Monday, January 28, 2008

Mr. Idiot shows polarity affects isotropic skew

In a comment, Mike pointed us to observation that the isotropic skew is dependent on the polarity of the column, which we know to be different in the Landis GCMS and IRMS runs. He also pointed us to the reference, which we include a snapshot of below, because there is a lot interesting there in very little space.

Note particularly the observations that overlaps are trouble, peak and baseline separation are important, and that Brenna's software isn't used in practice.

Handbook of Stable Isotope Analytical Techniques, Volume I, parts of pages 168 and 169. Click for bigger.

This is a survey article by one Wolfram Meier-Augenstein, and is citing J. Brenna.

Small world.


Mike Solberg said...

I have read in other places where WMA referenced Brenna's work too. Brenna's work is a basic part of the field of IRMS, as is WMA's. For the arbs to say that WMA's position on the comparison "across machines" was "scientifically totally unacceptable and fundamentally flawed" was just ludicrous. That's like me telling Brett Favre how to play quarterback. Perhaps WMA didn't make his point very well, but clearly the man is a world class authority on IRMS and it's connection to GCMS. The most they should have said was "we don't get it."

Anyway...wrt this isotropic skew I am not sure it matters much that the column was different for the GCMS and the IRMS. What matters is if they had some corrective software setting set improperly for the IRMS. I think Ali said that there probably was no systemic adjustment, and I think Simon Davis testified that (with all the corporate change of hands that went on in the IRMS business) no one really knows anymore what the LNDD software is doing, so perhaps there is no way to know if something was improperly set...but what does THAT say about LNDD's methods?