a what it looked like. I could spend days documenting it all. No telling how he really put it together.
At the end of the day I suspect many academics think that the letters behind their name allow them to mouth the words sometimes. The problem is when their work leaves their world and comes up for public discourse, it gets actually looked at. (I phrased that poorly)
Point being, it is my personal belief he just kludged anything together and expected his title to be its credibility.
His grammar and punctuation are worse than mine. I’m an engineer, he’s a Professor of English. I have an excuse. LOL
He claims he spent 3 weeks on this document and people have found literally dozens of falsehoods in it. It is literally so weak it is hard to fully attack his logic because he keeps it well hidden. This is the best he could do after 3 weeks? OY!
Anyway to ramble to a close, there are so many problems with the document, I threw up my hands, I can’t document them all. (But I hear thru the grapevine there are people smarter than me doing that for me.)
P
and to tip you off — the 4 has a”foot” problem
I saw he gave up on that one line but I figured since it was the end of the line I’d ignore it. He could make the case he just forgot to finish it or something. The TH was a given.
Michael — OUCH
Michael,
Now that I look at it more closely, that’s hardly the only problem. I’m not sure about the F (the copy quality is too bad for me to tell), but the S sure doesn’t have that huge serif on its back end. And check out the lower-case “t”! I don’t know what he used, but it isn’t what he said he did. The “Typewriter” “t” has a pronounced upwards turn at the right. The one in the memos basically goes down to baseline, levels out, and stops. Kinda like Times New Roman, actually 😉 (Actually the TNR “t” ticks upward at the end, but nothing like the Typewriter one does.)
Paul,
and to tip you off — the 4 has a “foot” problem
Why, yes, so it does. So basically the only digit he thought he could get away with was “1.” Lovely.
Well, I will be very curious to see how this shakes out, but for now I’ll leave it alone. No point finding twenty errors where a few will do.
When I first saw this the other day, it was the “9” in the years that stood out as obvious mismatches, contradicting Hailey’s claims that “the match is exact” and he had produced (or could produce) “a virtually flawless replica.” (See the 4 May 1972 letter comparison.)
As someone who has worked with type and PhotoShop professionally for more than 15 years I have to say that calling what the Nutty Professor produced and exact match is beyond ridiculous.
Listing the number of ways in which the characters do not match would be far too exhaustive for a comment. So I’ll just one more to those listed above. Look at the 9 in the line beginning May, 1972. The shape of the professor’s 9 bears no resemblance whatsoever to the one in the original forgery.
It is also beyond my ability to understand how anyone can assert that their production of a duplicate memo by digital methods proves that the original was not itself created digitally.
Like Michelle, I’m wondering whether anyone else is noticing all of the same logical and technical errors that I’m finding. Paul – you hint that someone is working on compiling all of this. Should we continue posting crtical points, or wait to see what comes out?
LSU Engineer:
Please keep making lists, if it’s not too much trouble. I think it will prove useful.
It’s even easier to prove it couldn’t be done on a typewriter. Try centering the text at the top. Not all that easy to do, which is why your perp never did it.
LSU-
If you guys want to make this an open thread on factual errors you find, that’s ok with me. I already called him enough names (ahem) so that is no longer needed.
But if you all want to make the list, I assure you, the powers that be will see it.
If nothing else, I can only assume the folks a USU are reading every word. If they see that lay people in their pajamas can find so many problems, perhaps they will revisit both their denunciation of us and their support for an academic embarrassment. (yeah right, I’m a comedian)
Seriously, have fun, they’ll use it.
Paul
Go Tigers!
The font in his replica doesn’t look to be the font he said he used in several instances (also notice the lowercase ‘r’), at least to my eye. Just looking at the blown up version of the word “Report”, there seems to be several inconsistencies from the ‘Typewiter Condensed’ characters provided.
ed,
Yes, the “r” is also wrong. Even if the curly bit had broken off or been eroded, there ought to be a trace in the original. And what’s more telling is that there’s no trace in the replica. Which was allegedly made on a computer. What gives? Or do we really need to ask at this point?
“mute” and “moot” mean different things.
AHW CRUD!
thanks Crash
P
bad fingers, bad!
Nice job with this story, Paul.
This post is where it needs to be at this point- quiet, thoughtful, inescapable.
You da man.
My $.02: In addition to all the other mismatches in the crucial overlay graphic (the digits in particular), there appear to be signs that individual characters were micropositioned to match the original forgery. Compare the two instances of the word ‘appointment,’ just above the forged signature. In Halley’s line, the leading ‘a’ is shifted above the baseline slightly, while the ‘e’ looks like it may have been dropped a bit. Slight vertical shifting also appears in ‘be conducted’ on the last line of text, where the ‘e’ in ‘be’ is not baseline-shifted, but the one in the next word is.
If the argument were that the second line was really done on a typewriter, then perhaps you could say its mechanism would cause letters to occasionally strike slightly above or below where they should have. But since we now know the text was mocked up in Photoshop (a process that introduces no such errors), the only explanation for these aberrations is an effort on the author’s part to nudge individual characters slightly so as to produce the spurious “exact match” he advertises in the figure’s caption. Occasionally he gets it slightly off, as can be expected given such a painstaking process.
Sometimes the tell-tale sign is the horizontal spacing. Look at the word ‘administrative.’ Not only are the ‘a’ characters not the same on the two lines, the space after each ‘a’ in Halley’s line (the ‘ad’ and ‘at’ pairs) differs considerably.
In short, I believe the ‘th’ isn’t the only example of such deceptive character positioning.
(Oh and yes, you really meant say such-and-such an argument was “moot,” meaning irrelevant.)
He asks “is the typing mechanical or digital?” and then answers that it was “typed on a machine.”
Seems pretty clear to me that he was trying to claim that it was created on a mechanical typewriter as opposed to modern electronic device.
Line up 1 million CBS executives in front of 1 million 1972-model IBM typewriters.
Seat one 12-year-old American child in front of a 2004 Microsoft Word word processor.
Give them all this address to re-type and center:
111th Fighter Interceptor Squadron
P.O. Box 34567
Houston, Texas 77034
Who do you bet will be first to center the address exactly as it is centered on Dan Rather
Your eyes saw that the TANG memos were forgeries because you are a computer expert/graphics art professional. My eyes saw it immediately because I am old enough to have done my work from high school onward on a typewriter. To make a superscript, you rolled the carriage upward half a line manually. It was necessary to hand onto the roll with one hand so it wouldn’t snap back. And then you typed the exponent with the other hand. But the font in the exponent was exactly the same size as the baseline font.
I haven’t gone through everything, but I read the professor’s original first draft that you posted and your arguments.
It seemed to me that the professor was trying to argue that the font used in the memo was a very close (virtually identical) match to the general family of fonts known as Typewriter.
I don’t think he ever suggested that he actually “typed” his version on a typewriter, but simply that he used the “Typewriter condensed” font for his comparison.
All of the other stuff regarding his cut and paste “additions” to his version do not ultimately detract from his basic arguments.
He believes that that the font in the memo is virtually identical to the “Typewriter” family of fonts commonly used in typewriters for many years, and different from Times New Roman and other modern electronic fonts. He also believes that the appearance of some of the letters in the memo suggests (but does not prove) that the memo was typed (at some point in time).
Oddly, one of his last suggestions seems the smartest and most straightforward means of resolving the question. Simply, examine a whole bunch of documents produced by the 111th during that general time period to see if those documents match the appearance of the memo.
Hasn’t anyone done that? It seems so obvious. What better way to disprove the possible authenticity of the memo than to contrast it against genuine documents from that unit during that time period.
Of course, even if there was a match, it wouldn’t prove the authenticity of the memos. But, it would probably elevate the discussion from the innuendo and character attacks often seen in these matters.
From what I’ve heard, it appears that these memos are fakes. The professor acknowleges that in his paper. He simply believes that the typographical analysis is not dispositive of the question.
You may have the expertise to dispute his arguments. But to argue against things he did not say only creates confusion.
Rather than continuing to critique Dr. Hailey’s work down to the last serifs or decenders etc., can someone here attempt to duplicate the Killian docs using American Typewriter Condensed and do an LGF flash comparing it to the original. That should put a lot of this to rest. Won’t be done on a typewriter, but Dr. Hailey’s work wasn’t done that way either.
The real test will come from the CBS review committee. I think the best thing that happened was Thornburgh’s appointment. From his Attorney General days he has knowledge of and access to the top forensic documents experts. And they will have valid credentials this time. Typography is only half the equation, the other half is the Air Force nomenclature and abbreviations used in the documents. I have no doubt but they will find that they are forgeries. I don’t think they will mince words either.
I think Dan Rather’s objections to Mr. Thornburgh come more from his prosecutorial abilities than his party affiliation.
This may bore everyone to tears, but I had a few hours to analyse a single section of Mr. Hailey’s work–the paragraphs around Figure 6, which supposedly demolish the idea the Times New Roman is present in the memos. I hardly touched the character-spacing issues, just responded to the single-character comparisons:
Figure 6: (comparison of the 3 “stays in a flight status” renditions):
1. Why is the TNR sample in bold? No one has asserted that the Killian memos were done in boldface.
2. Why does the TNR sample omit the word “a”?
3. Do you disagree that using TNR regular and using the full phrase would have produced a sample markedly closer to the appearance of the May 19 memo? In an academic research project, is it responsible to not display the best evidence that supports an opposing theory?
4. Does the IBM typewriter ball you refer to have any IBM part number or designation on it? (Please check the top and inside the ball). Do you know when it was available, and what machines it was available for?
5. How exactly was the “IBM Typewriter Font Condensed” sample created? Did you type in the phrase with a certain typeball, scan it, and present it unaltered? Or did you scan in a typed phrase and change the positions of the letters to simulate the character spacing of the May 19 memo? Or did you use a digital font–if so what font is it and what manipulations were done to create the published image? Cf. Figure 8’s caption, where you say you collected exemplars “from comparable documents in our archive.” Was the IBM sample created by mixing individual letters scanned from different documents?
6. In saying the IBM font is “condensed”, do you mean that there is a typeball with that designation? Or that you condensed the spacing between the characters to match the memo? Or that you altered the widths of individual characters to match the memo?
7. If you yourself condensed the IBM font in any way, are you aware of any 1972 typewriter which could definitely produce the same effect of condensation?
8. You state “One can see the same convex bases in both the Bush memos and the IBM font.” Why don’t your exemplars of the memo glyphs (Figure 8) display these convex bases?
9. If you think question 8 is explained by erratic digitalization of the memos, how can you be certain that you have correctly identified the “true” shapes of any of the glyphs? Isn’t it equally possible that the memo was composed in TNR or some very similar font, and all discrepancies with TNR’s glyphs are explained by poor digitization?
10. You state that “The strokes in the text have a consistent width in both the Bush memo and IBM font. (Times characters have inconsistent width.)” Please explain the nearly invisible serif and attenuation of the right arm of the memo’s “y”, which looks much more like TNR than IBM.
11. You state “The serifs are heavy and have consistent weight. (The Times serifs are short spurs.)” Please explain why the “i” of the memo’s “flight” has nearly lost its “flag”. Is this not more consistent with TNR than IBM? Please explain why the tails of all the memo’s “a”s are not as distinct as the IBM’s. Is this not more consistent with TNR than IBM? The memo’s “h” appears to have a “drooping flag”–is this not more consistent with TNR than IBM?
12. You state “Cross strokes on the “t”s are heavy in both the memo and IBM font. (Times New Roman cross strokes are fine.)” To my eye, the left-hand cross-stroke of all the memo “t”s is very weak–much more like TNR than IBM. Please specify why your estimate of the evidence is superior to mine. (Some of the IBM’s “t”s appear to have been cut off by digital manipulation–true?)
13. You state “The bowls of the Bush and IBM “a”s are of consistent width.” Is this untrue when the memo is compared to a proper sample of TNR regular? If not, how does it strengthen your case?
14. You state “The “e”s are similarly closed in American Typewriter and the memo, and their crossbars are more dense.” What font are you referring to specifically–the IBM typewriter ball, or ITC American Typewriter, or something else?
15. Since the image we are referring to has no “e”s in it, is it academically responsible to make the preceding claim? How can the reader evaluate your claim?
16. You state “The bottom serif in the “i” in the Bush memo is completely compatible with the IBM font. It is slab-like, and contains a comparable curve.” Please explain how you can come to this conclusion when the two “i”s of the memo are so different from one another: one is almost missing its upper serif, the other appears to have both a leftward and rightward upper serif. One “i” has a base curve, the other doesn’t.
17. The dot of the memo “i”s appears to be consistent in height with TNR but not with IBM (where the dot is lower). Please comment.
18. The memo’s “n” shows almost no upper-left serif, a prominent feature in the IBM sample. This seems more consistent with the subtler TNR serif. Please comment.
19. The image of the actual May 19 memo text in Figure 6 has much more detail than is contained in CBS’s PDF. (In my JPG file extracted from the PDF, the letter “f” is 17 pixels high and grayscale. In the image extracted from Hailey’s PDF, the “f” is 33 pixels high and has color information). Is your source for the memos different from CBS’s PDF? If so, please post it for us to examine. If you digitally manipulated the CBS JPG image, what exact steps did you take? What steps did you take to ensure than the manipulation did not add false detail to the image. Could any false detail have influenced your conclusions?
(Hint: the question of false detail arises again with Figure 13, the “broken e’s”)
20. Please publish a scan of the full character set of the “IBM Typewriter Font” ball you refer to, including the numerals and punctuation marks. Are there any typeball characters not included in the phrase “stays in a flight status” which are conspicuously unlike the memos’ characters? If so, why did you omit discussion of them from your document?
21. You state (following Figure 7, which I am not addressing in detail) “In no sense are the defining characteristics of the characters in “flight” in the Bush memo like the defining characteristics of the characters in the Times New Roman sample. On the other hand, every defining feature of the Bush memo is comparable to the critical features of American Typewriter bold condensed font.” Given the preceding questions and criticisms, is that a fair summary of the evidence?
Dummy, one of the Washington Post articles (unfortunately, I don’t remember which one) mentioned that several other memos from the same time period and location DID NOT match the CBS memos in many particulars – typeface, centering, alignment, word usage, lack of superscripting, and so on. On the other hand the CBS memos DID match each other very closely – despite being supposedly printed over a substantial period of time.
This was the professor’s suggestion (apparently he doesn’t read the Washington Post):
A productive approach is to examine more documents that came from the
111th during that period. If I am right, there will be documents that look exactly like the
Bush memos with flaws I believe I have identified; that would go some distance toward
proving provenance. On the other hand, if there are no documents that match the type, it
demonstrates that the documents are probably bogus.
Again, nothing has come close to matching. Obviously, it would be big news if ANY document from the time period looked similar to these memos, and he is far from the first to suggest this as a way to test the memos.
I zoomed into the memos and saw that Hailey’s reproduction had bits of colors around the edges.
If you are running Windows XP (I’m not, but I’ll give it a spin at work tomorrow), you should activate the ClearType font rendering (this is somewhere in the Control Panel). Type up a partial reproduction of the reproduction in Word (more than a word or two aren’t necessary), then take a screen cap, then zoom in it in Photoshop until you can see the blurring around the edges.
Cleartype is subpixel rendering. Zoomed up, it looks like it has bits of colors around the font. (TrueType would show only shades of grey.) If the colors around the edges matches what you’ve got in the memos,well, you’ve proved beyond a shade of doubt it’s a duplicate generated in Word.
Dummy,
Yes people have compared the phony documents to authenticated originals from Killian and the TANG base from the time GWB was there. Of course they aren’t even close.
Here was is a good comparison from WaPo:
http://www.washingtonpost.com/wp-srv/politics/daily/graphics/cbsdocs_091804.html