QandO
Questions and Observations| « Superscriptastic! | Main | Hodges retracts CBS authentication » |
|
September 10, 2004 To kern or not to kern? Bill, at InDC is retreating from the claim that the documents are kerned. I'm not sure why he's making the claim, because, as of yet, he's offered no explanation. In the analysis I posted last night after blowing up the images and comparing the spacing, I found two examples where it appears the letters are kerned. But, after seeing Bill's retraction, I decided to look at the issue again in Microsoft Word, rather than just looking at the available copies of the memos, and looking for overlapping horizontal spaces. The problem here, I think, is a confusion between kerning in Microsoft Word, and kerning within the font itself. Times New Roman (TNR) kerns the letters f, j, and by a tiny amount, r, in that they extend outside the horizontal space alloted to the letters. Microsoft Word adds additional kerning by further compressing the space between letters. Unfortunately, the effects of kerning in Word for 12pt TNR is hardly noticable, except in a few isolated cases involving capital letters. For example the letter combination "YA" appears slightly differently in TNR in kerned and unkerned Word documents. Based on that, I reproduced the 18 Aug 73 memo in both kerned and unkerned forms, in Microsoft word, and compared them to the original memo. After viewing all three side by side, I have come to the conclusion that the original document is not kerned. The kerning we see in the memo is solely an artifact of the font, not of additional kerning from Word. As such, I think I have to retract the kerning argument, because I simply can't say whether a typewriter version of TNR, if it was available, wasn't also similarly kerned internally. It is vitally important to ensure that we honestly examine and re-examine our arguments, to ensure that what we are presenting is as correct as we can make it. That includes making the appropriate retractions when further investigation proves us wrong. |
Well sh!+. I just retracted my kerning story. I feel like Kerry.
The damn memo was kerned before it wasn't.
The problem lies in the noise artifacts on the memo. Some letters appear to cross into other's but it's not from the typewriter / laser printer.
Still, getting within a PIXEL of perfection with the default Word setting still rings a bullcrap bell the size of Big Ben.
Posted by: Sharp as a Marble at September 10, 2004 10:02 PM
I don't quite follow the analysis here. I think the kerning issue is quite alive. I did my own simple analysis here and concluded the document in question indeed contained a kerned font.
The conclusion I make is that the document, if produced in 1973 or prior, was created on a typesetting system, not a typewriter. On typewriters, each character is "dumb", printing in its own space following the previous character.
Even a proportional space typewriter has to allocate a predetermined and fixed width to each character. No chance of overlapping character space unless you deliberately backspace-- hardly likely in a memo. Even still, you would not just happen to hit the word wrapping and line breaks identically to the current versions of Microsoft Word.
The docs are fake, probably produced on a personal computer using Microsoft Word and either an inkject or a laser printer.
Posted by: Roger Snowden at September 10, 2004 11:00 PM
Roger, You are confusing internal font kerning with documen t kerning It is, in fact, theoretically possible to create a character in a font where part of the character extends beyond the horizontal space alloted for the character. This allows the font to provide partial kerning--and, hence, better readability--without regard to the letters that precede or follow the kerned character.
The fact that its possible does not necessarily mean that it's likely, but it cannot be ruled out.
In any event, it's clear now that the document itself does not use the kerning features of Microsoft Word.
Posted by: Dale Franks at September 10, 2004 11:15 PM
I think we're getting into techie territory here, and need to keep our definitions straight. Here's my understanding of it:
1. In a computer font, Kerning properly refers to a specific pair of letters which the font designer thinks would look better if they were placed closer together. The designer specifies the optimal placement, but the letters will not be kerned unless the word processor is instructed to use the kerning table.
"YA" is a popular kerning combination. In the Aug. 18 memo, it appears to me that the letters "CYA" are not kerned, so if the memo was generated by a modern word processing program, that program did not have kerning turned on.
2. Word and Wordperfect allow "manual kerning". In this case, the user selects any pair of letters and slides the second letter left or right to please his own sense of aesthetics. (The option in Word is called "character spacing"). This is rarely used. There is no evidence of it in the memos, but I wouldn't expect it anyway.
3. Modern computer fonts have character width tables. These specify, among other things, how much whitespace will appear between the right edge of one character and the left edge of the next. These tables are crucial for proportional spacing. They are used by the word processor/printer whether the user likes it or not. N.B., a professional font designer would not refer to this as "kerning".
As pointed out, a peculiarity of bookish fonts like TNR is that the lowercase f has a negative right margin (I think "right sidebearing" is the term of art)--the letter following f will actually be nestled under the "f". This is particularly obvious with combinations like "fe" and "fr".
The memoes contain a number of f-combinations. It seems to me that the letters following the "f" do indeed nestle under the loop of the f, implying that they may have been produced by a computer font. However, the quality of CBS's scans is too poor to be certain. Has anyone started pestering CBS for high-quality scans of their most original versions?
If the memoes really do exhibit this particular type of kerning, defenders of the memoes must demonstrate that Killian had access to a typewriter that could produce this subtle effect. I find it very hard to believe that any 1970-ish office equipment could do it. The amount of the leftward shift at normal typewriter font size is...let's see... only about 1/60th of an inch. Until computers came along, it wasn't worth the bother on any equipment designed for a normal office.
Posted by: Passerby at September 10, 2004 11:25 PM
I worked in the computer graphics industry from 1984 to 1998. I know more about fonts and how various software and hardware reproduces them than any one in good mental health would want to. After a while you get to where you can recognize what fonts (even distinguishing between very similar ones - TNR vs. Palatino for example) are used in a document on sight and even hazard a pretty accurate guess as to what desktop publishing or word processing program was used to create them. It's analogous to an experienced mechanic being able to tell what's wrong with a car by listening to the engine.
As soon as I saw the August 18, 1973 document, before reading any of the opinions on it, I knew immediately at gut level that it was a fake. It is a modern document produced in Microsoft Word.
I'll spare you most of the technical mumbo-jumbo and cut straight to the two facts that cannot be argued:
1. Typewriters do not kern even when the design of the font calls for kerning of specific characters or character combinations - design specifications are sacrificed to the technological limitations of the hardware. CBS - show me a kerned document - any kerned document - typed on a 1972 typewriter or come clean and STFU.
2. The only systems capable of producing documents with kerned fonts available in 1972 were high end typesetting systems costing tens to hundreds of thousands of dollars. The document was composed on a non wysiwyg system by highly trained personnel and then output to film for transfer to a printing plate. These systems did not output to paper.
Posted by: Tim at September 13, 2004 10:56 AM
I do, however, suck at posting to message boards.
Posted by: Tim at September 13, 2004 10:59 AM
I agree with what you say - makes sense to me.
Looking for some propecia?
Posted by: propecia at November 15, 2004 01:41 PM
I agree with what you say - makes sense to me.
Looking for some propecia?
Posted by: propecia at November 15, 2004 01:41 PM
I agree with what you say - makes sense to me.
Looking for some propecia?
Posted by: propecia at November 15, 2004 04:46 PM