One result of the slide show's re-release was to send me off spending a whole lot more time than I had intended, or wanted, looking at the lab reports in the Landis documents.
Marc finds some things that are curious,
but after discussion, don't appear to be
earth-shattering. I suggest taking it as
a lesson in navigating the LDP to look for
something given a few pointers. Marc
here starts from leads in slide-show 2.1.
(You can click page images to get the original scan)
We've often agreed in this blog that the best Landis defense would come from some scientific challenge to the lab results. Reading the discussion over at DPF, though, I'm less optimistic that that can occur successfully. The debate is now focussing on where the threshholds should be set, and there doesn't seem to be consensus about that. The slide show makes various assertions as to where they should be, but there's considerable disagreement about how authoritative those assertions are.
In that context, arguments regarding the consistency of the lab findings become serious matters and not just quibbles, I think. We expect scientific results to be repeatable: same sample, same results, even if by different routes, at least in most cases. In that regard, the slide show raises some very serious questions, to which I'd like to add a little detail.
Before starting, let me warn everyone that even here nothing is entirely straightforward. The original sample (the A sample) was subjected to a fairly extensive battery of tests, for each of which a different preparation regimen was to be followed. Thus, broken into its many little aliquots and prepared for the different tests, as I understand it the single source becomes a whole range of differing samples. I suppose that means that the absolute readings of the various steroids might be different from one aliquot and test to another. But at the same time, it seems to me that the sainted T/E ratio ought to remain the same (or near enough) regardless.
A really pointed example of what look to me like significant inconsistencies in the results can be found in the three tests reported on USADA 054 (a screening test), USADA 212, and USADA 214 (the latter two confirmation tests). In fact, all three tests were run on the same day--Saturday, 7/22--by the same individual. Their results vary widely. The episode of these three tests is to me a microcosm of the general behavior of the lab.
Presumably, the test on p. 0054--the screening test--was performed first. It reports back a T/E ratio of 4.9. Had that been all, would the lab really have gone forward with all the rest of this pursuit? But the operator signals a "derivation inhibition," and checks the "Verify" box on the form. There is no explanation as to what led to this conclusion. (See below, by the way.)
At this point, I imagine,the tests on p. 0212 and p. 0214 were performed. Their results are altogether different from 0054's, both in regard to absolute values for T and E and the T/E ratio. Now, on the one hand, their absolute results might be different from 0054's because they were performing a different test from 0054, and needed a different preparation. But on the other hand, the results on 0212 and 0214 don't agree with each other in regard to the absolute values of T and E, even though they were performing the same test.
(Here again, there might be an obscure explanation. 0212 and 0214 may be using different samples. 0212's is identified as "995474te" while 0214's is "995474te SSH." But I wonder whether these really are different samples, and even if they were, I would want it confirmed that this accounted for their different results. I warned you that nothing was straightforward.)
In any case, 0212 and 0214 don't exactly agree with each other on the T/E ratio, either, though they are closer to each other, and far removed from 0054. 0054 reported 4.9, 0212 reported 10.7, and 0214 11.2. But all of that variation, coming with the same operator, on the same day, does not exactly foster confidence in the solid scientific underpinnings of the results.
(There are two interesting sidebars to this day's tangled story. First, when it came to writing up the T/E ratio results, LNDD used neither of these tests in its report on USADA 101. It uses a test from a 7/24, on p. 0092.
Second, the same test performed on 0054 was repeated on 7/25, apparently using the identical sample (995474H) despite the dread "derivation inhibition." Its results are reported on p. 0057.
The absolute results for the various steroids are completely different from 0054's. There's that fine scientific repeatability again. The T/E ratio, however, is very similar: 5.1. Go figure. The new operator notes: "Confirmation vial reinjected in screening." What in the world does that mean?)
Perhaps someone with experience of these tests can explain why such widely and, if I can use the term, "consistently" divergent results are normal and do not signal serious scientific failures. But no one has bothered to do so yet, and until someone does, I am looking wirh increasing skepticism at the pronouncements from the "authorities.”