Friday, November 17, 2006

Closer Look at the "Typos" Give More Doubt

By Correspondent Marc, who reconstructed the Ferret pages.

Click on a page image to see the original PDF scan.

The recent news took me back to the original Ferret documents where I noticed some anomalies I had not before.

Up until now, I was not too sanguine about the "label typo" defense, but looking again at certain pages of the lab report has made me a believer, at least to the extent of saying that the mistakes are serious enough to throw significant doubt on how reliably we can associate the laboratory conclusions with Floyd Landis. The relevant pages of the report (they relate to the A sample, I believe) are USADA 008 (= Ferret p. 3) and USADA 009.

I had been puzzled by the newspaper reports (Le Monde, Figaro, etc.) saying that the labelling error had been in typing 994 474 instead of 995 474 (Floyd’s actual ID number) since, on USADA 008 I had earlier noticed an error substituting 995 475 for 995 474 (that is, an error in the sixth digit rather than the third digit). As it turns out, though, I think both errors were made (by different people), and together they muddy the water enough to make me queasy about relying on the lab results.

[MORE]



It was looking at USADA 009 that showed me something I had not noticed before. On p. 009, the official in charge of "approving" the IRMS CIR--the isotope--results (this official and all others, including the "Caroline Loubet" who signed at least one of the letters in the packet Sam Abt received, can be identified from pages 019 and 020 of the report) repeats the results reported on p. 008 (the lab report). The original lab report is dated July 24; the approving report is dated July 25. No problem.

But the approving official clearly wrote the wrong ID number, originally writing 994 474, I think, and corrected it by writing a much larger 5 over the original wrong digit. (This, I think, must be the source of the newspaper reports of what the labelling error was.) Even this would scarcely be impressive, except that if you go back to the lab report, it looks very much as if the person "validating" the original test also wrote down the wrong ID number. It looks to me as if the correct number, 995 474, has been written on top of correction fluid whiting out an earlier error.

I found two aspects here significant.

  1. The newspaper reports make it sound as if the error was a simple typing error, made while preparing the final report for distribution. But in fact the error was not a typing error at all. It was a handwritten error--a copying error from one source to another, probably--made not once but twice, by two different people.
  2. The error was not made in some distant office during the final prettying up of the report for publication, it was made in the lab report of the isotopic test, then again in the immediate technical approval of the lab report. One explanation of these consistent errors so near the lab level would be that these errors are the result of a very early labelling error of the specimen or aliquot. But if that were the case, then I would say from a defense point of view that the association of the tested sample with Floyd is fatally flawed.
It is also possible that the approving officer put down the wrong number by copying the wrong number from the validating scientist's mistake on the lab report, realized it was wrong, corrected it by overwriting the 5 on his report, then went back and made the validating scientist correct the number on the lab report. But at this distance, how could you prove that was the sequence? Even if you asserted this were the case, very much the same doubt would remain about the reliability of the identification of the sample.

I don't think there can be an entirely innocent explanation for this double error. Somewhere--and I'd say quite early in the process--there was some sort of sloppiness. To these officials these are merely 6-digit numbers. (It is only supposed to be us who know that one of those 6-digit numbers is Floyd Landis.) Two officials make the same error in transcribing a particular specimen's number? Doesn't that suggest that the specimen came to them with the wrong number on it? For that matter, how would the second official know that the number was wrong, since it was merely a number? But such unanswered questions are the very heart of the Landis defense, and I am now convinced that there really are grounds for doubting the association of those results to Floyd.

But there were also some quite different questions that arose when I took my latest look at these documents.

The misidentification I had originally noticed (995 475 instead of 995 474) was committed by different person than the ones who made the error reported in the newspapers. You can see this different error on p. 008, just above the isotope results. It is clear to me that it is a simple error since this person has written Floyd's ID correctly in the lower portion of the form. (This is an "approving" official actually, and you can confirm that it is the same person who wrote both sections by the handwriting and the idiosyncratic way of doing dates). This official was in charge of approving the results of the gas spectrometer tests. In the lower part of the form, Floyd is reported as normal or negative for this battery of tests, though the official reports above that those test results are unreliable because there is evidence of masking.

Both notations by this official are dated July 25--that is to say, a day after the isotope tests. My new questions: if this is the A sample (before you had any reason to suspect wrongdoing), why would you do the isotopic test before you got the results back from the gas spectrometry tests? And why isn't there any discussion in the lab report as to why the initial normal/negative results were treated as suspect?

Couple the premature (or does it not seem to be, to others?) performance of the isotope tests with the inability of the IRMS officials to pin the right ID number on the sample, and you've got a situation murky enough to cast doubt on the reliabilty of the final results.

Marc

[end]

14 comments:

dan said...

TBV/Marc,

Excellent points. And very troubling.

- Rant

James said...

So how long is it going to take for USADA to fold. I'd say the latest news has certainly shifted the balance to the Landis side.

It's like when one of the kids at the bike shop lies about their bike being stolen. It starts out with an extravagant story about someone breaking into the basement, cutting the Kryptonite lock and then escaping before the cops came. Later it was some kid that was gonna beat up the other kid...then they left it in the front yard and some bigger kid came along and took it so can they just have another bike...then when the real truth comes out the back tire was flat so they traded it for another wheel that was missing half of the bearings...
Anyway, I guess we can start going back to the theories of a lab conspiracy. Or, the real truth…the lab ‘messed up’ and now everyone is trying to save face...

chip said...

There are many troubling pieces of this puzzle...such as why the same lab performed the tests on the "b" sample. It should be mandatory for a different certified lab to confirm th results of the first lab.
There comes a point where there is a cumulative effect of the "administrative errors" where all credibility is gone. Also, there is a document from Arnie Baker that I have that throws an astounding amount of doubt on the credibility of the results. Let me know if you would like for me to email it to you. In the report, there are pages of these same types of errors as well as breaches in established protocol and procedures as established by WADA...

tbv@trustbut.com said...

Please mail me anything you have.

The same lab does the B test because that is the WADA requirement. I don't understand why that is the requirement; perhaps it is to reduce the likelihood of getting different results which would tend to put the original lab under undesired scrutiny, and "let dopers off".

On the other hand, the policy does reduce cost and complexity, because it is not trivial to ship a sample safely and securely to another facility.

TBV

chip said...

That could create a conflict of interest in that the lab has an incentive to confirm their original report of the A sample. There is corruption at every turn including in the enforcement of the rules. This really bothers me...

I will email a copy of Baker's pro Landis propaganda :) to you.

Anonymous said...

As a very casual sports fan, after reading the IHT article mentioned here yesterday, I had to look up what squash is. I wasn't sure if it's a sport using balls and rackets, or what. The suggestion that Floyd is now involved in a plot to "out" the mistakes made regarding squash and alpinism is too deliciously funny. I haven't laughed so much for weeks. While you scientists discuss the tests and documents, I appreciate the humor. I hope Floyd and Amber are laughing this week.

marc said...

Let me add a long footnote to my already too long discussion of errors, handwritten or not. I wrote (in regard to p. 009): "But the approving official clearly wrote the wrong ID number, originally writing 994 474, I think, and corrected it by writing a much larger 5 over the original wrong digit. (This, I think, must be the source of the newspaper reports of what the labelling error was.)"

I no longer think that p. 009 was the source of the "typo" the lab was apologizing for. The newspaper accounts, after all, consistently reported that the error was in a document relating to the B sample. Looking through the pages dedicated to the B sample, I found--on p. 288--two genuine typographical errors: one mistakenly identifying the sample as 994 474 (the admitted error); the other mistakenly identifying the lab's own ID number for the sample as 478/07 instead of 178/07. Since these really are simple typos, I now believe that this page contains the error acknowledged and dismissed in the newspapers.

But that means that the handwritten errors I pointed out on pp. 008 and 009, regarding the A sample, remain unrecognized by the lab. And I continue to think them potentially more significant than the typed errors of p. 288.

Marc

Mark Reagan said...

Marc,

I am thoroughly enjoying you blog you are doing an awesome job "ferreting" out the French lab follies. However I have a small request. Could you post a point by point summary of the typo's, where they are found and who you think made them? So I can better understand who what and how, I just get lost in all the verbage. I think It would help me keep it straight. Again awesome job even if you don't have time for the request.

Thanks,

Mark R.

Gary said...

Again, why is there no oversight? The accountability with all of these organizations stinks. Again, the only ones accountable are the athletes. I'm not saying there aren't cheaters, some do cheat. And some labs make errors. And some governing organizations attack and deny before they work through due process.

On a side note, everyone should click on the link James put up. St. Louis Bike Works is a fantastic organization. They do amazing stuff and the shop is a hell of a cool place.

Cheryl from Maryland said...

So -- another review of the data by Marc finds additional labeling errors which are germane to the custody chain of the samples (and thus their reliability as evidence) and which aren't the errors to which the LNDD has admitted. I can't wait to see what reading the documents another time reveals.

marc said...

Hi Mark,

I get lost in all the pages, too. But after all the words explaining them, there are really only 3 pages involved (page references are to the page numbers that are prefixed USADA in the materials released by Floyd Landis and available through TbV). See if this helps.

(1) The French lab has admitted to one error in the official write-up of the tests on the B sample, identifying the sample by the number 994 474 (when Floyd's ID number is 995 474). In one newspaper story the error is characterized as an "administrative" error, in another story as a "typing" error. I think the error referred to is on p. 288, where the admitted error does occur. (A second error, which is unacknowledged by the lab, also occurs on that page. The lab's internal ID number for the sample is printed as 478/07, when the internal ID number for Floyd’s sample is actually 178/07.) It is possible that both these errors are simply typographical errors, though it remains fair to question that, and to ask for proof that the results reported don't really belong to some other sample than Floyd's.

(2) A different pair of errors, relating to the A sample, exists on pages 008 and 009. These are not typographical errors, since both are handwritten. Two different lab officials--certifying the results of the isotope test at two different levels--identified the A sample by a number different from Floyd's. In the instance on p. 009, it looks like the official originally wrote 994 474, then corrected it to 995 474 by overwriting the 4 in 994 with a larger 5. In the instance on p. 008 we can't tell what the original error was, since it seems to have been covered by white-out before Floyd's number was written in.

The lab has made no mention of these changes. I think these are obviously serious problems since their being handwritten, their being repeated by two different people, and their appearing much earlier in the testing process suggests that there may have been an early misidentification of a sample. The lab's silence on them also undermines confidence in the lab's oversight of the whole process.

(3) One other error occurs on p. 008. A different official suspects that something is masking a proper test of the sample, and identifies the sample by the number 995 475. Presumably, this masked sample is supposed to be Floyd's, and the same handwriting elsewhere on the page states that the original results of the gas spectrometry tests on Floyd's sample (995 474) are normal or negative. If 995 475 was written where 995 474 was intended, this error also is not acknowledged by the lab. It raises the same general problem of oversight and reliability, to which I would add: Why is there no discussion of Floyd's normal results or any explanation of why masking is suspected (if that sample really is Floyd's)?

Marc

Gene said...

where is the webcast mentioned in this article? ( http://tinyurl.com/y458mw )

Floyd Landis said, "I ask cyclists, cycling fans and anyone who is interested in fair treatment and due process to please watch this web cast and download Dr. Baker's presentation. Any fair minded individual who reviews the facts will see that what's fair is clear: the sanctions against me should be overturned immediately."

tbv@trustbut.com said...

Gene,

I suspect this refers to the versions of the presentation that will be posted later -- Monday by the press release, though I'd been led to believe it might be sometime tonight or over the weekend.

TBV

Anonymous said...

Many of these issues could have been easily avoided with better design of the testing protocol.

Why aren't the sample numbers sparse, with a checksum digit, like credit card numbers are?

Then most of these simple transposition and copying errors could be identified after the fact.