So they found something in 4 of 7 B samples. They found NOTHING in 7 of 7 A samples.
Sounds like the test is bullshit. Results appear non-repeatable with identical samples.
Actually, the poorly-written ESPN article says 1 of 7 "A" samples tested positive using a different testing method from the "B" samples:
- July 20: Stage 17, a mountain stage finishing in Morzine: This was Landis' comeback stage in which he left the peloton behind with a solo breakaway. He made up 8 minutes and scrambled back into second place overall, 30 seconds behind Pereiro.
USADA: Landis' "A" and "B" samples exceeded the allowable 4-to-1 testosterone-to-epitestosterone ratio and showed the presence of synthetic testosterone.
Also, as this article explains, different tests are used for the "A" and "B" samples. The "A" tests are just a "rough" test that measures the ratio of testosterone to epitestosterone (T/E ratio) to see if it's abnormal enough to trigger the more stringent "B" test (which actually measures for synthetic testosterone). A "normal" T/E ratio is 1-to-1, but anything below 4-to-1 is declared "negative," so a steroid user can pass the "A" test as long as they don't overdo it.
During the initial Tour de France testing, only 1 "B" sample was tested at all because only 1 "A" sample tested positive (triggering the "B" test). All of the "B" samples were tested later because the USADA requested it for the trial.